		Metadata-Version: 2.4
		Name: eigency
		Version: 3.4.0.6
		Summary: Cython interface between the numpy arrays and the Matrix/Array classes of the Eigen C++ library
		Author-email: Wouter Boomsma <wb@di.ku.dk>, Felipe Bordeu <felipebordeu@gmail.com>
		Maintainer-email: Felipe Bordeu <felipebordeu@gmail.com>, Wouter Boomsma <wb@di.ku.dk>
		License-Expression: LicenseRef-Wouter-Boomsma-1.0
		Project-URL: Homepage, https://github.com/eigency-org/eigency
		Classifier: Intended Audience :: Developers
		Classifier: Programming Language :: C++
		Classifier: Programming Language :: Cython
		Classifier: Programming Language :: Python :: 3.9
		Classifier: Programming Language :: Python :: 3.10
		Classifier: Programming Language :: Python :: 3.11
		Classifier: Programming Language :: Python :: 3.12
		Classifier: Programming Language :: Python :: 3.13
		Requires-Python: >=3.9
		Description-Content-Type: text/markdown
		License-File: LICENSE
		Requires-Dist: numpy>=1.23.5
		Dynamic: license-file

		# Eigency

		[![PyPI version](https://badge.fury.io/py/eigency.svg)](https://badge.fury.io/py/eigency)
		[![PEP 517](https://github.com/eigency-org/eigency/actions/workflows/build.yml/badge.svg)](https://github.com/eigency-org/eigency/actions/workflows/build.yml)
		[![pip wheel](https://github.com/eigency-org/eigency/actions/workflows/wheel.yml/badge.svg)](https://github.com/eigency-org/eigency/actions/workflows/wheel.yml)
		[![setup.py](https://github.com/eigency-org/eigency/actions/workflows/setup.yml/badge.svg)](https://github.com/eigency-org/eigency/actions/workflows/setup.yml)
		[![pre-commit](https://github.com/eigency-org/eigency/actions/workflows/pre-commit.yml/badge.svg)](https://github.com/eigency-org/eigency/actions/workflows/pre-commit.yml)

		Eigency is a Cython interface between Numpy arrays and Matrix/Array
		objects from the Eigen C++ library. It is intended to simplify the
		process of writing C++ extensions using the Eigen library. Eigency is
		designed to reuse the underlying storage of the arrays when passing
		data back and forth, and will thus avoid making unnecessary copies
		whenever possible. Only in cases where copies are explicitly requested
		by your C++ code will they be made.

		## Versioning

		Eigency uses a 4-number version (`N.N.N.N`) where the first 3 parts correspond to the embedded Eigen library version.
		The last part is a revision number of Eigency itself.

		## Installing

		Eigency is packaged as a source distribution (`sdist`) and available on PyPi.
		It can be easily installed using `pip`:
		```bash
		python -m pip install eigency
		```

		Requirement: `pip >= 18.0`

		If your `pip` is too old, then upgrade it using:
		```bash
		python -m pip install --upgrade pip
		```

		## Contributing

		For instructions on building and/or packaging Eigency from source,
		see the contributing guide [here](https://github.com/eigency-org/eigency/blob/master/CONTRIBUTING.md).

		## Usage

		Below is a description of a range of common usage scenarios. A full working
		example of both setup and these different use cases is available in the
		`test` directory distributed with the this package.

		### Setup

		To import eigency functionality, add the following to your `.pyx` file:
		```
		from eigency.core cimport *
		```

		In addition, in the `setup.py` file, the include directories must be
		set up to include the eigency includes. This can be done by calling
		the `get_includes` function in the `eigency` module:
		```
		import eigency
		...
		extensions = [
		Extension("module-dir-name/module-name", ["module-dir-name/module-name.pyx"],
		include_dirs = [".", "module-dir-name"] + eigency.get_includes()
		),
		]
		```

		Eigency includes a version of the Eigen library, and the `get_includes` function will include the path to this directory. If you
		have your own version of Eigen, just set the `include_eigen` option to False, and add your own path instead:
		```
		include_dirs = [".", "module-dir-name", 'path-to-own-eigen'] + eigency.get_includes(include_eigen=False)
		```

		### From Numpy to Eigen

		Assume we are writing a Cython interface to the following C++ function:

		```c++
		void function_w_mat_arg(const Eigen::Map<Eigen::MatrixXd> &mat) {
		std::cout << mat << "\n";
		}
		```

		Note that we use `Eigen::Map` to ensure that we can reuse the storage
		of the numpy array, thus avoiding making a copy. Assuming the C++ code
		is in a file called `functions.h`, the corresponding `.pyx` entry could look like this:

		```
		cdef extern from "functions.h":
		cdef void _function_w_mat_arg "function_w_mat_arg"(Map[MatrixXd] &)

		# This will be exposed to Python
		def function_w_mat_arg(np.ndarray array):
		return _function_w_mat_arg(Map[MatrixXd](array))
		```

		The last line contains the actual conversion. `Map` is an Eigency
		type that derives from the real Eigen map, and will take care of
		the conversion from the numpy array to the corresponding Eigen type.

		We can now call the C++ function directly from Python:
		```python
		>>> import numpy as np
		>>> import eigency_tests
		>>> x = np.array([[1.1, 2.2], [3.3, 4.4]])
		>>> eigency_tests.function_w_mat_arg(x)
		1.1 3.3
		2.2 4.4
		```
		(if you are wondering about why the matrix is transposed, please
		see the Storage layout section below).

		## Types matter

		The basic idea behind eigency is to share the underlying representation of a
		numpy array between Python and C++. This means that somewhere in the process,
		we need to make explicit which numerical types we are dealing with. In the
		function above, we specify that we expect an Eigen MatrixXd, which means
		that the numpy array must also contain double (i.e. float64) values. If we instead provide
		a numpy array of ints, we will get strange results.

		```python
		>>> import numpy as np
		>>> import eigency_tests
		>>> x = np.array([[1, 2], [3, 4]])
		>>> eigency_tests.function_w_mat_arg(x)
		4.94066e-324 1.4822e-323
		9.88131e-324 1.97626e-323
		```
		This is because we are explicitly asking C++ to interpret out python integer
		values as floats.

		To avoid this type of error, you can force your cython function to
		accept only numpy arrays of a specific type:

		```
		cdef extern from "functions.h":
		cdef void _function_w_mat_arg "function_w_mat_arg"(Map[MatrixXd] &)

		# This will be exposed to Python
		def function_w_mat_arg(np.ndarray[np.float64_t, ndim=2] array):
		return _function_w_mat_arg(Map[MatrixXd](array))
		```

		(Note that when using this technique to select the type, you also need to specify
		the dimensions of the array (this will default to 1)). Using this new definition,
		users will get an error when passing arrays of the wrong type:

		```python
		>>> import numpy as np
		>>> import eigency_tests
		>>> x = np.array([[1, 2], [3, 4]])
		>>> eigency_tests.function_w_mat_arg(x)
		Traceback (most recent call last):
		File "<stdin>", line 1, in <module>
		File "eigency_tests/eigency_tests.pyx", line 87, in eigency_tests.eigency_tests.function_w_mat_arg
		ValueError: Buffer dtype mismatch, expected 'float64_t' but got 'long'
		```

		Since it avoids many surprises, it is strongly recommended to use this technique
		to specify the full types of numpy arrays in your cython code whenever
		possible.

		### Writing Eigen Map types in Cython

		Since Cython does not support nested fused types, you cannot write types like `Map[Matrix[double, 2, 2]]`. In most cases, you won't need to, since you can just use Eigens convenience typedefs, such as `Map[VectorXd]`. If you need the additional flexibility of the full specification, you can use the `FlattenedMap` type, where all type arguments can be specified at top level, for instance `FlattenedMap[Matrix, double, _2, _3]` or `FlattenedMap[Matrix, double, _2, Dynamic]`. Note that dimensions must be prefixed with an underscore.

		Using full specifications of the Eigen types, the previous example would look like this:
		```
		cdef extern from "functions.h":
		cdef void _function_w_mat_arg "function_w_mat_arg" (FlattenedMap[Matrix, double, Dynamic, Dynamic] &)

		# This will be exposed to Python
		def function_w_mat_arg(np.ndarray[np.float64_t, ndim=2] array):
		return _function_w_mat_arg(FlattenedMap[Matrix, double, Dynamic, Dynamic](array))
		```

		`FlattenedType` takes four template parameters: arraytype, scalartype,
		rows and cols. Eigen supports a few other template arguments for
		setting the storage layout and Map strides. Since cython does not
		support default template arguments for fused types, we have instead
		defined separate types for this purpose. These are called
		`FlattenedMapWithOrder` and `FlattenedMapWithStride` with five and eight
		template arguments, respectively. For details on their use, see the section
		about storage layout below.

		### From Numpy to Eigen (insisting on a copy)

		Eigency will not complain if the C++ function you interface with does
		not take a Eigen Map object, but instead a regular Eigen Matrix or
		Array. However, in such cases, a copy will be made. Actually, the
		procedure is exactly the same as above. In the `.pyx` file, you still
		define everything exactly the same way as for the Map case described above.

		For instance, given the following C++ function:
		```c++
		void function_w_vec_arg_no_map(const Eigen::VectorXd &vec);
		```

		The Cython definitions would still look like this:

		```
		cdef extern from "functions.h":
		cdef void _function_w_vec_arg_no_map "function_w_vec_arg_no_map"(Map[VectorXd] &)

		# This will be exposed to Python
		def function_w_vec_arg_no_map(np.ndarray[np.float64_t] array):
		return _function_w_vec_arg_no_map(Map[VectorXd](array))
		```

		Cython will not mind the fact that the argument type in the extern
		declaration (a Map type) differs from the actual one in the `.h` file,
		as long as one can be assigned to the other. Since Map objects can be
		assigned to their corresponding Matrix/Array types this works
		seemlessly. But keep in mind that this assignment will make a copy of
		the underlying data.

		### Eigen to Numpy

		C++ functions returning a reference to an Eigen Matrix/Array can also
		be transferred to numpy arrays without copying their content. Assume
		we have a class with a single getter function that returns an Eigen
		matrix member:

		```c++
		class MyClass {
		public:
		MyClass():
		matrix(Eigen::Matrix3d::Constant(3.)) {
		}
		Eigen::MatrixXd &get_matrix() {
		return this->matrix;
		}
		private:
		Eigen::Matrix3d matrix;
		};
		```

		The Cython C++ class interface is specified as usual:

		```
		cdef cppclass _MyClass "MyClass":
		_MyClass "MyClass"() except +
		Matrix3d &get_matrix()
		```

		And the corresponding Python wrapper:

		```python
		cdef class MyClass:
		cdef _MyClass *thisptr;

		def __cinit__(self):
		self.thisptr = new _MyClass()

		def __dealloc__(self):
		del self.thisptr

		def get_matrix(self):
		return ndarray(self.thisptr.get_matrix())
		```

		This last line contains the actual conversion. Again, eigency has its
		own version of `ndarray`, that will take care of the conversion for
		you.

		Due to limitations in Cython, Eigency cannot deal with full
		Matrix/Array template specifications as return types
		(e.g. `Matrix[double, 4, 2]`). However, as a workaround, you can use
		`PlainObjectBase` as a return type in such cases (or in all cases if
		you prefer):

		```
		PlainObjectBase &get_matrix()
		```

		### Overriding default behavior

		The `ndarray` conversion type specifier will attempt do guess whether you want a copy
		or a view, depending on the return type. Most of the time, this is
		probably what you want. However, there might be cases where you want
		to override this behavior. For instance, functions returning const
		references will result in a copy of the array, since the const-ness
		cannot be enforced in Python. However, you can always override the
		default behavior by using the `ndarray_copy` or `ndarray_view`
		functions.

		Expanding the `MyClass` example from before:

		```c++
		class MyClass {
		public:
		...
		const Eigen::MatrixXd &get_const_matrix() {
		return this->matrix;
		}
		...
		};
		```

		With the corresponding cython interface specification
		The Cython C++ class interface is specified as usual:

		```
		cdef cppclass _MyClass "MyClass":
		...
		const Matrix3d &get_const_matrix()
		```

		The following would return a copy

		```python
		cdef class MyClass:
		...
		def get_const_matrix(self):
		return ndarray(self.thisptr.get_const_matrix())
		```

		while the following would force it to return a view

		```python
		cdef class MyClass:
		...
		def get_const_matrix(self):
		return ndarray_view(self.thisptr.get_const_matrix())
		```

		### Eigen to Numpy (non-reference return values)

		Functions returning an Eigen object (not a reference), are specified
		in a similar way. For instance, given the following C++ function:

		```c++
		Eigen::Matrix3d function_w_mat_retval();
		```

		The Cython code could be written as:

		```
		cdef extern from "functions.h":
		cdef Matrix3d _function_w_mat_retval "function_w_mat_retval" ()

		# This will be exposed to Python
		def function_w_mat_retval():
		return ndarray_copy(_function_w_mat_retval())
		```

		As mentioned above, you can replace `Matrix3d` (or any other Eigen return type) with
		`PlainObjectBase`, which is especially relevant when working with
		Eigen object that do not have an associated convenience typedef.

		Note that we use `ndarray_copy` instead of `ndarray` to explicitly
		state that a copy should be made. In c++11 compliant compilers, it
		will detect the rvalue reference and automatically make a copy even if
		you just use `ndarray` (see next section), but to ensure that it works
		also with older compilers it is recommended to always use
		`ndarray_copy` when returning newly constructed eigen values.


		### Corrupt data when returning non-map types

		The tendency of Eigency to avoid copies whenever possible can lead
		to corrupted data when returning non-map Eigen arrays. For instance,
		in the `function_w_mat_retval` from the previous section, a temporary
		value will be returned from C++, and we have to take care to make
		a copy of this data instead of letting the resulting numpy array
		refer directly to this memory. In C++11, this situation can be
		detected directly using rvalue references, and it will therefore
		automatically make a copy:
		```
		def function_w_mat_retval():
		# This works in C++11, because it detects the rvalue reference
		return ndarray(_function_w_mat_retval())
		```

		However, to make sure it works with older compilers,
		it is recommended to use the `ndarray_copy` conversion:

		```
		def function_w_mat_retval():
		# Explicit request for copy - this always works
		return ndarray_copy(_function_w_mat_retval())
		```

		### Storage layout - why arrays are sometimes transposed

		The default storage layout used in numpy and Eigen differ. Numpy uses
		a row-major layout (C-style) per default while Eigen uses a
		column-major layout (Fortran style) by default. In Eigency, we prioritize to
		avoid copying of data whenever possible, which can have unexpected
		consequences in some cases: There is no problem when passing values
		from C++ to Python - we just adjust the storage layout of the returned
		numpy array to match that of Eigen. However, since the storage layout
		is encoded into the _type_ of the Eigen array (or the type of the
		Map), we cannot automatically change the layout in the Python to C++ direction. In
		Eigency, we have therefore opted to return the transposed array/matrix
		in such cases. This provides the user with the flexibility to deal
		with the problem either in Python (use order="F" when constructing
		your numpy array), or on the C++ side: (1) explicitly define your
		argument to have the row-major storage layout, 2) manually set the Map
		stride, or 3) just call `.transpose()` on the received
		array/matrix).

		As an example, consider the case of a C++ function that both receives
		and returns a Eigen Map type, thus acting as a filter:

		```c++
		Eigen::Map<Eigen::ArrayXXd> function_filter(Eigen::Map<Eigen::ArrayXXd> &mat) {
		return mat;
		}
		```

		The Cython code could be:

		```
		cdef extern from "functions.h":
		...
		cdef Map[ArrayXXd] &_function_filter1 "function_filter1" (Map[ArrayXXd] &)

		def function_filter1(np.ndarray[np.float64_t, ndim=2] array):
		return ndarray(_function_filter1(Map[ArrayXXd](array)))

		```

		If we call this function from Python in the standard way, we will
		see that the array is transposed on the way from Python to C++, and
		remains that way when it is again returned to Python:

		```python
		>>> x = np.array([[1., 2., 3., 4.], [5., 6., 7., 8.]])
		>>> y = function_filter1(x)
		>>> print x
		[[ 1. 2. 3. 4.]
		[ 5. 6. 7. 8.]]
		>>> print y
		[[ 1. 5.]
		[ 2. 6.]
		[ 3. 7.]
		[ 4. 8.]]
		```

		The simplest way to avoid this is to tell numpy to use a
		column-major array layout instead of the default row-major
		layout. This can be done using the order='F' option:

		```python
		>>> x = np.array([[1., 2., 3., 4.], [5., 6., 7., 8.]], order='F')
		>>> y = function_filter1(x)
		>>> print x
		[[ 1. 2. 3. 4.]
		[ 5. 6. 7. 8.]]
		>>> print y
		[[ 1. 2. 3. 4.]
		[ 5. 6. 7. 8.]]
		```

		The other alternative is to tell Eigen to use RowMajor layout. This
		requires changing the C++ function definition:

		```c++
		typedef Eigen::Map<Eigen::Array<double, Eigen::Dynamic, Eigen::Dynamic, Eigen::RowMajor> > RowMajorArrayMap;

		RowMajorArrayMap &function_filter2(RowMajorArrayMap &mat) {
		return mat;
		}
		```

		To write the corresponding Cython definition, we need the expanded version of
		`FlattenedMap` called `FlattenedMapWithOrder`, which allows us to specify
		the storage order:

		```
		cdef extern from "functions.h":
		...
		cdef PlainObjectBase _function_filter2 "function_filter2" (FlattenedMapWithOrder[Array, double, Dynamic, Dynamic, RowMajor])

		def function_filter2(np.ndarray[np.float64_t, ndim=2] array):
		return ndarray(_function_filter2(FlattenedMapWithOrder[Array, double, Dynamic, Dynamic, RowMajor](array)))
		```

		Another alternative is to keep the array itself in RowMajor format,
		but use different stride values for the Map type:

		```c++
		typedef Eigen::Map<Eigen::ArrayXXd, Eigen::Unaligned, Eigen::Stride<1, Eigen::Dynamic> > CustomStrideMap;

		CustomStrideMap &function_filter3(CustomStrideMap &);
		```

		In this case, in Cython, we need to use the even more extended
		`FlattenedMap` type called `FlattenedMapWithStride`, taking eight
		arguments:

		```
		cdef extern from "functions.h":
		...
		cdef PlainObjectBase _function_filter3 "function_filter3" (FlattenedMapWithStride[Array, double, Dynamic, Dynamic, ColMajor, Unaligned, _1, Dynamic])

		def function_filter3(np.ndarray[np.float64_t, ndim=2] array):
		return ndarray(_function_filter3(FlattenedMapWithStride[Array, double, Dynamic, Dynamic, ColMajor, Unaligned, _1, Dynamic](array)))
		```

		In all three cases, the returned array will now be of the same shape
		as the original.

		### Long double support

		Eigency provides new shorthands for Eigen `long double` and complex `long double` Matrix and Array types.
		Examples:
		```
		Vector4ld
		Matrix3ld
		Vector2cld
		Matrix4cld
		Array3Xld
		ArrayXXcld
		```

		These typedefs are available in the `eigency` namespace when including the `eigency` header:
		```c++
		#include "eigency.h"

		void receive_long_double_matrix(Eigen::Map<eigency::MatrixXld> &mat) {
		// use long double eigen matrix
		}
		```

		Use Cython (.pyx) to create Python binding to your C++ function:
		```python
		cdef extern from "functions.h":
		cdef void _receive_long_double_matrix "receive_long_double_matrix"(Map[MatrixXld] &)

		def send_long_double_ndarray(np.ndarray[np.longdouble_t, ndim=2] array):
		return _receive_long_double_matrix(Map[MatrixXld](array))
		```

		Invoke in Python:
		```python
		import numpy as np
		import my_module

		x = np.array([[1.1, 2.2], [3.3, 4.4]], dtype=np.longdouble)
		my_module.send_long_double_ndarray(x)
		```

+1

eigency.egg-info/requires.txt

numpy>=1.23.5

+365

eigency.egg-info/SOURCES.txt

		.gitmodules
		LICENSE
		README.md
		pyproject.toml
		setup.cfg
		setup.py
		eigency/__init__.py
		eigency/conversions.cpp
		eigency/conversions.pxd
		eigency/conversions.pyx
		eigency/conversions_api.h
		eigency/core.cpp
		eigency/core.pxd
		eigency/core.pyx
		eigency/eigency.h
		eigency/eigency_cpp.h
		eigency.egg-info/PKG-INFO
		eigency.egg-info/SOURCES.txt
		eigency.egg-info/dependency_links.txt
		eigency.egg-info/requires.txt
		eigency.egg-info/top_level.txt
		eigency/eigen/COPYING.APACHE
		eigency/eigen/COPYING.BSD
		eigency/eigen/COPYING.GPL
		eigency/eigen/COPYING.LGPL
		eigency/eigen/COPYING.MINPACK
		eigency/eigen/COPYING.MPL2
		eigency/eigen/COPYING.README
		eigency/eigen/Eigen/Cholesky
		eigency/eigen/Eigen/CholmodSupport
		eigency/eigen/Eigen/Core
		eigency/eigen/Eigen/Dense
		eigency/eigen/Eigen/Eigen
		eigency/eigen/Eigen/Eigenvalues
		eigency/eigen/Eigen/Geometry
		eigency/eigen/Eigen/Householder
		eigency/eigen/Eigen/IterativeLinearSolvers
		eigency/eigen/Eigen/Jacobi
		eigency/eigen/Eigen/KLUSupport
		eigency/eigen/Eigen/LU
		eigency/eigen/Eigen/MetisSupport
		eigency/eigen/Eigen/OrderingMethods
		eigency/eigen/Eigen/PaStiXSupport
		eigency/eigen/Eigen/PardisoSupport
		eigency/eigen/Eigen/QR
		eigency/eigen/Eigen/QtAlignedMalloc
		eigency/eigen/Eigen/SPQRSupport
		eigency/eigen/Eigen/SVD
		eigency/eigen/Eigen/Sparse
		eigency/eigen/Eigen/SparseCholesky
		eigency/eigen/Eigen/SparseCore
		eigency/eigen/Eigen/SparseLU
		eigency/eigen/Eigen/SparseQR
		eigency/eigen/Eigen/StdDeque
		eigency/eigen/Eigen/StdList
		eigency/eigen/Eigen/StdVector
		eigency/eigen/Eigen/SuperLUSupport
		eigency/eigen/Eigen/UmfPackSupport
		eigency/eigen/Eigen/src/Cholesky/LDLT.h
		eigency/eigen/Eigen/src/Cholesky/LLT.h
		eigency/eigen/Eigen/src/Cholesky/LLT_LAPACKE.h
		eigency/eigen/Eigen/src/CholmodSupport/CholmodSupport.h
		eigency/eigen/Eigen/src/Core/ArithmeticSequence.h
		eigency/eigen/Eigen/src/Core/Array.h
		eigency/eigen/Eigen/src/Core/ArrayBase.h
		eigency/eigen/Eigen/src/Core/ArrayWrapper.h
		eigency/eigen/Eigen/src/Core/Assign.h
		eigency/eigen/Eigen/src/Core/AssignEvaluator.h
		eigency/eigen/Eigen/src/Core/Assign_MKL.h
		eigency/eigen/Eigen/src/Core/BandMatrix.h
		eigency/eigen/Eigen/src/Core/Block.h
		eigency/eigen/Eigen/src/Core/BooleanRedux.h
		eigency/eigen/Eigen/src/Core/CommaInitializer.h
		eigency/eigen/Eigen/src/Core/ConditionEstimator.h
		eigency/eigen/Eigen/src/Core/CoreEvaluators.h
		eigency/eigen/Eigen/src/Core/CoreIterators.h
		eigency/eigen/Eigen/src/Core/CwiseBinaryOp.h
		eigency/eigen/Eigen/src/Core/CwiseNullaryOp.h
		eigency/eigen/Eigen/src/Core/CwiseTernaryOp.h
		eigency/eigen/Eigen/src/Core/CwiseUnaryOp.h
		eigency/eigen/Eigen/src/Core/CwiseUnaryView.h
		eigency/eigen/Eigen/src/Core/DenseBase.h
		eigency/eigen/Eigen/src/Core/DenseCoeffsBase.h
		eigency/eigen/Eigen/src/Core/DenseStorage.h
		eigency/eigen/Eigen/src/Core/Diagonal.h
		eigency/eigen/Eigen/src/Core/DiagonalMatrix.h
		eigency/eigen/Eigen/src/Core/DiagonalProduct.h
		eigency/eigen/Eigen/src/Core/Dot.h
		eigency/eigen/Eigen/src/Core/EigenBase.h
		eigency/eigen/Eigen/src/Core/ForceAlignedAccess.h
		eigency/eigen/Eigen/src/Core/Fuzzy.h
		eigency/eigen/Eigen/src/Core/GeneralProduct.h
		eigency/eigen/Eigen/src/Core/GenericPacketMath.h
		eigency/eigen/Eigen/src/Core/GlobalFunctions.h
		eigency/eigen/Eigen/src/Core/IO.h
		eigency/eigen/Eigen/src/Core/IndexedView.h
		eigency/eigen/Eigen/src/Core/Inverse.h
		eigency/eigen/Eigen/src/Core/Map.h
		eigency/eigen/Eigen/src/Core/MapBase.h
		eigency/eigen/Eigen/src/Core/MathFunctions.h
		eigency/eigen/Eigen/src/Core/MathFunctionsImpl.h
		eigency/eigen/Eigen/src/Core/Matrix.h
		eigency/eigen/Eigen/src/Core/MatrixBase.h
		eigency/eigen/Eigen/src/Core/NestByValue.h
		eigency/eigen/Eigen/src/Core/NoAlias.h
		eigency/eigen/Eigen/src/Core/NumTraits.h
		eigency/eigen/Eigen/src/Core/PartialReduxEvaluator.h
		eigency/eigen/Eigen/src/Core/PermutationMatrix.h
		eigency/eigen/Eigen/src/Core/PlainObjectBase.h
		eigency/eigen/Eigen/src/Core/Product.h
		eigency/eigen/Eigen/src/Core/ProductEvaluators.h
		eigency/eigen/Eigen/src/Core/Random.h
		eigency/eigen/Eigen/src/Core/Redux.h
		eigency/eigen/Eigen/src/Core/Ref.h
		eigency/eigen/Eigen/src/Core/Replicate.h
		eigency/eigen/Eigen/src/Core/Reshaped.h
		eigency/eigen/Eigen/src/Core/ReturnByValue.h
		eigency/eigen/Eigen/src/Core/Reverse.h
		eigency/eigen/Eigen/src/Core/Select.h
		eigency/eigen/Eigen/src/Core/SelfAdjointView.h
		eigency/eigen/Eigen/src/Core/SelfCwiseBinaryOp.h
		eigency/eigen/Eigen/src/Core/Solve.h
		eigency/eigen/Eigen/src/Core/SolveTriangular.h
		eigency/eigen/Eigen/src/Core/SolverBase.h
		eigency/eigen/Eigen/src/Core/StableNorm.h
		eigency/eigen/Eigen/src/Core/StlIterators.h
		eigency/eigen/Eigen/src/Core/Stride.h
		eigency/eigen/Eigen/src/Core/Swap.h
		eigency/eigen/Eigen/src/Core/Transpose.h
		eigency/eigen/Eigen/src/Core/Transpositions.h
		eigency/eigen/Eigen/src/Core/TriangularMatrix.h
		eigency/eigen/Eigen/src/Core/VectorBlock.h
		eigency/eigen/Eigen/src/Core/VectorwiseOp.h
		eigency/eigen/Eigen/src/Core/Visitor.h
		eigency/eigen/Eigen/src/Core/arch/AVX/Complex.h
		eigency/eigen/Eigen/src/Core/arch/AVX/MathFunctions.h
		eigency/eigen/Eigen/src/Core/arch/AVX/PacketMath.h
		eigency/eigen/Eigen/src/Core/arch/AVX/TypeCasting.h
		eigency/eigen/Eigen/src/Core/arch/AVX512/Complex.h
		eigency/eigen/Eigen/src/Core/arch/AVX512/MathFunctions.h
		eigency/eigen/Eigen/src/Core/arch/AVX512/PacketMath.h
		eigency/eigen/Eigen/src/Core/arch/AVX512/TypeCasting.h
		eigency/eigen/Eigen/src/Core/arch/AltiVec/Complex.h
		eigency/eigen/Eigen/src/Core/arch/AltiVec/MathFunctions.h
		eigency/eigen/Eigen/src/Core/arch/AltiVec/MatrixProduct.h
		eigency/eigen/Eigen/src/Core/arch/AltiVec/MatrixProductCommon.h
		eigency/eigen/Eigen/src/Core/arch/AltiVec/MatrixProductMMA.h
		eigency/eigen/Eigen/src/Core/arch/AltiVec/PacketMath.h
		eigency/eigen/Eigen/src/Core/arch/CUDA/Complex.h
		eigency/eigen/Eigen/src/Core/arch/Default/BFloat16.h
		eigency/eigen/Eigen/src/Core/arch/Default/ConjHelper.h
		eigency/eigen/Eigen/src/Core/arch/Default/GenericPacketMathFunctions.h
		eigency/eigen/Eigen/src/Core/arch/Default/GenericPacketMathFunctionsFwd.h
		eigency/eigen/Eigen/src/Core/arch/Default/Half.h
		eigency/eigen/Eigen/src/Core/arch/Default/Settings.h
		eigency/eigen/Eigen/src/Core/arch/Default/TypeCasting.h
		eigency/eigen/Eigen/src/Core/arch/GPU/MathFunctions.h
		eigency/eigen/Eigen/src/Core/arch/GPU/PacketMath.h
		eigency/eigen/Eigen/src/Core/arch/GPU/TypeCasting.h
		eigency/eigen/Eigen/src/Core/arch/HIP/hcc/math_constants.h
		eigency/eigen/Eigen/src/Core/arch/MSA/Complex.h
		eigency/eigen/Eigen/src/Core/arch/MSA/MathFunctions.h
		eigency/eigen/Eigen/src/Core/arch/MSA/PacketMath.h
		eigency/eigen/Eigen/src/Core/arch/NEON/Complex.h
		eigency/eigen/Eigen/src/Core/arch/NEON/GeneralBlockPanelKernel.h
		eigency/eigen/Eigen/src/Core/arch/NEON/MathFunctions.h
		eigency/eigen/Eigen/src/Core/arch/NEON/PacketMath.h
		eigency/eigen/Eigen/src/Core/arch/NEON/TypeCasting.h
		eigency/eigen/Eigen/src/Core/arch/SSE/Complex.h
		eigency/eigen/Eigen/src/Core/arch/SSE/MathFunctions.h
		eigency/eigen/Eigen/src/Core/arch/SSE/PacketMath.h
		eigency/eigen/Eigen/src/Core/arch/SSE/TypeCasting.h
		eigency/eigen/Eigen/src/Core/arch/SVE/MathFunctions.h
		eigency/eigen/Eigen/src/Core/arch/SVE/PacketMath.h
		eigency/eigen/Eigen/src/Core/arch/SVE/TypeCasting.h
		eigency/eigen/Eigen/src/Core/arch/SYCL/InteropHeaders.h
		eigency/eigen/Eigen/src/Core/arch/SYCL/MathFunctions.h
		eigency/eigen/Eigen/src/Core/arch/SYCL/PacketMath.h
		eigency/eigen/Eigen/src/Core/arch/SYCL/SyclMemoryModel.h
		eigency/eigen/Eigen/src/Core/arch/SYCL/TypeCasting.h
		eigency/eigen/Eigen/src/Core/arch/ZVector/Complex.h
		eigency/eigen/Eigen/src/Core/arch/ZVector/MathFunctions.h
		eigency/eigen/Eigen/src/Core/arch/ZVector/PacketMath.h
		eigency/eigen/Eigen/src/Core/functors/AssignmentFunctors.h
		eigency/eigen/Eigen/src/Core/functors/BinaryFunctors.h
		eigency/eigen/Eigen/src/Core/functors/NullaryFunctors.h
		eigency/eigen/Eigen/src/Core/functors/StlFunctors.h
		eigency/eigen/Eigen/src/Core/functors/TernaryFunctors.h
		eigency/eigen/Eigen/src/Core/functors/UnaryFunctors.h
		eigency/eigen/Eigen/src/Core/products/GeneralBlockPanelKernel.h
		eigency/eigen/Eigen/src/Core/products/GeneralMatrixMatrix.h
		eigency/eigen/Eigen/src/Core/products/GeneralMatrixMatrixTriangular.h
		eigency/eigen/Eigen/src/Core/products/GeneralMatrixMatrixTriangular_BLAS.h
		eigency/eigen/Eigen/src/Core/products/GeneralMatrixMatrix_BLAS.h
		eigency/eigen/Eigen/src/Core/products/GeneralMatrixVector.h
		eigency/eigen/Eigen/src/Core/products/GeneralMatrixVector_BLAS.h
		eigency/eigen/Eigen/src/Core/products/Parallelizer.h
		eigency/eigen/Eigen/src/Core/products/SelfadjointMatrixMatrix.h
		eigency/eigen/Eigen/src/Core/products/SelfadjointMatrixMatrix_BLAS.h
		eigency/eigen/Eigen/src/Core/products/SelfadjointMatrixVector.h
		eigency/eigen/Eigen/src/Core/products/SelfadjointMatrixVector_BLAS.h
		eigency/eigen/Eigen/src/Core/products/SelfadjointProduct.h
		eigency/eigen/Eigen/src/Core/products/SelfadjointRank2Update.h
		eigency/eigen/Eigen/src/Core/products/TriangularMatrixMatrix.h
		eigency/eigen/Eigen/src/Core/products/TriangularMatrixMatrix_BLAS.h
		eigency/eigen/Eigen/src/Core/products/TriangularMatrixVector.h
		eigency/eigen/Eigen/src/Core/products/TriangularMatrixVector_BLAS.h
		eigency/eigen/Eigen/src/Core/products/TriangularSolverMatrix.h
		eigency/eigen/Eigen/src/Core/products/TriangularSolverMatrix_BLAS.h
		eigency/eigen/Eigen/src/Core/products/TriangularSolverVector.h
		eigency/eigen/Eigen/src/Core/util/BlasUtil.h
		eigency/eigen/Eigen/src/Core/util/ConfigureVectorization.h
		eigency/eigen/Eigen/src/Core/util/Constants.h
		eigency/eigen/Eigen/src/Core/util/DisableStupidWarnings.h
		eigency/eigen/Eigen/src/Core/util/ForwardDeclarations.h
		eigency/eigen/Eigen/src/Core/util/IndexedViewHelper.h
		eigency/eigen/Eigen/src/Core/util/IntegralConstant.h
		eigency/eigen/Eigen/src/Core/util/MKL_support.h
		eigency/eigen/Eigen/src/Core/util/Macros.h
		eigency/eigen/Eigen/src/Core/util/Memory.h
		eigency/eigen/Eigen/src/Core/util/Meta.h
		eigency/eigen/Eigen/src/Core/util/NonMPL2.h
		eigency/eigen/Eigen/src/Core/util/ReenableStupidWarnings.h
		eigency/eigen/Eigen/src/Core/util/ReshapedHelper.h
		eigency/eigen/Eigen/src/Core/util/StaticAssert.h
		eigency/eigen/Eigen/src/Core/util/SymbolicIndex.h
		eigency/eigen/Eigen/src/Core/util/XprHelper.h
		eigency/eigen/Eigen/src/Eigenvalues/ComplexEigenSolver.h
		eigency/eigen/Eigen/src/Eigenvalues/ComplexSchur.h
		eigency/eigen/Eigen/src/Eigenvalues/ComplexSchur_LAPACKE.h
		eigency/eigen/Eigen/src/Eigenvalues/EigenSolver.h
		eigency/eigen/Eigen/src/Eigenvalues/GeneralizedEigenSolver.h
		eigency/eigen/Eigen/src/Eigenvalues/GeneralizedSelfAdjointEigenSolver.h
		eigency/eigen/Eigen/src/Eigenvalues/HessenbergDecomposition.h
		eigency/eigen/Eigen/src/Eigenvalues/MatrixBaseEigenvalues.h
		eigency/eigen/Eigen/src/Eigenvalues/RealQZ.h
		eigency/eigen/Eigen/src/Eigenvalues/RealSchur.h
		eigency/eigen/Eigen/src/Eigenvalues/RealSchur_LAPACKE.h
		eigency/eigen/Eigen/src/Eigenvalues/SelfAdjointEigenSolver.h
		eigency/eigen/Eigen/src/Eigenvalues/SelfAdjointEigenSolver_LAPACKE.h
		eigency/eigen/Eigen/src/Eigenvalues/Tridiagonalization.h
		eigency/eigen/Eigen/src/Geometry/AlignedBox.h
		eigency/eigen/Eigen/src/Geometry/AngleAxis.h
		eigency/eigen/Eigen/src/Geometry/EulerAngles.h
		eigency/eigen/Eigen/src/Geometry/Homogeneous.h
		eigency/eigen/Eigen/src/Geometry/Hyperplane.h
		eigency/eigen/Eigen/src/Geometry/OrthoMethods.h
		eigency/eigen/Eigen/src/Geometry/ParametrizedLine.h
		eigency/eigen/Eigen/src/Geometry/Quaternion.h
		eigency/eigen/Eigen/src/Geometry/Rotation2D.h
		eigency/eigen/Eigen/src/Geometry/RotationBase.h
		eigency/eigen/Eigen/src/Geometry/Scaling.h
		eigency/eigen/Eigen/src/Geometry/Transform.h
		eigency/eigen/Eigen/src/Geometry/Translation.h
		eigency/eigen/Eigen/src/Geometry/Umeyama.h
		eigency/eigen/Eigen/src/Geometry/arch/Geometry_SIMD.h
		eigency/eigen/Eigen/src/Householder/BlockHouseholder.h
		eigency/eigen/Eigen/src/Householder/Householder.h
		eigency/eigen/Eigen/src/Householder/HouseholderSequence.h
		eigency/eigen/Eigen/src/IterativeLinearSolvers/BasicPreconditioners.h
		eigency/eigen/Eigen/src/IterativeLinearSolvers/BiCGSTAB.h
		eigency/eigen/Eigen/src/IterativeLinearSolvers/ConjugateGradient.h
		eigency/eigen/Eigen/src/IterativeLinearSolvers/IncompleteCholesky.h
		eigency/eigen/Eigen/src/IterativeLinearSolvers/IncompleteLUT.h
		eigency/eigen/Eigen/src/IterativeLinearSolvers/IterativeSolverBase.h
		eigency/eigen/Eigen/src/IterativeLinearSolvers/LeastSquareConjugateGradient.h
		eigency/eigen/Eigen/src/IterativeLinearSolvers/SolveWithGuess.h
		eigency/eigen/Eigen/src/Jacobi/Jacobi.h
		eigency/eigen/Eigen/src/KLUSupport/KLUSupport.h
		eigency/eigen/Eigen/src/LU/Determinant.h
		eigency/eigen/Eigen/src/LU/FullPivLU.h
		eigency/eigen/Eigen/src/LU/InverseImpl.h
		eigency/eigen/Eigen/src/LU/PartialPivLU.h
		eigency/eigen/Eigen/src/LU/PartialPivLU_LAPACKE.h
		eigency/eigen/Eigen/src/LU/arch/InverseSize4.h
		eigency/eigen/Eigen/src/MetisSupport/MetisSupport.h
		eigency/eigen/Eigen/src/OrderingMethods/Amd.h
		eigency/eigen/Eigen/src/OrderingMethods/Eigen_Colamd.h
		eigency/eigen/Eigen/src/OrderingMethods/Ordering.h
		eigency/eigen/Eigen/src/PaStiXSupport/PaStiXSupport.h
		eigency/eigen/Eigen/src/PardisoSupport/PardisoSupport.h
		eigency/eigen/Eigen/src/QR/ColPivHouseholderQR.h
		eigency/eigen/Eigen/src/QR/ColPivHouseholderQR_LAPACKE.h
		eigency/eigen/Eigen/src/QR/CompleteOrthogonalDecomposition.h
		eigency/eigen/Eigen/src/QR/FullPivHouseholderQR.h
		eigency/eigen/Eigen/src/QR/HouseholderQR.h
		eigency/eigen/Eigen/src/QR/HouseholderQR_LAPACKE.h
		eigency/eigen/Eigen/src/SPQRSupport/SuiteSparseQRSupport.h
		eigency/eigen/Eigen/src/SVD/BDCSVD.h
		eigency/eigen/Eigen/src/SVD/JacobiSVD.h
		eigency/eigen/Eigen/src/SVD/JacobiSVD_LAPACKE.h
		eigency/eigen/Eigen/src/SVD/SVDBase.h
		eigency/eigen/Eigen/src/SVD/UpperBidiagonalization.h
		eigency/eigen/Eigen/src/SparseCholesky/SimplicialCholesky.h
		eigency/eigen/Eigen/src/SparseCholesky/SimplicialCholesky_impl.h
		eigency/eigen/Eigen/src/SparseCore/AmbiVector.h
		eigency/eigen/Eigen/src/SparseCore/CompressedStorage.h
		eigency/eigen/Eigen/src/SparseCore/ConservativeSparseSparseProduct.h
		eigency/eigen/Eigen/src/SparseCore/MappedSparseMatrix.h
		eigency/eigen/Eigen/src/SparseCore/SparseAssign.h
		eigency/eigen/Eigen/src/SparseCore/SparseBlock.h
		eigency/eigen/Eigen/src/SparseCore/SparseColEtree.h
		eigency/eigen/Eigen/src/SparseCore/SparseCompressedBase.h
		eigency/eigen/Eigen/src/SparseCore/SparseCwiseBinaryOp.h
		eigency/eigen/Eigen/src/SparseCore/SparseCwiseUnaryOp.h
		eigency/eigen/Eigen/src/SparseCore/SparseDenseProduct.h
		eigency/eigen/Eigen/src/SparseCore/SparseDiagonalProduct.h
		eigency/eigen/Eigen/src/SparseCore/SparseDot.h
		eigency/eigen/Eigen/src/SparseCore/SparseFuzzy.h
		eigency/eigen/Eigen/src/SparseCore/SparseMap.h
		eigency/eigen/Eigen/src/SparseCore/SparseMatrix.h
		eigency/eigen/Eigen/src/SparseCore/SparseMatrixBase.h
		eigency/eigen/Eigen/src/SparseCore/SparsePermutation.h
		eigency/eigen/Eigen/src/SparseCore/SparseProduct.h
		eigency/eigen/Eigen/src/SparseCore/SparseRedux.h
		eigency/eigen/Eigen/src/SparseCore/SparseRef.h
		eigency/eigen/Eigen/src/SparseCore/SparseSelfAdjointView.h
		eigency/eigen/Eigen/src/SparseCore/SparseSolverBase.h
		eigency/eigen/Eigen/src/SparseCore/SparseSparseProductWithPruning.h
		eigency/eigen/Eigen/src/SparseCore/SparseTranspose.h
		eigency/eigen/Eigen/src/SparseCore/SparseTriangularView.h
		eigency/eigen/Eigen/src/SparseCore/SparseUtil.h
		eigency/eigen/Eigen/src/SparseCore/SparseVector.h
		eigency/eigen/Eigen/src/SparseCore/SparseView.h
		eigency/eigen/Eigen/src/SparseCore/TriangularSolver.h
		eigency/eigen/Eigen/src/SparseLU/SparseLU.h
		eigency/eigen/Eigen/src/SparseLU/SparseLUImpl.h
		eigency/eigen/Eigen/src/SparseLU/SparseLU_Memory.h
		eigency/eigen/Eigen/src/SparseLU/SparseLU_Structs.h
		eigency/eigen/Eigen/src/SparseLU/SparseLU_SupernodalMatrix.h
		eigency/eigen/Eigen/src/SparseLU/SparseLU_Utils.h
		eigency/eigen/Eigen/src/SparseLU/SparseLU_column_bmod.h
		eigency/eigen/Eigen/src/SparseLU/SparseLU_column_dfs.h
		eigency/eigen/Eigen/src/SparseLU/SparseLU_copy_to_ucol.h
		eigency/eigen/Eigen/src/SparseLU/SparseLU_gemm_kernel.h
		eigency/eigen/Eigen/src/SparseLU/SparseLU_heap_relax_snode.h
		eigency/eigen/Eigen/src/SparseLU/SparseLU_kernel_bmod.h
		eigency/eigen/Eigen/src/SparseLU/SparseLU_panel_bmod.h
		eigency/eigen/Eigen/src/SparseLU/SparseLU_panel_dfs.h
		eigency/eigen/Eigen/src/SparseLU/SparseLU_pivotL.h
		eigency/eigen/Eigen/src/SparseLU/SparseLU_pruneL.h
		eigency/eigen/Eigen/src/SparseLU/SparseLU_relax_snode.h
		eigency/eigen/Eigen/src/SparseQR/SparseQR.h
		eigency/eigen/Eigen/src/StlSupport/StdDeque.h
		eigency/eigen/Eigen/src/StlSupport/StdList.h
		eigency/eigen/Eigen/src/StlSupport/StdVector.h
		eigency/eigen/Eigen/src/StlSupport/details.h
		eigency/eigen/Eigen/src/SuperLUSupport/SuperLUSupport.h
		eigency/eigen/Eigen/src/UmfPackSupport/UmfPackSupport.h
		eigency/eigen/Eigen/src/misc/Image.h
		eigency/eigen/Eigen/src/misc/Kernel.h
		eigency/eigen/Eigen/src/misc/RealSvd2x2.h
		eigency/eigen/Eigen/src/misc/blas.h
		eigency/eigen/Eigen/src/misc/lapack.h
		eigency/eigen/Eigen/src/misc/lapacke.h
		eigency/eigen/Eigen/src/misc/lapacke_mangling.h
		eigency/eigen/Eigen/src/plugins/ArrayCwiseBinaryOps.h
		eigency/eigen/Eigen/src/plugins/ArrayCwiseUnaryOps.h
		eigency/eigen/Eigen/src/plugins/BlockMethods.h
		eigency/eigen/Eigen/src/plugins/CommonCwiseBinaryOps.h
		eigency/eigen/Eigen/src/plugins/CommonCwiseUnaryOps.h
		eigency/eigen/Eigen/src/plugins/IndexedViewMethods.h
		eigency/eigen/Eigen/src/plugins/MatrixCwiseBinaryOps.h
		eigency/eigen/Eigen/src/plugins/MatrixCwiseUnaryOps.h
		eigency/eigen/Eigen/src/plugins/ReshapedMethods.h

+1

eigency.egg-info/top_level.txt

eigency

+20

LICENSE

		Copyright (c) 2016 Wouter Boomsma

		Permission is hereby granted, free of charge, to any person obtaining
		a copy of this software and associated documentation files (the
		"Software"), to deal in the Software without restriction, including
		without limitation the rights to use, copy, modify, merge, publish,
		distribute, sublicense, and/or sell copies of the Software, and to
		permit persons to whom the Software is furnished to do so, subject to
		the following conditions:

		The above copyright notice and this permission notice shall be
		included in all copies or substantial portions of the Software.

		THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
		EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
		MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
		NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
		LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
		OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
		WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

+566

PKG-INFO

		Metadata-Version: 2.4
		Name: eigency
		Version: 3.4.0.6
		Summary: Cython interface between the numpy arrays and the Matrix/Array classes of the Eigen C++ library
		Author-email: Wouter Boomsma <wb@di.ku.dk>, Felipe Bordeu <felipebordeu@gmail.com>
		Maintainer-email: Felipe Bordeu <felipebordeu@gmail.com>, Wouter Boomsma <wb@di.ku.dk>
		License-Expression: LicenseRef-Wouter-Boomsma-1.0
		Project-URL: Homepage, https://github.com/eigency-org/eigency
		Classifier: Intended Audience :: Developers
		Classifier: Programming Language :: C++
		Classifier: Programming Language :: Cython
		Classifier: Programming Language :: Python :: 3.9
		Classifier: Programming Language :: Python :: 3.10
		Classifier: Programming Language :: Python :: 3.11
		Classifier: Programming Language :: Python :: 3.12
		Classifier: Programming Language :: Python :: 3.13
		Requires-Python: >=3.9
		Description-Content-Type: text/markdown
		License-File: LICENSE
		Requires-Dist: numpy>=1.23.5
		Dynamic: license-file

		# Eigency

		[![PyPI version](https://badge.fury.io/py/eigency.svg)](https://badge.fury.io/py/eigency)
		[![PEP 517](https://github.com/eigency-org/eigency/actions/workflows/build.yml/badge.svg)](https://github.com/eigency-org/eigency/actions/workflows/build.yml)
		[![pip wheel](https://github.com/eigency-org/eigency/actions/workflows/wheel.yml/badge.svg)](https://github.com/eigency-org/eigency/actions/workflows/wheel.yml)
		[![setup.py](https://github.com/eigency-org/eigency/actions/workflows/setup.yml/badge.svg)](https://github.com/eigency-org/eigency/actions/workflows/setup.yml)
		[![pre-commit](https://github.com/eigency-org/eigency/actions/workflows/pre-commit.yml/badge.svg)](https://github.com/eigency-org/eigency/actions/workflows/pre-commit.yml)

		Eigency is a Cython interface between Numpy arrays and Matrix/Array
		objects from the Eigen C++ library. It is intended to simplify the
		process of writing C++ extensions using the Eigen library. Eigency is
		designed to reuse the underlying storage of the arrays when passing
		data back and forth, and will thus avoid making unnecessary copies
		whenever possible. Only in cases where copies are explicitly requested
		by your C++ code will they be made.

		## Versioning

		Eigency uses a 4-number version (`N.N.N.N`) where the first 3 parts correspond to the embedded Eigen library version.
		The last part is a revision number of Eigency itself.

		## Installing

		Eigency is packaged as a source distribution (`sdist`) and available on PyPi.
		It can be easily installed using `pip`:
		```bash
		python -m pip install eigency
		```

		Requirement: `pip >= 18.0`

		If your `pip` is too old, then upgrade it using:
		```bash
		python -m pip install --upgrade pip
		```

		## Contributing

		For instructions on building and/or packaging Eigency from source,
		see the contributing guide [here](https://github.com/eigency-org/eigency/blob/master/CONTRIBUTING.md).

		## Usage

		Below is a description of a range of common usage scenarios. A full working
		example of both setup and these different use cases is available in the
		`test` directory distributed with the this package.

		### Setup

		To import eigency functionality, add the following to your `.pyx` file:
		```
		from eigency.core cimport *
		```

		In addition, in the `setup.py` file, the include directories must be
		set up to include the eigency includes. This can be done by calling
		the `get_includes` function in the `eigency` module:
		```
		import eigency
		...
		extensions = [
		Extension("module-dir-name/module-name", ["module-dir-name/module-name.pyx"],
		include_dirs = [".", "module-dir-name"] + eigency.get_includes()
		),
		]
		```

		Eigency includes a version of the Eigen library, and the `get_includes` function will include the path to this directory. If you
		have your own version of Eigen, just set the `include_eigen` option to False, and add your own path instead:
		```
		include_dirs = [".", "module-dir-name", 'path-to-own-eigen'] + eigency.get_includes(include_eigen=False)
		```

		### From Numpy to Eigen

		Assume we are writing a Cython interface to the following C++ function:

		```c++
		void function_w_mat_arg(const Eigen::Map<Eigen::MatrixXd> &mat) {
		std::cout << mat << "\n";
		}
		```

		Note that we use `Eigen::Map` to ensure that we can reuse the storage
		of the numpy array, thus avoiding making a copy. Assuming the C++ code
		is in a file called `functions.h`, the corresponding `.pyx` entry could look like this:

		```
		cdef extern from "functions.h":
		cdef void _function_w_mat_arg "function_w_mat_arg"(Map[MatrixXd] &)

		# This will be exposed to Python
		def function_w_mat_arg(np.ndarray array):
		return _function_w_mat_arg(Map[MatrixXd](array))
		```

		The last line contains the actual conversion. `Map` is an Eigency
		type that derives from the real Eigen map, and will take care of
		the conversion from the numpy array to the corresponding Eigen type.

		We can now call the C++ function directly from Python:
		```python
		>>> import numpy as np
		>>> import eigency_tests
		>>> x = np.array([[1.1, 2.2], [3.3, 4.4]])
		>>> eigency_tests.function_w_mat_arg(x)
		1.1 3.3
		2.2 4.4
		```
		(if you are wondering about why the matrix is transposed, please
		see the Storage layout section below).

		## Types matter

		The basic idea behind eigency is to share the underlying representation of a
		numpy array between Python and C++. This means that somewhere in the process,
		we need to make explicit which numerical types we are dealing with. In the
		function above, we specify that we expect an Eigen MatrixXd, which means
		that the numpy array must also contain double (i.e. float64) values. If we instead provide
		a numpy array of ints, we will get strange results.

		```python
		>>> import numpy as np
		>>> import eigency_tests
		>>> x = np.array([[1, 2], [3, 4]])
		>>> eigency_tests.function_w_mat_arg(x)
		4.94066e-324 1.4822e-323
		9.88131e-324 1.97626e-323
		```
		This is because we are explicitly asking C++ to interpret out python integer
		values as floats.

		To avoid this type of error, you can force your cython function to
		accept only numpy arrays of a specific type:

		```
		cdef extern from "functions.h":
		cdef void _function_w_mat_arg "function_w_mat_arg"(Map[MatrixXd] &)

		# This will be exposed to Python
		def function_w_mat_arg(np.ndarray[np.float64_t, ndim=2] array):
		return _function_w_mat_arg(Map[MatrixXd](array))
		```

		(Note that when using this technique to select the type, you also need to specify
		the dimensions of the array (this will default to 1)). Using this new definition,
		users will get an error when passing arrays of the wrong type:

		```python
		>>> import numpy as np
		>>> import eigency_tests
		>>> x = np.array([[1, 2], [3, 4]])
		>>> eigency_tests.function_w_mat_arg(x)
		Traceback (most recent call last):
		File "<stdin>", line 1, in <module>
		File "eigency_tests/eigency_tests.pyx", line 87, in eigency_tests.eigency_tests.function_w_mat_arg
		ValueError: Buffer dtype mismatch, expected 'float64_t' but got 'long'
		```

		Since it avoids many surprises, it is strongly recommended to use this technique
		to specify the full types of numpy arrays in your cython code whenever
		possible.

		### Writing Eigen Map types in Cython

		Since Cython does not support nested fused types, you cannot write types like `Map[Matrix[double, 2, 2]]`. In most cases, you won't need to, since you can just use Eigens convenience typedefs, such as `Map[VectorXd]`. If you need the additional flexibility of the full specification, you can use the `FlattenedMap` type, where all type arguments can be specified at top level, for instance `FlattenedMap[Matrix, double, _2, _3]` or `FlattenedMap[Matrix, double, _2, Dynamic]`. Note that dimensions must be prefixed with an underscore.

		Using full specifications of the Eigen types, the previous example would look like this:
		```
		cdef extern from "functions.h":
		cdef void _function_w_mat_arg "function_w_mat_arg" (FlattenedMap[Matrix, double, Dynamic, Dynamic] &)

		# This will be exposed to Python
		def function_w_mat_arg(np.ndarray[np.float64_t, ndim=2] array):
		return _function_w_mat_arg(FlattenedMap[Matrix, double, Dynamic, Dynamic](array))
		```

		`FlattenedType` takes four template parameters: arraytype, scalartype,
		rows and cols. Eigen supports a few other template arguments for
		setting the storage layout and Map strides. Since cython does not
		support default template arguments for fused types, we have instead
		defined separate types for this purpose. These are called
		`FlattenedMapWithOrder` and `FlattenedMapWithStride` with five and eight
		template arguments, respectively. For details on their use, see the section
		about storage layout below.

		### From Numpy to Eigen (insisting on a copy)

		Eigency will not complain if the C++ function you interface with does
		not take a Eigen Map object, but instead a regular Eigen Matrix or
		Array. However, in such cases, a copy will be made. Actually, the
		procedure is exactly the same as above. In the `.pyx` file, you still
		define everything exactly the same way as for the Map case described above.

		For instance, given the following C++ function:
		```c++
		void function_w_vec_arg_no_map(const Eigen::VectorXd &vec);
		```

		The Cython definitions would still look like this:

		```
		cdef extern from "functions.h":
		cdef void _function_w_vec_arg_no_map "function_w_vec_arg_no_map"(Map[VectorXd] &)

		# This will be exposed to Python
		def function_w_vec_arg_no_map(np.ndarray[np.float64_t] array):
		return _function_w_vec_arg_no_map(Map[VectorXd](array))
		```

		Cython will not mind the fact that the argument type in the extern
		declaration (a Map type) differs from the actual one in the `.h` file,
		as long as one can be assigned to the other. Since Map objects can be
		assigned to their corresponding Matrix/Array types this works
		seemlessly. But keep in mind that this assignment will make a copy of
		the underlying data.

		### Eigen to Numpy

		C++ functions returning a reference to an Eigen Matrix/Array can also
		be transferred to numpy arrays without copying their content. Assume
		we have a class with a single getter function that returns an Eigen
		matrix member:

		```c++
		class MyClass {
		public:
		MyClass():
		matrix(Eigen::Matrix3d::Constant(3.)) {
		}
		Eigen::MatrixXd &get_matrix() {
		return this->matrix;
		}
		private:
		Eigen::Matrix3d matrix;
		};
		```

		The Cython C++ class interface is specified as usual:

		```
		cdef cppclass _MyClass "MyClass":
		_MyClass "MyClass"() except +
		Matrix3d &get_matrix()
		```

		And the corresponding Python wrapper:

		```python
		cdef class MyClass:
		cdef _MyClass *thisptr;

		def __cinit__(self):
		self.thisptr = new _MyClass()

		def __dealloc__(self):
		del self.thisptr

		def get_matrix(self):
		return ndarray(self.thisptr.get_matrix())
		```

		This last line contains the actual conversion. Again, eigency has its
		own version of `ndarray`, that will take care of the conversion for
		you.

		Due to limitations in Cython, Eigency cannot deal with full
		Matrix/Array template specifications as return types
		(e.g. `Matrix[double, 4, 2]`). However, as a workaround, you can use
		`PlainObjectBase` as a return type in such cases (or in all cases if
		you prefer):

		```
		PlainObjectBase &get_matrix()
		```

		### Overriding default behavior

		The `ndarray` conversion type specifier will attempt do guess whether you want a copy
		or a view, depending on the return type. Most of the time, this is
		probably what you want. However, there might be cases where you want
		to override this behavior. For instance, functions returning const
		references will result in a copy of the array, since the const-ness
		cannot be enforced in Python. However, you can always override the
		default behavior by using the `ndarray_copy` or `ndarray_view`
		functions.

		Expanding the `MyClass` example from before:

		```c++
		class MyClass {
		public:
		...
		const Eigen::MatrixXd &get_const_matrix() {
		return this->matrix;
		}
		...
		};
		```

		With the corresponding cython interface specification
		The Cython C++ class interface is specified as usual:

		```
		cdef cppclass _MyClass "MyClass":
		...
		const Matrix3d &get_const_matrix()
		```

		The following would return a copy

		```python
		cdef class MyClass:
		...
		def get_const_matrix(self):
		return ndarray(self.thisptr.get_const_matrix())
		```

		while the following would force it to return a view

		```python
		cdef class MyClass:
		...
		def get_const_matrix(self):
		return ndarray_view(self.thisptr.get_const_matrix())
		```

		### Eigen to Numpy (non-reference return values)

		Functions returning an Eigen object (not a reference), are specified
		in a similar way. For instance, given the following C++ function:

		```c++
		Eigen::Matrix3d function_w_mat_retval();
		```

		The Cython code could be written as:

		```
		cdef extern from "functions.h":
		cdef Matrix3d _function_w_mat_retval "function_w_mat_retval" ()

		# This will be exposed to Python
		def function_w_mat_retval():
		return ndarray_copy(_function_w_mat_retval())
		```

		As mentioned above, you can replace `Matrix3d` (or any other Eigen return type) with
		`PlainObjectBase`, which is especially relevant when working with
		Eigen object that do not have an associated convenience typedef.

		Note that we use `ndarray_copy` instead of `ndarray` to explicitly
		state that a copy should be made. In c++11 compliant compilers, it
		will detect the rvalue reference and automatically make a copy even if
		you just use `ndarray` (see next section), but to ensure that it works
		also with older compilers it is recommended to always use
		`ndarray_copy` when returning newly constructed eigen values.


		### Corrupt data when returning non-map types

		The tendency of Eigency to avoid copies whenever possible can lead
		to corrupted data when returning non-map Eigen arrays. For instance,
		in the `function_w_mat_retval` from the previous section, a temporary
		value will be returned from C++, and we have to take care to make
		a copy of this data instead of letting the resulting numpy array
		refer directly to this memory. In C++11, this situation can be
		detected directly using rvalue references, and it will therefore
		automatically make a copy:
		```
		def function_w_mat_retval():
		# This works in C++11, because it detects the rvalue reference
		return ndarray(_function_w_mat_retval())
		```

		However, to make sure it works with older compilers,
		it is recommended to use the `ndarray_copy` conversion:

		```
		def function_w_mat_retval():
		# Explicit request for copy - this always works
		return ndarray_copy(_function_w_mat_retval())
		```

		### Storage layout - why arrays are sometimes transposed

		The default storage layout used in numpy and Eigen differ. Numpy uses
		a row-major layout (C-style) per default while Eigen uses a
		column-major layout (Fortran style) by default. In Eigency, we prioritize to
		avoid copying of data whenever possible, which can have unexpected
		consequences in some cases: There is no problem when passing values
		from C++ to Python - we just adjust the storage layout of the returned
		numpy array to match that of Eigen. However, since the storage layout
		is encoded into the _type_ of the Eigen array (or the type of the
		Map), we cannot automatically change the layout in the Python to C++ direction. In
		Eigency, we have therefore opted to return the transposed array/matrix
		in such cases. This provides the user with the flexibility to deal
		with the problem either in Python (use order="F" when constructing
		your numpy array), or on the C++ side: (1) explicitly define your
		argument to have the row-major storage layout, 2) manually set the Map
		stride, or 3) just call `.transpose()` on the received
		array/matrix).

		As an example, consider the case of a C++ function that both receives
		and returns a Eigen Map type, thus acting as a filter:

		```c++
		Eigen::Map<Eigen::ArrayXXd> function_filter(Eigen::Map<Eigen::ArrayXXd> &mat) {
		return mat;
		}
		```

		The Cython code could be:

		```
		cdef extern from "functions.h":
		...
		cdef Map[ArrayXXd] &_function_filter1 "function_filter1" (Map[ArrayXXd] &)

		def function_filter1(np.ndarray[np.float64_t, ndim=2] array):
		return ndarray(_function_filter1(Map[ArrayXXd](array)))

		```

		If we call this function from Python in the standard way, we will
		see that the array is transposed on the way from Python to C++, and
		remains that way when it is again returned to Python:

		```python
		>>> x = np.array([[1., 2., 3., 4.], [5., 6., 7., 8.]])
		>>> y = function_filter1(x)
		>>> print x
		[[ 1. 2. 3. 4.]
		[ 5. 6. 7. 8.]]
		>>> print y
		[[ 1. 5.]
		[ 2. 6.]
		[ 3. 7.]
		[ 4. 8.]]
		```

		The simplest way to avoid this is to tell numpy to use a
		column-major array layout instead of the default row-major
		layout. This can be done using the order='F' option:

		```python
		>>> x = np.array([[1., 2., 3., 4.], [5., 6., 7., 8.]], order='F')
		>>> y = function_filter1(x)
		>>> print x
		[[ 1. 2. 3. 4.]
		[ 5. 6. 7. 8.]]
		>>> print y
		[[ 1. 2. 3. 4.]
		[ 5. 6. 7. 8.]]
		```

		The other alternative is to tell Eigen to use RowMajor layout. This
		requires changing the C++ function definition:

		```c++
		typedef Eigen::Map<Eigen::Array<double, Eigen::Dynamic, Eigen::Dynamic, Eigen::RowMajor> > RowMajorArrayMap;

		RowMajorArrayMap &function_filter2(RowMajorArrayMap &mat) {
		return mat;
		}
		```

		To write the corresponding Cython definition, we need the expanded version of
		`FlattenedMap` called `FlattenedMapWithOrder`, which allows us to specify
		the storage order:

		```
		cdef extern from "functions.h":
		...
		cdef PlainObjectBase _function_filter2 "function_filter2" (FlattenedMapWithOrder[Array, double, Dynamic, Dynamic, RowMajor])

		def function_filter2(np.ndarray[np.float64_t, ndim=2] array):
		return ndarray(_function_filter2(FlattenedMapWithOrder[Array, double, Dynamic, Dynamic, RowMajor](array)))
		```

		Another alternative is to keep the array itself in RowMajor format,
		but use different stride values for the Map type:

		```c++
		typedef Eigen::Map<Eigen::ArrayXXd, Eigen::Unaligned, Eigen::Stride<1, Eigen::Dynamic> > CustomStrideMap;

		CustomStrideMap &function_filter3(CustomStrideMap &);
		```

		In this case, in Cython, we need to use the even more extended
		`FlattenedMap` type called `FlattenedMapWithStride`, taking eight
		arguments:

		```
		cdef extern from "functions.h":
		...
		cdef PlainObjectBase _function_filter3 "function_filter3" (FlattenedMapWithStride[Array, double, Dynamic, Dynamic, ColMajor, Unaligned, _1, Dynamic])

		def function_filter3(np.ndarray[np.float64_t, ndim=2] array):
		return ndarray(_function_filter3(FlattenedMapWithStride[Array, double, Dynamic, Dynamic, ColMajor, Unaligned, _1, Dynamic](array)))
		```

		In all three cases, the returned array will now be of the same shape
		as the original.

		### Long double support

		Eigency provides new shorthands for Eigen `long double` and complex `long double` Matrix and Array types.
		Examples:
		```
		Vector4ld
		Matrix3ld
		Vector2cld
		Matrix4cld
		Array3Xld
		ArrayXXcld
		```

		These typedefs are available in the `eigency` namespace when including the `eigency` header:
		```c++
		#include "eigency.h"

		void receive_long_double_matrix(Eigen::Map<eigency::MatrixXld> &mat) {
		// use long double eigen matrix
		}
		```

		Use Cython (.pyx) to create Python binding to your C++ function:
		```python
		cdef extern from "functions.h":
		cdef void _receive_long_double_matrix "receive_long_double_matrix"(Map[MatrixXld] &)

		def send_long_double_ndarray(np.ndarray[np.longdouble_t, ndim=2] array):
		return _receive_long_double_matrix(Map[MatrixXld](array))
		```

		Invoke in Python:
		```python
		import numpy as np
		import my_module

		x = np.array([[1.1, 2.2], [3.3, 4.4]], dtype=np.longdouble)
		my_module.send_long_double_ndarray(x)
		```

+46

pyproject.toml

		[build-system]
		requires = [
		"setuptools>=60.0.0",
		"setuptools_scm>=6.4.0",
		"oldest-supported-numpy; python_version<='3.8'",
		"numpy>=2.0.0; python_version>'3.8'",
		"Cython>=3.0.0"
		]
		build-backend = "setuptools.build_meta"

		[project]
		name = "eigency"
		dynamic = ["version"]
		description = "Cython interface between the numpy arrays and the Matrix/Array classes of the Eigen C++ library"

		readme = "README.md"

		requires-python = ">=3.9"
		license = "LicenseRef-Wouter-Boomsma-1.0"
		license-files = ["LICENSE"]

		dependencies = [
		"numpy>=1.23.5",
		]

		authors = [
		{name = "Wouter Boomsma", email = "wb@di.ku.dk"},
		{name = "Felipe Bordeu", email = "felipebordeu@gmail.com"}
		]
		maintainers = [
		{name = "Felipe Bordeu", email = "felipebordeu@gmail.com"},
		{name = "Wouter Boomsma", email = "wb@di.ku.dk"}
		]

		classifiers = [
		"Intended Audience :: Developers",
		"Programming Language :: C++",
		"Programming Language :: Cython",
		"Programming Language :: Python :: 3.9",
		"Programming Language :: Python :: 3.10",
		"Programming Language :: Python :: 3.11",
		"Programming Language :: Python :: 3.12",
		"Programming Language :: Python :: 3.13",
		]
		[project.urls]
		Homepage = "https://github.com/eigency-org/eigency"

+544

README.md

		# Eigency

		[![PyPI version](https://badge.fury.io/py/eigency.svg)](https://badge.fury.io/py/eigency)
		[![PEP 517](https://github.com/eigency-org/eigency/actions/workflows/build.yml/badge.svg)](https://github.com/eigency-org/eigency/actions/workflows/build.yml)
		[![pip wheel](https://github.com/eigency-org/eigency/actions/workflows/wheel.yml/badge.svg)](https://github.com/eigency-org/eigency/actions/workflows/wheel.yml)
		[![setup.py](https://github.com/eigency-org/eigency/actions/workflows/setup.yml/badge.svg)](https://github.com/eigency-org/eigency/actions/workflows/setup.yml)
		[![pre-commit](https://github.com/eigency-org/eigency/actions/workflows/pre-commit.yml/badge.svg)](https://github.com/eigency-org/eigency/actions/workflows/pre-commit.yml)

		Eigency is a Cython interface between Numpy arrays and Matrix/Array
		objects from the Eigen C++ library. It is intended to simplify the
		process of writing C++ extensions using the Eigen library. Eigency is
		designed to reuse the underlying storage of the arrays when passing
		data back and forth, and will thus avoid making unnecessary copies
		whenever possible. Only in cases where copies are explicitly requested
		by your C++ code will they be made.

		## Versioning

		Eigency uses a 4-number version (`N.N.N.N`) where the first 3 parts correspond to the embedded Eigen library version.
		The last part is a revision number of Eigency itself.

		## Installing

		Eigency is packaged as a source distribution (`sdist`) and available on PyPi.
		It can be easily installed using `pip`:
		```bash
		python -m pip install eigency
		```

		Requirement: `pip >= 18.0`

		If your `pip` is too old, then upgrade it using:
		```bash
		python -m pip install --upgrade pip
		```

		## Contributing

		For instructions on building and/or packaging Eigency from source,
		see the contributing guide [here](https://github.com/eigency-org/eigency/blob/master/CONTRIBUTING.md).

		## Usage

		Below is a description of a range of common usage scenarios. A full working
		example of both setup and these different use cases is available in the
		`test` directory distributed with the this package.

		### Setup

		To import eigency functionality, add the following to your `.pyx` file:
		```
		from eigency.core cimport *
		```

		In addition, in the `setup.py` file, the include directories must be
		set up to include the eigency includes. This can be done by calling
		the `get_includes` function in the `eigency` module:
		```
		import eigency
		...
		extensions = [
		Extension("module-dir-name/module-name", ["module-dir-name/module-name.pyx"],
		include_dirs = [".", "module-dir-name"] + eigency.get_includes()
		),
		]
		```

		Eigency includes a version of the Eigen library, and the `get_includes` function will include the path to this directory. If you
		have your own version of Eigen, just set the `include_eigen` option to False, and add your own path instead:
		```
		include_dirs = [".", "module-dir-name", 'path-to-own-eigen'] + eigency.get_includes(include_eigen=False)
		```

		### From Numpy to Eigen

		Assume we are writing a Cython interface to the following C++ function:

		```c++
		void function_w_mat_arg(const Eigen::Map<Eigen::MatrixXd> &mat) {
		std::cout << mat << "\n";
		}
		```

		Note that we use `Eigen::Map` to ensure that we can reuse the storage
		of the numpy array, thus avoiding making a copy. Assuming the C++ code
		is in a file called `functions.h`, the corresponding `.pyx` entry could look like this:

		```
		cdef extern from "functions.h":
		cdef void _function_w_mat_arg "function_w_mat_arg"(Map[MatrixXd] &)

		# This will be exposed to Python
		def function_w_mat_arg(np.ndarray array):
		return _function_w_mat_arg(Map[MatrixXd](array))
		```

		The last line contains the actual conversion. `Map` is an Eigency
		type that derives from the real Eigen map, and will take care of
		the conversion from the numpy array to the corresponding Eigen type.

		We can now call the C++ function directly from Python:
		```python
		>>> import numpy as np
		>>> import eigency_tests
		>>> x = np.array([[1.1, 2.2], [3.3, 4.4]])
		>>> eigency_tests.function_w_mat_arg(x)
		1.1 3.3
		2.2 4.4
		```
		(if you are wondering about why the matrix is transposed, please
		see the Storage layout section below).

		## Types matter

		The basic idea behind eigency is to share the underlying representation of a
		numpy array between Python and C++. This means that somewhere in the process,
		we need to make explicit which numerical types we are dealing with. In the
		function above, we specify that we expect an Eigen MatrixXd, which means
		that the numpy array must also contain double (i.e. float64) values. If we instead provide
		a numpy array of ints, we will get strange results.

		```python
		>>> import numpy as np
		>>> import eigency_tests
		>>> x = np.array([[1, 2], [3, 4]])
		>>> eigency_tests.function_w_mat_arg(x)
		4.94066e-324 1.4822e-323
		9.88131e-324 1.97626e-323
		```
		This is because we are explicitly asking C++ to interpret out python integer
		values as floats.

		To avoid this type of error, you can force your cython function to
		accept only numpy arrays of a specific type:

		```
		cdef extern from "functions.h":
		cdef void _function_w_mat_arg "function_w_mat_arg"(Map[MatrixXd] &)

		# This will be exposed to Python
		def function_w_mat_arg(np.ndarray[np.float64_t, ndim=2] array):
		return _function_w_mat_arg(Map[MatrixXd](array))
		```

		(Note that when using this technique to select the type, you also need to specify
		the dimensions of the array (this will default to 1)). Using this new definition,
		users will get an error when passing arrays of the wrong type:

		```python
		>>> import numpy as np
		>>> import eigency_tests
		>>> x = np.array([[1, 2], [3, 4]])
		>>> eigency_tests.function_w_mat_arg(x)
		Traceback (most recent call last):
		File "<stdin>", line 1, in <module>
		File "eigency_tests/eigency_tests.pyx", line 87, in eigency_tests.eigency_tests.function_w_mat_arg
		ValueError: Buffer dtype mismatch, expected 'float64_t' but got 'long'
		```

		Since it avoids many surprises, it is strongly recommended to use this technique
		to specify the full types of numpy arrays in your cython code whenever
		possible.

		### Writing Eigen Map types in Cython

		Since Cython does not support nested fused types, you cannot write types like `Map[Matrix[double, 2, 2]]`. In most cases, you won't need to, since you can just use Eigens convenience typedefs, such as `Map[VectorXd]`. If you need the additional flexibility of the full specification, you can use the `FlattenedMap` type, where all type arguments can be specified at top level, for instance `FlattenedMap[Matrix, double, _2, _3]` or `FlattenedMap[Matrix, double, _2, Dynamic]`. Note that dimensions must be prefixed with an underscore.

		Using full specifications of the Eigen types, the previous example would look like this:
		```
		cdef extern from "functions.h":
		cdef void _function_w_mat_arg "function_w_mat_arg" (FlattenedMap[Matrix, double, Dynamic, Dynamic] &)

		# This will be exposed to Python
		def function_w_mat_arg(np.ndarray[np.float64_t, ndim=2] array):
		return _function_w_mat_arg(FlattenedMap[Matrix, double, Dynamic, Dynamic](array))
		```

		`FlattenedType` takes four template parameters: arraytype, scalartype,
		rows and cols. Eigen supports a few other template arguments for
		setting the storage layout and Map strides. Since cython does not
		support default template arguments for fused types, we have instead
		defined separate types for this purpose. These are called
		`FlattenedMapWithOrder` and `FlattenedMapWithStride` with five and eight
		template arguments, respectively. For details on their use, see the section
		about storage layout below.

		### From Numpy to Eigen (insisting on a copy)

		Eigency will not complain if the C++ function you interface with does
		not take a Eigen Map object, but instead a regular Eigen Matrix or
		Array. However, in such cases, a copy will be made. Actually, the
		procedure is exactly the same as above. In the `.pyx` file, you still
		define everything exactly the same way as for the Map case described above.

		For instance, given the following C++ function:
		```c++
		void function_w_vec_arg_no_map(const Eigen::VectorXd &vec);
		```

		The Cython definitions would still look like this:

		```
		cdef extern from "functions.h":
		cdef void _function_w_vec_arg_no_map "function_w_vec_arg_no_map"(Map[VectorXd] &)

		# This will be exposed to Python
		def function_w_vec_arg_no_map(np.ndarray[np.float64_t] array):
		return _function_w_vec_arg_no_map(Map[VectorXd](array))
		```

		Cython will not mind the fact that the argument type in the extern
		declaration (a Map type) differs from the actual one in the `.h` file,
		as long as one can be assigned to the other. Since Map objects can be
		assigned to their corresponding Matrix/Array types this works
		seemlessly. But keep in mind that this assignment will make a copy of
		the underlying data.

		### Eigen to Numpy

		C++ functions returning a reference to an Eigen Matrix/Array can also
		be transferred to numpy arrays without copying their content. Assume
		we have a class with a single getter function that returns an Eigen
		matrix member:

		```c++
		class MyClass {
		public:
		MyClass():
		matrix(Eigen::Matrix3d::Constant(3.)) {
		}
		Eigen::MatrixXd &get_matrix() {
		return this->matrix;
		}
		private:
		Eigen::Matrix3d matrix;
		};
		```

		The Cython C++ class interface is specified as usual:

		```
		cdef cppclass _MyClass "MyClass":
		_MyClass "MyClass"() except +
		Matrix3d &get_matrix()
		```

		And the corresponding Python wrapper:

		```python
		cdef class MyClass:
		cdef _MyClass *thisptr;

		def __cinit__(self):
		self.thisptr = new _MyClass()

		def __dealloc__(self):
		del self.thisptr

		def get_matrix(self):
		return ndarray(self.thisptr.get_matrix())
		```

		This last line contains the actual conversion. Again, eigency has its
		own version of `ndarray`, that will take care of the conversion for
		you.

		Due to limitations in Cython, Eigency cannot deal with full
		Matrix/Array template specifications as return types
		(e.g. `Matrix[double, 4, 2]`). However, as a workaround, you can use
		`PlainObjectBase` as a return type in such cases (or in all cases if
		you prefer):

		```
		PlainObjectBase &get_matrix()
		```

		### Overriding default behavior

		The `ndarray` conversion type specifier will attempt do guess whether you want a copy
		or a view, depending on the return type. Most of the time, this is
		probably what you want. However, there might be cases where you want
		to override this behavior. For instance, functions returning const
		references will result in a copy of the array, since the const-ness
		cannot be enforced in Python. However, you can always override the
		default behavior by using the `ndarray_copy` or `ndarray_view`
		functions.

		Expanding the `MyClass` example from before:

		```c++
		class MyClass {
		public:
		...
		const Eigen::MatrixXd &get_const_matrix() {
		return this->matrix;
		}
		...
		};
		```

		With the corresponding cython interface specification
		The Cython C++ class interface is specified as usual:

		```
		cdef cppclass _MyClass "MyClass":
		...
		const Matrix3d &get_const_matrix()
		```

		The following would return a copy

		```python
		cdef class MyClass:
		...
		def get_const_matrix(self):
		return ndarray(self.thisptr.get_const_matrix())
		```

		while the following would force it to return a view

		```python
		cdef class MyClass:
		...
		def get_const_matrix(self):
		return ndarray_view(self.thisptr.get_const_matrix())
		```

		### Eigen to Numpy (non-reference return values)

		Functions returning an Eigen object (not a reference), are specified
		in a similar way. For instance, given the following C++ function:

		```c++
		Eigen::Matrix3d function_w_mat_retval();
		```

		The Cython code could be written as:

		```
		cdef extern from "functions.h":
		cdef Matrix3d _function_w_mat_retval "function_w_mat_retval" ()

		# This will be exposed to Python
		def function_w_mat_retval():
		return ndarray_copy(_function_w_mat_retval())
		```

		As mentioned above, you can replace `Matrix3d` (or any other Eigen return type) with
		`PlainObjectBase`, which is especially relevant when working with
		Eigen object that do not have an associated convenience typedef.

		Note that we use `ndarray_copy` instead of `ndarray` to explicitly
		state that a copy should be made. In c++11 compliant compilers, it
		will detect the rvalue reference and automatically make a copy even if
		you just use `ndarray` (see next section), but to ensure that it works
		also with older compilers it is recommended to always use
		`ndarray_copy` when returning newly constructed eigen values.


		### Corrupt data when returning non-map types

		The tendency of Eigency to avoid copies whenever possible can lead
		to corrupted data when returning non-map Eigen arrays. For instance,
		in the `function_w_mat_retval` from the previous section, a temporary
		value will be returned from C++, and we have to take care to make
		a copy of this data instead of letting the resulting numpy array
		refer directly to this memory. In C++11, this situation can be
		detected directly using rvalue references, and it will therefore
		automatically make a copy:
		```
		def function_w_mat_retval():
		# This works in C++11, because it detects the rvalue reference
		return ndarray(_function_w_mat_retval())
		```

		However, to make sure it works with older compilers,
		it is recommended to use the `ndarray_copy` conversion:

		```
		def function_w_mat_retval():
		# Explicit request for copy - this always works
		return ndarray_copy(_function_w_mat_retval())
		```

		### Storage layout - why arrays are sometimes transposed

		The default storage layout used in numpy and Eigen differ. Numpy uses
		a row-major layout (C-style) per default while Eigen uses a
		column-major layout (Fortran style) by default. In Eigency, we prioritize to
		avoid copying of data whenever possible, which can have unexpected
		consequences in some cases: There is no problem when passing values
		from C++ to Python - we just adjust the storage layout of the returned
		numpy array to match that of Eigen. However, since the storage layout
		is encoded into the _type_ of the Eigen array (or the type of the
		Map), we cannot automatically change the layout in the Python to C++ direction. In
		Eigency, we have therefore opted to return the transposed array/matrix
		in such cases. This provides the user with the flexibility to deal
		with the problem either in Python (use order="F" when constructing
		your numpy array), or on the C++ side: (1) explicitly define your
		argument to have the row-major storage layout, 2) manually set the Map
		stride, or 3) just call `.transpose()` on the received
		array/matrix).

		As an example, consider the case of a C++ function that both receives
		and returns a Eigen Map type, thus acting as a filter:

		```c++
		Eigen::Map<Eigen::ArrayXXd> function_filter(Eigen::Map<Eigen::ArrayXXd> &mat) {
		return mat;
		}
		```

		The Cython code could be:

		```
		cdef extern from "functions.h":
		...
		cdef Map[ArrayXXd] &_function_filter1 "function_filter1" (Map[ArrayXXd] &)

		def function_filter1(np.ndarray[np.float64_t, ndim=2] array):
		return ndarray(_function_filter1(Map[ArrayXXd](array)))

		```

		If we call this function from Python in the standard way, we will
		see that the array is transposed on the way from Python to C++, and
		remains that way when it is again returned to Python:

		```python
		>>> x = np.array([[1., 2., 3., 4.], [5., 6., 7., 8.]])
		>>> y = function_filter1(x)
		>>> print x
		[[ 1. 2. 3. 4.]
		[ 5. 6. 7. 8.]]
		>>> print y
		[[ 1. 5.]
		[ 2. 6.]
		[ 3. 7.]
		[ 4. 8.]]
		```

		The simplest way to avoid this is to tell numpy to use a
		column-major array layout instead of the default row-major
		layout. This can be done using the order='F' option:

		```python
		>>> x = np.array([[1., 2., 3., 4.], [5., 6., 7., 8.]], order='F')
		>>> y = function_filter1(x)
		>>> print x
		[[ 1. 2. 3. 4.]
		[ 5. 6. 7. 8.]]
		>>> print y
		[[ 1. 2. 3. 4.]
		[ 5. 6. 7. 8.]]
		```

		The other alternative is to tell Eigen to use RowMajor layout. This
		requires changing the C++ function definition:

		```c++
		typedef Eigen::Map<Eigen::Array<double, Eigen::Dynamic, Eigen::Dynamic, Eigen::RowMajor> > RowMajorArrayMap;

		RowMajorArrayMap &function_filter2(RowMajorArrayMap &mat) {
		return mat;
		}
		```

		To write the corresponding Cython definition, we need the expanded version of
		`FlattenedMap` called `FlattenedMapWithOrder`, which allows us to specify
		the storage order:

		```
		cdef extern from "functions.h":
		...
		cdef PlainObjectBase _function_filter2 "function_filter2" (FlattenedMapWithOrder[Array, double, Dynamic, Dynamic, RowMajor])

		def function_filter2(np.ndarray[np.float64_t, ndim=2] array):
		return ndarray(_function_filter2(FlattenedMapWithOrder[Array, double, Dynamic, Dynamic, RowMajor](array)))
		```

		Another alternative is to keep the array itself in RowMajor format,
		but use different stride values for the Map type:

		```c++
		typedef Eigen::Map<Eigen::ArrayXXd, Eigen::Unaligned, Eigen::Stride<1, Eigen::Dynamic> > CustomStrideMap;

		CustomStrideMap &function_filter3(CustomStrideMap &);
		```

		In this case, in Cython, we need to use the even more extended
		`FlattenedMap` type called `FlattenedMapWithStride`, taking eight
		arguments:

		```
		cdef extern from "functions.h":
		...
		cdef PlainObjectBase _function_filter3 "function_filter3" (FlattenedMapWithStride[Array, double, Dynamic, Dynamic, ColMajor, Unaligned, _1, Dynamic])

		def function_filter3(np.ndarray[np.float64_t, ndim=2] array):
		return ndarray(_function_filter3(FlattenedMapWithStride[Array, double, Dynamic, Dynamic, ColMajor, Unaligned, _1, Dynamic](array)))
		```

		In all three cases, the returned array will now be of the same shape
		as the original.

		### Long double support

		Eigency provides new shorthands for Eigen `long double` and complex `long double` Matrix and Array types.
		Examples:
		```
		Vector4ld
		Matrix3ld
		Vector2cld
		Matrix4cld
		Array3Xld
		ArrayXXcld
		```

		These typedefs are available in the `eigency` namespace when including the `eigency` header:
		```c++
		#include "eigency.h"

		void receive_long_double_matrix(Eigen::Map<eigency::MatrixXld> &mat) {
		// use long double eigen matrix
		}
		```

		Use Cython (.pyx) to create Python binding to your C++ function:
		```python
		cdef extern from "functions.h":
		cdef void _receive_long_double_matrix "receive_long_double_matrix"(Map[MatrixXld] &)

		def send_long_double_ndarray(np.ndarray[np.longdouble_t, ndim=2] array):
		return _receive_long_double_matrix(Map[MatrixXld](array))
		```

		Invoke in Python:
		```python
		import numpy as np
		import my_module

		x = np.array([[1.1, 2.2], [3.3, 4.4]], dtype=np.longdouble)
		my_module.send_long_double_ndarray(x)
		```

+8

setup.cfg

		[sdist]
		owner = root
		group = root

		[egg_info]
		tag_build =
		tag_date = 0

+95

setup.py

		import os
		import sys
		from os.path import basename, join

		import numpy as np
		from setuptools import find_namespace_packages, setup
		from setuptools.extension import Extension

		sys.path.append(".")

		__package_name__ = "eigency"
		include_dirs = [np.get_include()]
		if "EIGEN_INC" in os.environ:
		useSystemEigen = True
		include_dirs.append(os.environ["EIGEN_INC"])
		else:
		useSystemEigen = False
		import eigency # noqa: E402

		__eigen_dir__ = eigency.get_eigency_eigen_dir()
		__eigen_lib_dir__ = join(basename(__eigen_dir__), "Eigen")
		include_dirs.append(__eigen_dir__)
		# Not all users may have cython installed. If they only want this as a means
		# to access the Eigen header files to compile their own C++ code, then they
		# may not have cython already installed. Therefore, only require cython
		# for cases where the user will need to build the .cpp files from the .pyx
		# files (typically from a git clone) and not for other pip installations.
		# cf. discussion in PR #26.

		# Follow the pattern recommended here:
		# http://cython.readthedocs.io/en/latest/src/reference/compilation.html#distributing-cython-modules
		try:
		from Cython.Build import cythonize

		# Maybe make this a command line option?
		USE_CYTHON = True
		ext = ".pyx"
		except ImportError:
		USE_CYTHON = False
		ext = ".cpp"

		extensions = [
		Extension(
		"eigency.conversions",
		["eigency/conversions" + ext],
		include_dirs=include_dirs,
		language="c++",
		define_macros=[("NPY_NO_DEPRECATED_API", "NPY_1_7_API_VERSION")],
		),
		Extension(
		"eigency.core",
		["eigency/core" + ext],
		include_dirs=include_dirs,
		language="c++",
		define_macros=[("NPY_NO_DEPRECATED_API", "NPY_1_7_API_VERSION")],
		),
		]

		if USE_CYTHON:
		extensions = cythonize(
		extensions,
		compiler_directives=dict(
		language_level="3",
		),
		)

		long_description = open("README.md").read()

		eigen_data_files = []

		exclude_package_data = {}
		if not useSystemEigen:
		for root, dirs, files in os.walk(join(__eigen_dir__, "Eigen")):
		for f in files:
		if f.endswith(".h"):
		eigen_data_files.append(join(root, f))
		eigen_data_files.append(join(__eigen_lib_dir__, "*"))
		exclude_package_data = {__package_name__: [join(__eigen_lib_dir__, "CMakeLists.txt")]}

		setup(
		name=__package_name__,
		use_scm_version=True,
		ext_modules=extensions,
		packages=find_namespace_packages(
		include=[
		"eigency",
		"eigency.eigen",
		"eigency.eigen.Eigen",
		"eigency.eigen.Eigen.*",
		],
		),
		include_package_data=True,
		package_data={__package_name__: [".h", ".pxd", "*.pyx"] + eigen_data_files},
		exclude_package_data=exclude_package_data,
		)

eigency/conversions.cpython-310-darwin.so

Sorry, the diff of this file is too big to display

eigency/core.cpython-310-darwin.so

Sorry, the diff of this file is not supported yet

-20

licenses/LICENSE

		Copyright (c) 2016 Wouter Boomsma

		Permission is hereby granted, free of charge, to any person obtaining
		a copy of this software and associated documentation files (the
		"Software"), to deal in the Software without restriction, including
		without limitation the rights to use, copy, modify, merge, publish,
		distribute, sublicense, and/or sell copies of the Software, and to
		permit persons to whom the Software is furnished to do so, subject to
		the following conditions:

		The above copyright notice and this permission notice shall be
		included in all copies or substantial portions of the Software.

		THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
		EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
		MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
		NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
		LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
		OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
		WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

-566

METADATA

		Metadata-Version: 2.4
		Name: eigency
		Version: 3.4.0.5
		Summary: Cython interface between the numpy arrays and the Matrix/Array classes of the Eigen C++ library
		Author-email: Wouter Boomsma <wb@di.ku.dk>, Felipe Bordeu <felipebordeu@gmail.com>
		Maintainer-email: Felipe Bordeu <felipebordeu@gmail.com>, Wouter Boomsma <wb@di.ku.dk>
		License-Expression: LicenseRef-Wouter-Boomsma-1.0
		Project-URL: Homepage, https://github.com/eigency-org/eigency
		Classifier: Intended Audience :: Developers
		Classifier: Programming Language :: C++
		Classifier: Programming Language :: Cython
		Classifier: Programming Language :: Python :: 3.9
		Classifier: Programming Language :: Python :: 3.10
		Classifier: Programming Language :: Python :: 3.11
		Classifier: Programming Language :: Python :: 3.12
		Classifier: Programming Language :: Python :: 3.13
		Requires-Python: >=3.9
		Description-Content-Type: text/markdown
		License-File: LICENSE
		Requires-Dist: numpy>=1.23.5
		Dynamic: license-file

		# Eigency

		[![PyPI version](https://badge.fury.io/py/eigency.svg)](https://badge.fury.io/py/eigency)
		[![PEP 517](https://github.com/eigency-org/eigency/actions/workflows/build.yml/badge.svg)](https://github.com/eigency-org/eigency/actions/workflows/build.yml)
		[![pip wheel](https://github.com/eigency-org/eigency/actions/workflows/wheel.yml/badge.svg)](https://github.com/eigency-org/eigency/actions/workflows/wheel.yml)
		[![setup.py](https://github.com/eigency-org/eigency/actions/workflows/setup.yml/badge.svg)](https://github.com/eigency-org/eigency/actions/workflows/setup.yml)
		[![pre-commit](https://github.com/eigency-org/eigency/actions/workflows/pre-commit.yml/badge.svg)](https://github.com/eigency-org/eigency/actions/workflows/pre-commit.yml)

		Eigency is a Cython interface between Numpy arrays and Matrix/Array
		objects from the Eigen C++ library. It is intended to simplify the
		process of writing C++ extensions using the Eigen library. Eigency is
		designed to reuse the underlying storage of the arrays when passing
		data back and forth, and will thus avoid making unnecessary copies
		whenever possible. Only in cases where copies are explicitly requested
		by your C++ code will they be made.

		## Versioning

		Eigency uses a 4-number version (`N.N.N.N`) where the first 3 parts correspond to the embedded Eigen library version.
		The last part is a revision number of Eigency itself.

		## Installing

		Eigency is packaged as a source distribution (`sdist`) and available on PyPi.
		It can be easily installed using `pip`:
		```bash
		python -m pip install eigency
		```

		Requirement: `pip >= 18.0`

		If your `pip` is too old, then upgrade it using:
		```bash
		python -m pip install --upgrade pip
		```

		## Contributing

		For instructions on building and/or packaging Eigency from source,
		see the contributing guide [here](https://github.com/eigency-org/eigency/blob/master/CONTRIBUTING.md).

		## Usage

		Below is a description of a range of common usage scenarios. A full working
		example of both setup and these different use cases is available in the
		`test` directory distributed with the this package.

		### Setup

		To import eigency functionality, add the following to your `.pyx` file:
		```
		from eigency.core cimport *
		```

		In addition, in the `setup.py` file, the include directories must be
		set up to include the eigency includes. This can be done by calling
		the `get_includes` function in the `eigency` module:
		```
		import eigency
		...
		extensions = [
		Extension("module-dir-name/module-name", ["module-dir-name/module-name.pyx"],
		include_dirs = [".", "module-dir-name"] + eigency.get_includes()
		),
		]
		```

		Eigency includes a version of the Eigen library, and the `get_includes` function will include the path to this directory. If you
		have your own version of Eigen, just set the `include_eigen` option to False, and add your own path instead:
		```
		include_dirs = [".", "module-dir-name", 'path-to-own-eigen'] + eigency.get_includes(include_eigen=False)
		```

		### From Numpy to Eigen

		Assume we are writing a Cython interface to the following C++ function:

		```c++
		void function_w_mat_arg(const Eigen::Map<Eigen::MatrixXd> &mat) {
		std::cout << mat << "\n";
		}
		```

		Note that we use `Eigen::Map` to ensure that we can reuse the storage
		of the numpy array, thus avoiding making a copy. Assuming the C++ code
		is in a file called `functions.h`, the corresponding `.pyx` entry could look like this:

		```
		cdef extern from "functions.h":
		cdef void _function_w_mat_arg "function_w_mat_arg"(Map[MatrixXd] &)

		# This will be exposed to Python
		def function_w_mat_arg(np.ndarray array):
		return _function_w_mat_arg(Map[MatrixXd](array))
		```

		The last line contains the actual conversion. `Map` is an Eigency
		type that derives from the real Eigen map, and will take care of
		the conversion from the numpy array to the corresponding Eigen type.

		We can now call the C++ function directly from Python:
		```python
		>>> import numpy as np
		>>> import eigency_tests
		>>> x = np.array([[1.1, 2.2], [3.3, 4.4]])
		>>> eigency_tests.function_w_mat_arg(x)
		1.1 3.3
		2.2 4.4
		```
		(if you are wondering about why the matrix is transposed, please
		see the Storage layout section below).

		## Types matter

		The basic idea behind eigency is to share the underlying representation of a
		numpy array between Python and C++. This means that somewhere in the process,
		we need to make explicit which numerical types we are dealing with. In the
		function above, we specify that we expect an Eigen MatrixXd, which means
		that the numpy array must also contain double (i.e. float64) values. If we instead provide
		a numpy array of ints, we will get strange results.

		```python
		>>> import numpy as np
		>>> import eigency_tests
		>>> x = np.array([[1, 2], [3, 4]])
		>>> eigency_tests.function_w_mat_arg(x)
		4.94066e-324 1.4822e-323
		9.88131e-324 1.97626e-323
		```
		This is because we are explicitly asking C++ to interpret out python integer
		values as floats.

		To avoid this type of error, you can force your cython function to
		accept only numpy arrays of a specific type:

		```
		cdef extern from "functions.h":
		cdef void _function_w_mat_arg "function_w_mat_arg"(Map[MatrixXd] &)

		# This will be exposed to Python
		def function_w_mat_arg(np.ndarray[np.float64_t, ndim=2] array):
		return _function_w_mat_arg(Map[MatrixXd](array))
		```

		(Note that when using this technique to select the type, you also need to specify
		the dimensions of the array (this will default to 1)). Using this new definition,
		users will get an error when passing arrays of the wrong type:

		```python
		>>> import numpy as np
		>>> import eigency_tests
		>>> x = np.array([[1, 2], [3, 4]])
		>>> eigency_tests.function_w_mat_arg(x)
		Traceback (most recent call last):
		File "<stdin>", line 1, in <module>
		File "eigency_tests/eigency_tests.pyx", line 87, in eigency_tests.eigency_tests.function_w_mat_arg
		ValueError: Buffer dtype mismatch, expected 'float64_t' but got 'long'
		```

		Since it avoids many surprises, it is strongly recommended to use this technique
		to specify the full types of numpy arrays in your cython code whenever
		possible.

		### Writing Eigen Map types in Cython

		Since Cython does not support nested fused types, you cannot write types like `Map[Matrix[double, 2, 2]]`. In most cases, you won't need to, since you can just use Eigens convenience typedefs, such as `Map[VectorXd]`. If you need the additional flexibility of the full specification, you can use the `FlattenedMap` type, where all type arguments can be specified at top level, for instance `FlattenedMap[Matrix, double, _2, _3]` or `FlattenedMap[Matrix, double, _2, Dynamic]`. Note that dimensions must be prefixed with an underscore.

		Using full specifications of the Eigen types, the previous example would look like this:
		```
		cdef extern from "functions.h":
		cdef void _function_w_mat_arg "function_w_mat_arg" (FlattenedMap[Matrix, double, Dynamic, Dynamic] &)

		# This will be exposed to Python
		def function_w_mat_arg(np.ndarray[np.float64_t, ndim=2] array):
		return _function_w_mat_arg(FlattenedMap[Matrix, double, Dynamic, Dynamic](array))
		```

		`FlattenedType` takes four template parameters: arraytype, scalartype,
		rows and cols. Eigen supports a few other template arguments for
		setting the storage layout and Map strides. Since cython does not
		support default template arguments for fused types, we have instead
		defined separate types for this purpose. These are called
		`FlattenedMapWithOrder` and `FlattenedMapWithStride` with five and eight
		template arguments, respectively. For details on their use, see the section
		about storage layout below.

		### From Numpy to Eigen (insisting on a copy)

		Eigency will not complain if the C++ function you interface with does
		not take a Eigen Map object, but instead a regular Eigen Matrix or
		Array. However, in such cases, a copy will be made. Actually, the
		procedure is exactly the same as above. In the `.pyx` file, you still
		define everything exactly the same way as for the Map case described above.

		For instance, given the following C++ function:
		```c++
		void function_w_vec_arg_no_map(const Eigen::VectorXd &vec);
		```

		The Cython definitions would still look like this:

		```
		cdef extern from "functions.h":
		cdef void _function_w_vec_arg_no_map "function_w_vec_arg_no_map"(Map[VectorXd] &)

		# This will be exposed to Python
		def function_w_vec_arg_no_map(np.ndarray[np.float64_t] array):
		return _function_w_vec_arg_no_map(Map[VectorXd](array))
		```

		Cython will not mind the fact that the argument type in the extern
		declaration (a Map type) differs from the actual one in the `.h` file,
		as long as one can be assigned to the other. Since Map objects can be
		assigned to their corresponding Matrix/Array types this works
		seemlessly. But keep in mind that this assignment will make a copy of
		the underlying data.

		### Eigen to Numpy

		C++ functions returning a reference to an Eigen Matrix/Array can also
		be transferred to numpy arrays without copying their content. Assume
		we have a class with a single getter function that returns an Eigen
		matrix member:

		```c++
		class MyClass {
		public:
		MyClass():
		matrix(Eigen::Matrix3d::Constant(3.)) {
		}
		Eigen::MatrixXd &get_matrix() {
		return this->matrix;
		}
		private:
		Eigen::Matrix3d matrix;
		};
		```

		The Cython C++ class interface is specified as usual:

		```
		cdef cppclass _MyClass "MyClass":
		_MyClass "MyClass"() except +
		Matrix3d &get_matrix()
		```

		And the corresponding Python wrapper:

		```python
		cdef class MyClass:
		cdef _MyClass *thisptr;

		def __cinit__(self):
		self.thisptr = new _MyClass()

		def __dealloc__(self):
		del self.thisptr

		def get_matrix(self):
		return ndarray(self.thisptr.get_matrix())
		```

		This last line contains the actual conversion. Again, eigency has its
		own version of `ndarray`, that will take care of the conversion for
		you.

		Due to limitations in Cython, Eigency cannot deal with full
		Matrix/Array template specifications as return types
		(e.g. `Matrix[double, 4, 2]`). However, as a workaround, you can use
		`PlainObjectBase` as a return type in such cases (or in all cases if
		you prefer):

		```
		PlainObjectBase &get_matrix()
		```

		### Overriding default behavior

		The `ndarray` conversion type specifier will attempt do guess whether you want a copy
		or a view, depending on the return type. Most of the time, this is
		probably what you want. However, there might be cases where you want
		to override this behavior. For instance, functions returning const
		references will result in a copy of the array, since the const-ness
		cannot be enforced in Python. However, you can always override the
		default behavior by using the `ndarray_copy` or `ndarray_view`
		functions.

		Expanding the `MyClass` example from before:

		```c++
		class MyClass {
		public:
		...
		const Eigen::MatrixXd &get_const_matrix() {
		return this->matrix;
		}
		...
		};
		```

		With the corresponding cython interface specification
		The Cython C++ class interface is specified as usual:

		```
		cdef cppclass _MyClass "MyClass":
		...
		const Matrix3d &get_const_matrix()
		```

		The following would return a copy

		```python
		cdef class MyClass:
		...
		def get_const_matrix(self):
		return ndarray(self.thisptr.get_const_matrix())
		```

		while the following would force it to return a view

		```python
		cdef class MyClass:
		...
		def get_const_matrix(self):
		return ndarray_view(self.thisptr.get_const_matrix())
		```

		### Eigen to Numpy (non-reference return values)

		Functions returning an Eigen object (not a reference), are specified
		in a similar way. For instance, given the following C++ function:

		```c++
		Eigen::Matrix3d function_w_mat_retval();
		```

		The Cython code could be written as:

		```
		cdef extern from "functions.h":
		cdef Matrix3d _function_w_mat_retval "function_w_mat_retval" ()

		# This will be exposed to Python
		def function_w_mat_retval():
		return ndarray_copy(_function_w_mat_retval())
		```

		As mentioned above, you can replace `Matrix3d` (or any other Eigen return type) with
		`PlainObjectBase`, which is especially relevant when working with
		Eigen object that do not have an associated convenience typedef.

		Note that we use `ndarray_copy` instead of `ndarray` to explicitly
		state that a copy should be made. In c++11 compliant compilers, it
		will detect the rvalue reference and automatically make a copy even if
		you just use `ndarray` (see next section), but to ensure that it works
		also with older compilers it is recommended to always use
		`ndarray_copy` when returning newly constructed eigen values.


		### Corrupt data when returning non-map types

		The tendency of Eigency to avoid copies whenever possible can lead
		to corrupted data when returning non-map Eigen arrays. For instance,
		in the `function_w_mat_retval` from the previous section, a temporary
		value will be returned from C++, and we have to take care to make
		a copy of this data instead of letting the resulting numpy array
		refer directly to this memory. In C++11, this situation can be
		detected directly using rvalue references, and it will therefore
		automatically make a copy:
		```
		def function_w_mat_retval():
		# This works in C++11, because it detects the rvalue reference
		return ndarray(_function_w_mat_retval())
		```

		However, to make sure it works with older compilers,
		it is recommended to use the `ndarray_copy` conversion:

		```
		def function_w_mat_retval():
		# Explicit request for copy - this always works
		return ndarray_copy(_function_w_mat_retval())
		```

		### Storage layout - why arrays are sometimes transposed

		The default storage layout used in numpy and Eigen differ. Numpy uses
		a row-major layout (C-style) per default while Eigen uses a
		column-major layout (Fortran style) by default. In Eigency, we prioritize to
		avoid copying of data whenever possible, which can have unexpected
		consequences in some cases: There is no problem when passing values
		from C++ to Python - we just adjust the storage layout of the returned
		numpy array to match that of Eigen. However, since the storage layout
		is encoded into the _type_ of the Eigen array (or the type of the
		Map), we cannot automatically change the layout in the Python to C++ direction. In
		Eigency, we have therefore opted to return the transposed array/matrix
		in such cases. This provides the user with the flexibility to deal
		with the problem either in Python (use order="F" when constructing
		your numpy array), or on the C++ side: (1) explicitly define your
		argument to have the row-major storage layout, 2) manually set the Map
		stride, or 3) just call `.transpose()` on the received
		array/matrix).

		As an example, consider the case of a C++ function that both receives
		and returns a Eigen Map type, thus acting as a filter:

		```c++
		Eigen::Map<Eigen::ArrayXXd> function_filter(Eigen::Map<Eigen::ArrayXXd> &mat) {
		return mat;
		}
		```

		The Cython code could be:

		```
		cdef extern from "functions.h":
		...
		cdef Map[ArrayXXd] &_function_filter1 "function_filter1" (Map[ArrayXXd] &)

		def function_filter1(np.ndarray[np.float64_t, ndim=2] array):
		return ndarray(_function_filter1(Map[ArrayXXd](array)))

		```

		If we call this function from Python in the standard way, we will
		see that the array is transposed on the way from Python to C++, and
		remains that way when it is again returned to Python:

		```python
		>>> x = np.array([[1., 2., 3., 4.], [5., 6., 7., 8.]])
		>>> y = function_filter1(x)
		>>> print x
		[[ 1. 2. 3. 4.]
		[ 5. 6. 7. 8.]]
		>>> print y
		[[ 1. 5.]
		[ 2. 6.]
		[ 3. 7.]
		[ 4. 8.]]
		```

		The simplest way to avoid this is to tell numpy to use a
		column-major array layout instead of the default row-major
		layout. This can be done using the order='F' option:

		```python
		>>> x = np.array([[1., 2., 3., 4.], [5., 6., 7., 8.]], order='F')
		>>> y = function_filter1(x)
		>>> print x
		[[ 1. 2. 3. 4.]
		[ 5. 6. 7. 8.]]
		>>> print y
		[[ 1. 2. 3. 4.]
		[ 5. 6. 7. 8.]]
		```

		The other alternative is to tell Eigen to use RowMajor layout. This
		requires changing the C++ function definition:

		```c++
		typedef Eigen::Map<Eigen::Array<double, Eigen::Dynamic, Eigen::Dynamic, Eigen::RowMajor> > RowMajorArrayMap;

		RowMajorArrayMap &function_filter2(RowMajorArrayMap &mat) {
		return mat;
		}
		```

		To write the corresponding Cython definition, we need the expanded version of
		`FlattenedMap` called `FlattenedMapWithOrder`, which allows us to specify
		the storage order:

		```
		cdef extern from "functions.h":
		...
		cdef PlainObjectBase _function_filter2 "function_filter2" (FlattenedMapWithOrder[Array, double, Dynamic, Dynamic, RowMajor])

		def function_filter2(np.ndarray[np.float64_t, ndim=2] array):
		return ndarray(_function_filter2(FlattenedMapWithOrder[Array, double, Dynamic, Dynamic, RowMajor](array)))
		```

		Another alternative is to keep the array itself in RowMajor format,
		but use different stride values for the Map type:

		```c++
		typedef Eigen::Map<Eigen::ArrayXXd, Eigen::Unaligned, Eigen::Stride<1, Eigen::Dynamic> > CustomStrideMap;

		CustomStrideMap &function_filter3(CustomStrideMap &);
		```

		In this case, in Cython, we need to use the even more extended
		`FlattenedMap` type called `FlattenedMapWithStride`, taking eight
		arguments:

		```
		cdef extern from "functions.h":
		...
		cdef PlainObjectBase _function_filter3 "function_filter3" (FlattenedMapWithStride[Array, double, Dynamic, Dynamic, ColMajor, Unaligned, _1, Dynamic])

		def function_filter3(np.ndarray[np.float64_t, ndim=2] array):
		return ndarray(_function_filter3(FlattenedMapWithStride[Array, double, Dynamic, Dynamic, ColMajor, Unaligned, _1, Dynamic](array)))
		```

		In all three cases, the returned array will now be of the same shape
		as the original.

		### Long double support

		Eigency provides new shorthands for Eigen `long double` and complex `long double` Matrix and Array types.
		Examples:
		```
		Vector4ld
		Matrix3ld
		Vector2cld
		Matrix4cld
		Array3Xld
		ArrayXXcld
		```

		These typedefs are available in the `eigency` namespace when including the `eigency` header:
		```c++
		#include "eigency.h"

		void receive_long_double_matrix(Eigen::Map<eigency::MatrixXld> &mat) {
		// use long double eigen matrix
		}
		```

		Use Cython (.pyx) to create Python binding to your C++ function:
		```python
		cdef extern from "functions.h":
		cdef void _receive_long_double_matrix "receive_long_double_matrix"(Map[MatrixXld] &)

		def send_long_double_ndarray(np.ndarray[np.longdouble_t, ndim=2] array):
		return _receive_long_double_matrix(Map[MatrixXld](array))
		```

		Invoke in Python:
		```python
		import numpy as np
		import my_module

		x = np.array([[1.1, 2.2], [3.3, 4.4]], dtype=np.longdouble)
		my_module.send_long_double_ndarray(x)
		```

-361

RECORD

		eigency-3.4.0.5.dist-info/RECORD,,
		eigency-3.4.0.5.dist-info/WHEEL,sha256=cMZt8yy9snWrcJ6ukBKliUyeAPJLnzk9nJajsmR7UPI,136
		eigency-3.4.0.5.dist-info/top_level.txt,sha256=DBJvIgHQMhMUeEDbqHHPmrsyr-0YM1UyIC4Ho9YT264,8
		eigency-3.4.0.5.dist-info/METADATA,sha256=NoGUGhdGaC8fgABnfhfmpKqAjxqAIh5ixu7O8UC1QRY,19602
		eigency-3.4.0.5.dist-info/licenses/LICENSE,sha256=ON97Ls1S-bG2sTZRKI4qJzP_hh4_1ZnFh9CDmFfb2g0,1058
		eigency/conversions.pxd,sha256=DS6p4JGfBsoWdTSRLk224zSv0zU_8hXIC8LKHEBK8Gc,1170
		eigency/core.pxd,sha256=sPaXoQ8cg7_i3JIq3xPamdTp6vTtUOkWeUJ3XWYNfVo,22405
		eigency/conversions.cpp,sha256=b-EjAddHSyCGUusLOc2uMH43tMnz-WnWL8No8GvP0nU,1818851
		eigency/core.cpp,sha256=mllqks7P6Db-7-w83cFfk5GIS8m3lFOU9LoaYjs5XVU,320226
		eigency/eigency_cpp.h,sha256=7NU7GNe7nDifjZ7OuN_caDu2SB4prLKnLPnDEEDolIc,19834
		eigency/__init__.py,sha256=Cg7sAXYZ9kV0eEYhUdPWx4Gd366BMFWttkKrYx-1PX4,412
		eigency/eigency.h,sha256=NLVNfM_nwXj2aXTrrRGAN9rqLcAXJJLzX8wvbEaKw2A,2257
		eigency/conversions.cpython-310-darwin.so,sha256=zWd3iAU2JFZzHYt4oMOJvf51ZOjJPp7j2Yfn6xoZ2Ws,398928
		eigency/conversions_api.h,sha256=RHGj9aEnbuYw22_X0i2pOG9eylRktjmzXV4sSQS0ioo,37356
		eigency/core.pyx,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
		eigency/conversions.pyx,sha256=mTXZSS-pjD4NF4keuNghgsWWFJkeZ3ePCgxjEREH4D4,1662
		eigency/core.cpython-310-darwin.so,sha256=gRcV6QE87i2Cm_6ZZctI_Y9B6hhE4Qdu5H8Sz-MIa2Y,55976
		eigency/eigen/COPYING.LGPL,sha256=3GJlINzVOiL3J68-5Cx3DlbJemT-OtsGN5nYqwMv5VE,26530
		eigency/eigen/COPYING.BSD,sha256=UZKNzjYhPFMzujFy6EfXNdTG6bf_JyKjJsSQZxVbgus,1517
		eigency/eigen/COPYING.APACHE,sha256=AzeQAaexKi7Jl6JVVCR9mFJws1PBDVuv7prI1lGYILc,11362
		eigency/eigen/COPYING.MPL2,sha256=-rPda9qyJvHAhjCx3ZF-Efy07F4eAg4sFvg6ChOGPoU,16726
		eigency/eigen/COPYING.GPL,sha256=jOtLnuWt7d5Hsx6XXB2QxzrSe2sWWh3NgMfFRetluQM,35147
		eigency/eigen/COPYING.README,sha256=yDIwt3DxfvE4bqH9NoEnHdmKqTZGvb-1v_OhtwUP_50,779
		eigency/eigen/COPYING.MINPACK,sha256=yHt_juiPYZXpF0OCDAA1SDNYOu8JG3Li1KScjijnmKA,2193
		eigency/eigen/Eigen/CholmodSupport,sha256=JvDRvQZ3eFOU8fTrLN33Y0I_cDwJQbPlg3PpVbATAyc,1900
		eigency/eigen/Eigen/StdVector,sha256=fosQYHV6EUApj_OQXfcO21mtEIP1BWgd8w3J3Jlj2Ks,803
		eigency/eigen/Eigen/StdDeque,sha256=wYhRZ6RHglq3w1VzQiw--iPexi9bYwrDIv_mWODuUkI,797
		eigency/eigen/Eigen/Core,sha256=uK5fd6MNSLyvYipW9wR7qosvt5D123Nhqh1X9CT6PB4,12799
		eigency/eigen/Eigen/SparseLU,sha256=XnRux2MFSTHxbt5cWIqrK0TXVf69rymfXuTMoDRqhbY,1814
		eigency/eigen/Eigen/OrderingMethods,sha256=JfrIHKwzWAF-Tcsg-Tr9D5zmXWBBN28v9dlv2dCEvXI,2451
		eigency/eigen/Eigen/QtAlignedMalloc,sha256=xGYpHG59Iw6Ldrclza0e2Cwkm_AfXsO8tvP2UB7Q_5c,900
		eigency/eigen/Eigen/Householder,sha256=GhjqJlnbFc6T3g9a_xNaU2r5cB2pr08CZxDlbe8jpVk,829
		eigency/eigen/Eigen/PardisoSupport,sha256=UOxQA985n6v1GGIOIg9xxxf_OgOnIYuGUKl03DrQaxU,1116
		eigency/eigen/Eigen/SparseCore,sha256=IyPB4xed_Gq_NIFht4wb9ezQyqj0UscMrKmhzf3Wpq4,2240
		eigency/eigen/Eigen/Jacobi,sha256=rzAgyRuOktBzTBE_sVEdlqDWsFgSiRHYmOAXdiijWMg,894
		eigency/eigen/Eigen/Dense,sha256=UyXJ1NaT5QTuASxSda1LELE1pwHL2HwURm0uoh6iLuU,122
		eigency/eigen/Eigen/SPQRSupport,sha256=T6LXO5mMB5RoLWPf-__3lb6kGn0YZlhAscNX5m_xyJ0,1162
		eigency/eigen/Eigen/QR,sha256=53rhAQ6d8dyw5ah1YEjDiAzrAjcR6NG1aeAbA58TCCA,1272
		eigency/eigen/Eigen/Eigen,sha256=sEc7XXwM2TEdSpBUgWLozxYJBEdocff7dJ6WJOmz39E,35
		eigency/eigen/Eigen/SparseQR,sha256=dDNFiuAILyaqw7VE8I5XGKXRIsR5qTFCUrQRzPHZuPs,1195
		eigency/eigen/Eigen/SVD,sha256=EVWjLv1mTLQzM6NPqKy8vCb6M62u3-6OIO3fZ0Nq6IY,1584
		eigency/eigen/Eigen/Cholesky,sha256=SMcsLXuY_bRRZ10u4DhQ7P3JfqQbzlXtdUsD2kqwm9A,1161
		eigency/eigen/Eigen/UmfPackSupport,sha256=AD1B-gjN5oscRphJn6m_nIA867J650hhWFFRn0LcVqw,1382
		eigency/eigen/Eigen/IterativeLinearSolvers,sha256=hP-jisHvO-AFv_UAt4tSnB4WT8UGN4Nv-GSP6hJ7dPc,2083
		eigency/eigen/Eigen/LU,sha256=0A-rF2kizdSiMf8z6tWPwtshqdkwNLz3ugwqNRplS6Q,1268
		eigency/eigen/Eigen/Sparse,sha256=vjUbbuZSp6IOOVPQ6uIW1NAcLCa6e2pGS_kI2zRJkHs,888
		eigency/eigen/Eigen/Geometry,sha256=84oUXuUT0fp1v8_HlPvEfJ3mWjr12ftxngV6xjMCwvg,1940
		eigency/eigen/Eigen/SuperLUSupport,sha256=iqYB8_oea_wDt5lma_OevELUSRel-yqqSJvq_WOQH3I,2243
		eigency/eigen/Eigen/StdList,sha256=A1QOKXrpbtA-go2aQzu39aXpOqg8b1CmwnHHj5KD1mc,726
		eigency/eigen/Eigen/MetisSupport,sha256=5G1R0wZZYkCKQDSA8snK1OfWH7UJweSmlSdHUSdJaaI,991
		eigency/eigen/Eigen/SparseCholesky,sha256=bjZKchFd6lcGHhNqbsgEzyHF0815qkQ5_Ss1L1id6ps,1235
		eigency/eigen/Eigen/Eigenvalues,sha256=8ZWKms880-H1GNhx19xgLoEBszzjUAi47o4XIIRb2gc,1777
		eigency/eigen/Eigen/PaStiXSupport,sha256=9CO5Vxa4BPlSEjTfvFitGi-tmhb4S88W3wu4iu7MhUM,1751
		eigency/eigen/Eigen/KLUSupport,sha256=vEYx66RCVNLRAih_AcArDKeTKRyXa1CbfDUoFX_tL2Q,1389
		eigency/eigen/Eigen/src/CholmodSupport/CholmodSupport.h,sha256=_XiNvLNTzR8Oyp-m_2V2CzNY-7RF7n3v7x-nyYNKXnk,25441
		eigency/eigen/Eigen/src/misc/blas.h,sha256=dOWodSZy7a3ZNjf23H_rilBCfqlPmsXR7VbcLa4PrWc,30560
		eigency/eigen/Eigen/src/misc/lapacke.h,sha256=Bmc-lq9luQfIPUJJW3VDYbQr2_nvxjU9UfLnAjlsrrc,1058369
		eigency/eigen/Eigen/src/misc/Kernel.h,sha256=IIiRDtPS_STfPS980xZLyEV8eCqOAFi39K3yKKHT7ro,2742
		eigency/eigen/Eigen/src/misc/RealSvd2x2.h,sha256=X-B5X0XNkqCbSMAiWe90O6y76G-JDLDAclwMA_tWiGo,1748
		eigency/eigen/Eigen/src/misc/Image.h,sha256=-Pn7ZkwUYgVhdflW2Td77dFQ6MRfuJo570RjtjaGxKE,2913
		eigency/eigen/Eigen/src/misc/lapack.h,sha256=_9_gqkbLMiEgKp8lguOez21a1mYmzf1w9AmFlSHWg0Y,7834
		eigency/eigen/Eigen/src/misc/lapacke_mangling.h,sha256=fXH97lOh4M0BVoq6LxUvLh5Pu4SehNdA47Ax6Bn7-iI,474
		eigency/eigen/Eigen/src/Core/Reverse.h,sha256=lmCGSuTPyF3sQNgueaWYtx5cWPUwNL3TDHQJPFq5oVQ,7522
		eigency/eigen/Eigen/src/Core/ConditionEstimator.h,sha256=uSdt-5gLRfIu3k4I7uRH9eAIR0AOZvJnlnaSIz5CTCc,6990
		eigency/eigen/Eigen/src/Core/PermutationMatrix.h,sha256=xOo3Hcj9crxhDZe56OJB2XZcCNjWqT5m2FH8CVzZzmA,20748
		eigency/eigen/Eigen/src/Core/Assign.h,sha256=dKQAmY90XhiXo_Kd4MHxDmrS6XJp9ylFq7JZ5mF_O2I,2738
		eigency/eigen/Eigen/src/Core/MathFunctionsImpl.h,sha256=lvc1eFdCOZ_8YQmm2KQb8jknibWnE8nL-y8Il2dJEkc,7156
		eigency/eigen/Eigen/src/Core/CwiseUnaryOp.h,sha256=ptT-VJWaOq0EShXxJmiLRFujF8jNj2DDpOGGMuMBXJA,3937
		eigency/eigen/Eigen/src/Core/DiagonalMatrix.h,sha256=Kg1DVgs8zq-jetJsx7RKlRVF5WqH0X54uJbG0GPGRxg,14670
		eigency/eigen/Eigen/src/Core/StlIterators.h,sha256=yudREy61voHxl0k2-akEGTWqipHX0AZbSDAeTPV7EMQ,21641
		eigency/eigen/Eigen/src/Core/VectorBlock.h,sha256=1_VF-6FHnjOeMEB5nsGgAIskLLr7RWEAy5XnzWRYa0k,3488
		eigency/eigen/Eigen/src/Core/ReturnByValue.h,sha256=ijhUmPC6DGdGb-FRr0LlU3vTxddbidQLc7wQPHcwlmU,4284
		eigency/eigen/Eigen/src/Core/Solve.h,sha256=0UEeTDezRD00It_7piJacK4Yaw7ZCj8Ju-wUcn7fhvI,6837
		eigency/eigen/Eigen/src/Core/ArithmeticSequence.h,sha256=8vgcI6huL7rFhH2DSPfdEPD-OYEoNNNLVERuG8pE-68,19214
		eigency/eigen/Eigen/src/Core/Product.h,sha256=R5WGCVcHEtjAZsjnUp_OM0bwzHlKzNOQ91PZcD1IQs8,7336
		eigency/eigen/Eigen/src/Core/MatrixBase.h,sha256=P8p3jdYQkj98CJ_m9j5gAAXxT-3BVoUVkfypmaKOWSs,23856
		eigency/eigen/Eigen/src/Core/Stride.h,sha256=XvhjHxIi1JnpwnMmbsu4B3iRJosg3RHHv1foOz-3w30,4212
		eigency/eigen/Eigen/src/Core/Map.h,sha256=N9PDYRqEYInLMKDXjd3EXTIphbtj-k31aA9vM8AWoHs,7256
		eigency/eigen/Eigen/src/Core/ForceAlignedAccess.h,sha256=g5uk6NSZjvNzhM24_mbkq6A3rvqCTaxjsQLwFrNnAXU,4909
		eigency/eigen/Eigen/src/Core/MathFunctions.h,sha256=s8bNFz1KyHnlaKxXCAQt0LsWCAXGupxV2gXIE0ZW91g,60784
		eigency/eigen/Eigen/src/Core/SelfCwiseBinaryOp.h,sha256=TQ8CFAN4yP0ouwXS8T3FDdhnvbwt3c4iP-N7gtvxerU,1697
		eigency/eigen/Eigen/src/Core/BandMatrix.h,sha256=ZusMpQuT9NzcCDqiH0Md2QUsMZkRNW2QJVdvDNnGyJM,14075
		eigency/eigen/Eigen/src/Core/Inverse.h,sha256=Xu8vJuEFNJiB6xQIzC62dLJkySTUW6P2v867yt_msQw,3503
		eigency/eigen/Eigen/src/Core/Assign_MKL.h,sha256=dPYwhbhTcHKTPsFzuRtDcL7RUXeMZDWicqlBtnj4tJA,12488
		eigency/eigen/Eigen/src/Core/Swap.h,sha256=cq3ST1UzR38ao94FITvncu7qmhPWfbgtAvdPzOUB6kA,2765
		eigency/eigen/Eigen/src/Core/ProductEvaluators.h,sha256=2IY694zzCciY008Oizi7TBs3C1YipIBkIR9bOwrZHrA,53832
		eigency/eigen/Eigen/src/Core/MapBase.h,sha256=7uz3jWkieES7KRTck21U22rNn6jvjQ4pqScrhJON4aA,11281
		eigency/eigen/Eigen/src/Core/Matrix.h,sha256=_YTfwccLBClS0pnznwVpT4LKHxfKAT6zNnV6PPcebRE,24343
		eigency/eigen/Eigen/src/Core/EigenBase.h,sha256=9nQxmKrOAik2Su84QtekyuxMZIJoES1dyIdQ0TnwJho,5841
		eigency/eigen/Eigen/src/Core/StableNorm.h,sha256=SI5NEBo9ABOxkbgMGPwcQumIczthCbcEv6QMz3fT0Yc,8700
		eigency/eigen/Eigen/src/Core/GenericPacketMath.h,sha256=FatABDPfUC3DlhmTRVtWPFXV-kfFqTIE0SrSp5cwr90,38812
		eigency/eigen/Eigen/src/Core/CwiseBinaryOp.h,sha256=AxbMB-jrRUh65PJDJ5EWjhwTQHEgAU67P0FhP_xuMCw,7909
		eigency/eigen/Eigen/src/Core/Visitor.h,sha256=0y8bT7AGWjEr8TooYc88PS_wjSvvStiwFLUAYa6IAqQ,11997
		eigency/eigen/Eigen/src/Core/DenseStorage.h,sha256=I16S84EpH_dd3j9b2Djc9-TGyAuoOUBVDib37kt8DIg,25360
		eigency/eigen/Eigen/src/Core/DiagonalProduct.h,sha256=_6cfNHpZFMnpNB6oK7Qr97b3uNBFS8F7xWKa3bghOMs,988
		eigency/eigen/Eigen/src/Core/CwiseUnaryView.h,sha256=5ZqmH2GmdTLUproXOERwPH6Y4o17jH6c7gqp0w1rlnU,5551
		eigency/eigen/Eigen/src/Core/NoAlias.h,sha256=Ip1s5MuKGSHKzWNPb-ZOaYxUB3MCSHxiICuEWnRGYY0,3620
		eigency/eigen/Eigen/src/Core/DenseCoeffsBase.h,sha256=shKz8IklvEP3gDcJl28Iz0qD2t1UXVWLqRWZ6SoAllk,24484
		eigency/eigen/Eigen/src/Core/PartialReduxEvaluator.h,sha256=hj7Bxbfe3LjOvaTLnDi2_HfcddIpDb_yw5p9uYzjOjg,9207
		eigency/eigen/Eigen/src/Core/IndexedView.h,sha256=X6kQiPumLzTPCrTmy7q6iKNrfEsZXebS9nyW4pPzL-w,9620
		eigency/eigen/Eigen/src/Core/Fuzzy.h,sha256=TaSjzj16hbMNNZd2WEFeSKTBBR7Wuh3Uxhy8jTUAJ9M,5759
		eigency/eigen/Eigen/src/Core/TriangularMatrix.h,sha256=zEwgkLv57QKwo2EpgJ-yw5oy0r0u_qBdDeux6v6EgRE,38277
		eigency/eigen/Eigen/src/Core/IO.h,sha256=iaTrSS9KSgqeIe_2ZmODFS0PRe-C1KGHnmEnfa-0rU8,8238
		eigency/eigen/Eigen/src/Core/Array.h,sha256=39fKPydmTk45GiUd5NpStFdQf8lCUWohqz6Ln6T-jaQ,16782
		eigency/eigen/Eigen/src/Core/SolverBase.h,sha256=M57lChZDARbiviOylEHsf7kLc_8r9r6KOeHHT7Phca4,6170
		eigency/eigen/Eigen/src/Core/AssignEvaluator.h,sha256=E5UBHDUEfSI5Oeas89CW6DrsdhKG6aEJZ0pbdeBtZaM,41673
		eigency/eigen/Eigen/src/Core/Replicate.h,sha256=b3eh2lzUvOvhy7kNUv1_wj0ZpG6LG48qgI7Q7L-b2-4,5656
		eigency/eigen/Eigen/src/Core/CoreIterators.h,sha256=MLxTh9kMqEHm7IVobSElMYoZ1P2qBah_rm_6sISPdJo,4745
		eigency/eigen/Eigen/src/Core/ArrayBase.h,sha256=h6XAellrUiXUmw2iWF2B0Ub38jt6DaeiDFHrW3Rq_F4,8217
		eigency/eigen/Eigen/src/Core/Block.h,sha256=EJKGf-xoAEOIsLo403rWA8x8tdA03Ci2_jqC6SrYiK0,18648
		eigency/eigen/Eigen/src/Core/Dot.h,sha256=SVnXKm92-UFx5WJF5zMtVVp3UdKdy-BLOq6iyEGODvU,11654
		eigency/eigen/Eigen/src/Core/PlainObjectBase.h,sha256=-qVMt5_XTbX9X7PG9g_CUjmOvDUp6WB8_eIjIkQWIzU,49193
		eigency/eigen/Eigen/src/Core/SelfAdjointView.h,sha256=awX8Mcrm-Ubbf89EOe-hPrIc2bGQOueRDMdKYRuV9o8,14999
		eigency/eigen/Eigen/src/Core/GeneralProduct.h,sha256=u3pQ7qYb-HzruMX_JVPjZB4ed9YvVhg44SZb9CrIg6A,21679
		eigency/eigen/Eigen/src/Core/SolveTriangular.h,sha256=8psntHTv5Wkru_Hxhj8N20zPA54-2gJ5b1TDOInchqE,9368
		eigency/eigen/Eigen/src/Core/CoreEvaluators.h,sha256=6J0oxusVbLZZ9YqGGX_eJArEyGEaPdKRHkGipw0CejM,63841
		eigency/eigen/Eigen/src/Core/CwiseNullaryOp.h,sha256=cPW5ApD9Ddx10LNh324u_0-7vZl_W4kyFArFgYAgJN8,36282
		eigency/eigen/Eigen/src/Core/VectorwiseOp.h,sha256=taEMnk7xoASOi9F-UUNUseFkex1wDVkHbB-jTs8EvD0,35168
		eigency/eigen/Eigen/src/Core/ArrayWrapper.h,sha256=i9CnjmTAjDy95K81Nv_m4GKhLOnHMhxDoUaXBKk3zOQ,7018
		eigency/eigen/Eigen/src/Core/BooleanRedux.h,sha256=Udvh6C02Obp9MLeU4KmYLPizv5iZVWiJcNS81yPHCqU,4429
		eigency/eigen/Eigen/src/Core/Transpositions.h,sha256=OQ83uchr7R8XCMX3eGtGCWZYIDcKQ3oiLaMgowx7mGc,13567
		eigency/eigen/Eigen/src/Core/NestByValue.h,sha256=XpO9qFQqjr-zIdyRmbEQpTri-48sIOELbyTkN5sKJ8s,2520
		eigency/eigen/Eigen/src/Core/Ref.h,sha256=9KCGVMNnveUblYknmVbCOht0eniluU60x-Q_Fq4mLrw,17821
		eigency/eigen/Eigen/src/Core/Reshaped.h,sha256=UQGxKCGDdGABPnCtstVC9eYjzdZUHTYEr2KNKnjefhQ,17033
		eigency/eigen/Eigen/src/Core/Transpose.h,sha256=qhBLL99zcYMdT87iOD3spTAMSYBbmaPIg-E_QLEiV1A,17606
		eigency/eigen/Eigen/src/Core/Redux.h,sha256=3Ntognx_kGOvdFyaDh6Fz1PApptn09SIATSi747R8_Y,19195
		eigency/eigen/Eigen/src/Core/NumTraits.h,sha256=YDlXi0wX8kTrbgsY8puAictK_71EMTZeT-iySCWRmfE,12884
		eigency/eigen/Eigen/src/Core/CommaInitializer.h,sha256=qAJ2QnYid9sBFKj070CQMDVIQSHVtO35pQ1cEOEaGyA,5981
		eigency/eigen/Eigen/src/Core/GlobalFunctions.h,sha256=NTFhnsI7k43_IwOaPw3Ykr5TP2PWgcfWWWZZ27CuMaI,11543
		eigency/eigen/Eigen/src/Core/Random.h,sha256=_JWxS-8mkmR8CJGyWQbKyYjpUZ5fcZ2w_Z5k4PrwpsI,7756
		eigency/eigen/Eigen/src/Core/Diagonal.h,sha256=DohU_IOblBq_OwllK1_sT6WnzTpErPy1J0p3c7abL_8,9870
		eigency/eigen/Eigen/src/Core/DenseBase.h,sha256=ho6oVZqhHldM5zShTqVvXOUDJ0ay1z9MjHkw-JxZgic,31529
		eigency/eigen/Eigen/src/Core/CwiseTernaryOp.h,sha256=l9ngPHPhoQE1LYsA7hkQE7a2rPAOkMiH3VS-pFUd574,8256
		eigency/eigen/Eigen/src/Core/Select.h,sha256=C5vueNWTUBIRCHeLutIDpGPYHHG_tc1zK4WcAmBaHrU,6143
		eigency/eigen/Eigen/src/Core/products/SelfadjointRank2Update.h,sha256=GjBksbLTUh8dSwIroBAzFAZd3pNmaGzgwq9y_AezhAY,4126
		eigency/eigen/Eigen/src/Core/products/GeneralBlockPanelKernel.h,sha256=sYJHOdyCm91ch9TiA1eUIw9eqOMAeXfjeDmeGVYrmPM,108448
		eigency/eigen/Eigen/src/Core/products/GeneralMatrixMatrixTriangular_BLAS.h,sha256=lz3gcCibomI0C11GP9WWJfLxJWkfozIuaeKPWNp8KoQ,6936
		eigency/eigen/Eigen/src/Core/products/GeneralMatrixMatrix_BLAS.h,sha256=ec1yDwUGvK3GAOot40BaRFdTbko1vrQHgWH3X3YTT2c,5106
		eigency/eigen/Eigen/src/Core/products/GeneralMatrixVector_BLAS.h,sha256=mZtYGatUC0-bugfLV3Y81DPJy8H-pm4F2hKqQQZOdpM,6368
		eigency/eigen/Eigen/src/Core/products/GeneralMatrixMatrix.h,sha256=33qhvQHkXEfnHqOM5XnYH1t5qlXe1A8haFgClGVk7CA,20104
		eigency/eigen/Eigen/src/Core/products/Parallelizer.h,sha256=LFS04pfAyxh-5HrsMZ7q00Om_P2z9BTUJ1i_1o_0k40,5582
		eigency/eigen/Eigen/src/Core/products/GeneralMatrixMatrixTriangular.h,sha256=tBWrOcXAPqby8_dJumI0JLnx4PfqHn33HhYPLuOReW4,15948
		eigency/eigen/Eigen/src/Core/products/TriangularMatrixVector.h,sha256=pFx7Lk-jLFI3L51jXGWZoYRPkFPt2RCHdOZJcrC9yP4,14722
		eigency/eigen/Eigen/src/Core/products/SelfadjointMatrixVector.h,sha256=X9VMtikMlDeHUmXltnN5f0zbbkXHkf4CR7QeCY4zEFU,9958
		eigency/eigen/Eigen/src/Core/products/TriangularSolverMatrix.h,sha256=e755tKu-x4ef6xyEwr-SuOUDm6p6e8_u6d7LOr9URVA,14678
		eigency/eigen/Eigen/src/Core/products/GeneralMatrixVector.h,sha256=6ghVCEVie_PXwK0w7XBh1euVkLH7VI_T2P3pBnB7EfM,21724
		eigency/eigen/Eigen/src/Core/products/SelfadjointProduct.h,sha256=DkewFi5s_R2NwxTXy8EFUpk2sDqAL1RsEibRG2o42-Y,6164
		eigency/eigen/Eigen/src/Core/products/TriangularMatrixVector_BLAS.h,sha256=SsT_NUCdiHXlevXOyArnxTe2QDTicg1AH766Tscmh0k,10571
		eigency/eigen/Eigen/src/Core/products/TriangularMatrixMatrix_BLAS.h,sha256=sJu-wvyIjUQml9JtYlcfGpkbShLzR3SJMPNZLVN8Odo,13867
		eigency/eigen/Eigen/src/Core/products/SelfadjointMatrixVector_BLAS.h,sha256=EKrXRBTWeuks1FjFJn-ExEHdLD3oJs1PkzOqBCzHaxw,5209
		eigency/eigen/Eigen/src/Core/products/TriangularSolverMatrix_BLAS.h,sha256=2cKW56cfUCooBLv4pwS09ehmtZeNb5QWhcNUi6XmqEI,6707
		eigency/eigen/Eigen/src/Core/products/SelfadjointMatrixMatrix_BLAS.h,sha256=dO0QzTSlEY0DALfJHNCLv-VqyWWdGKOD640qrzoHDOw,11570
		eigency/eigen/Eigen/src/Core/products/TriangularSolverVector.h,sha256=X-vXZ8N5pXDF-RiweJD3XIe_muTWsPVlRW8nWzCNEL4,5882
		eigency/eigen/Eigen/src/Core/products/TriangularMatrixMatrix.h,sha256=inJs7op1I5ker51aTO8raRwxg5TWT0svVeE1iKWZ7RY,20987
		eigency/eigen/Eigen/src/Core/products/SelfadjointMatrixMatrix.h,sha256=4gQBLDrdPARSB7Xk5-7qX4O_PanQxvfih6zqHFiksi0,21354
		eigency/eigen/Eigen/src/Core/util/NonMPL2.h,sha256=RekKBVaidiOpEb9sm8gcs9_LyeyC3OKOSvf4JV6l2tc,85
		eigency/eigen/Eigen/src/Core/util/ReshapedHelper.h,sha256=IiSNYQUioI1RXHJZb45ppb0PaTBUWphGxXXaVRGbNRI,1432
		eigency/eigen/Eigen/src/Core/util/BlasUtil.h,sha256=R4hliGvYhqOw9uIh_5uBG-N0InTR-iYMf71SWivFVxY,23156
		eigency/eigen/Eigen/src/Core/util/StaticAssert.h,sha256=FuzHTQMY4HyEd5nmdXqLTd13adbwcHehgKfe8-620b0,10676
		eigency/eigen/Eigen/src/Core/util/Constants.h,sha256=ttU2oem4zBxoX3xs95Tr5wpnIJk3D-V-R19soXOSNqA,21931
		eigency/eigen/Eigen/src/Core/util/XprHelper.h,sha256=91FpxKLKPlGo5qxCP_NK2Ce_cyZMy736OxGtGfjvEN4,35762
		eigency/eigen/Eigen/src/Core/util/MKL_support.h,sha256=SH8MIg1AuEGIdrXH2QMtNZc3TN_3kNQCLo9O6pMs9iA,4268
		eigency/eigen/Eigen/src/Core/util/Macros.h,sha256=jXMlm0ukgubb0Rsx2Ih_01DfsINnANMnk0Rrumqxrno,52909
		eigency/eigen/Eigen/src/Core/util/DisableStupidWarnings.h,sha256=Ytqy175AkKqgXj0sZYGySbc2Vya7dQbNyRBVL8g4bFU,4892
		eigency/eigen/Eigen/src/Core/util/SymbolicIndex.h,sha256=2x8GtyOpQZncKMYU5NkdJatRYS08G9cFYnN2_x0Mnv4,12003
		eigency/eigen/Eigen/src/Core/util/ForwardDeclarations.h,sha256=ZhdE6dxa1vvWMloGjxiNVqfUxTouV3vfAtIDLoP5NUE,15555
		eigency/eigen/Eigen/src/Core/util/Memory.h,sha256=4lVlqmrj5hLCa5LP5zGl35DX7fpme-W5h8KAHHCyyVw,46661
		eigency/eigen/Eigen/src/Core/util/Meta.h,sha256=TuaFXI83ZpuMhkgzQsVvmtBvLfxpq_5GkSmKjSqUQgI,29336
		eigency/eigen/Eigen/src/Core/util/ConfigureVectorization.h,sha256=Puw_386FDIesp7rEXOK_GJiASoReKalUjD6Y5Mgq01o,19876
		eigency/eigen/Eigen/src/Core/util/IntegralConstant.h,sha256=MsGgGP33vR824xB3MDJWbzUa_ztaANXyOeJNDCgJ7mk,10949
		eigency/eigen/Eigen/src/Core/util/ReenableStupidWarnings.h,sha256=4CANL3bylEgx4MEpc_qnL6uwRFu6_WXHrlFquM1nreU,1024
		eigency/eigen/Eigen/src/Core/util/IndexedViewHelper.h,sha256=6I-WuqgpdRGkOQAtI7CuMARenFQ4pLmzaA5BgtbittU,6696
		eigency/eigen/Eigen/src/Core/functors/UnaryFunctors.h,sha256=kFshHJT8zGBHriYmc8ZEq4KcgnXpi29LiXjHPLNJRio,40146
		eigency/eigen/Eigen/src/Core/functors/TernaryFunctors.h,sha256=D8PBqipa0vHKnNvPN6YzcSzARSsQftY_akzxDWcREWg,607
		eigency/eigen/Eigen/src/Core/functors/AssignmentFunctors.h,sha256=2PGxIkRB-s5e_kJx1zVGg-wWtcvo7InnxyjBP_hb8WA,6686
		eigency/eigen/Eigen/src/Core/functors/BinaryFunctors.h,sha256=CwfNbUDL-WVhgk64Cq5a9xnc_hxNNJsBEQCs21eCGZc,20921
		eigency/eigen/Eigen/src/Core/functors/StlFunctors.h,sha256=cYI-b8vomEe7fK7xnoN7Hu0cMkQs340QQQOIjE_-S7A,4998
		eigency/eigen/Eigen/src/Core/functors/NullaryFunctors.h,sha256=GQpjDoSCvAuUODxnubXXpB88JssQiU1UWXz949c7pws,8334
		eigency/eigen/Eigen/src/Core/arch/SSE/PacketMath.h,sha256=peCG6Wv3h2vZ2gxDEX9IJDXQTNy8TPnBBMFyRCmAQTA,64465
		eigency/eigen/Eigen/src/Core/arch/SSE/MathFunctions.h,sha256=cAcwShFbn0K0g-TiyoNqDsV41oqJo_kMBI3evwdGc-U,6765
		eigency/eigen/Eigen/src/Core/arch/SSE/TypeCasting.h,sha256=Uq2Z3RJeG1t7ZICopSGTROezDe9xwI_nTRQwSFaELmA,3650
		eigency/eigen/Eigen/src/Core/arch/SSE/Complex.h,sha256=h6bwXhEi-IZVCphcy3kVC-CZmTUtfLzoa77tZIohV6k,14251
		eigency/eigen/Eigen/src/Core/arch/GPU/PacketMath.h,sha256=2KKjUpwiDgzbxMXoRwB7Fg5Bs98NsGn9RWqgvd0LoD0,57047
		eigency/eigen/Eigen/src/Core/arch/GPU/MathFunctions.h,sha256=WNBXkzPWzZx-a9VYKr_u3A2mS80eARwiY8LdflRmT1c,2695
		eigency/eigen/Eigen/src/Core/arch/GPU/TypeCasting.h,sha256=HR1OOHLb9NmrIsW2dkRBgmWgpkHibWw0DUgJyqO8Lpg,2256
		eigency/eigen/Eigen/src/Core/arch/MSA/PacketMath.h,sha256=PaxsMC3KR2tfvd6eWHG3-vSEc_100mAsB-dIOiy4ZM4,33615
		eigency/eigen/Eigen/src/Core/arch/MSA/MathFunctions.h,sha256=8WDhQFA0s_u9LNZOh6J8DQ-vkAchObk3v263HzaYPyg,16159
		eigency/eigen/Eigen/src/Core/arch/MSA/Complex.h,sha256=mD175hijPLKuPWMgJG8Tn1c5uyUzdRLMJSzeY2SHkWQ,17541
		eigency/eigen/Eigen/src/Core/arch/CUDA/Complex.h,sha256=of74nv05x33Y09JHA06CnpMGnhKM--ya7wKOi5XHhu0,17317
		eigency/eigen/Eigen/src/Core/arch/ZVector/PacketMath.h,sha256=F01Hv3MIEyKand_VxS4kvVHMgyBC4YoaepnphPKiIXY,36894
		eigency/eigen/Eigen/src/Core/arch/ZVector/MathFunctions.h,sha256=AL5v8pbey2Oeuyd75kVVmD6O2q98waSnkKYnszN19Es,8024
		eigency/eigen/Eigen/src/Core/arch/ZVector/Complex.h,sha256=5kgD19Zhq9EwxI7PhkRBwLnA1cXD4uioUIo7h_DbLF8,16728
		eigency/eigen/Eigen/src/Core/arch/Default/Half.h,sha256=vN9CBcF-3crLwXXBpdGI2ys2yeUCmzgpB20F39pAb00,35534
		eigency/eigen/Eigen/src/Core/arch/Default/Settings.h,sha256=Qbh6le3EAOdUX3JzxZdCq8ld765sp82F-WLVGnqPt5A,1746
		eigency/eigen/Eigen/src/Core/arch/Default/TypeCasting.h,sha256=Tv47egCZ2kG0IEAACpxLkr1I77NblQXEU7Xz0wT7T3U,3746
		eigency/eigen/Eigen/src/Core/arch/Default/BFloat16.h,sha256=HD-vLoEqRj0L4cfaszaO3mr-hNNPBCwAe8Xa03Xpa9k,26903
		eigency/eigen/Eigen/src/Core/arch/Default/GenericPacketMathFunctionsFwd.h,sha256=a9Lk1ShhdtC4I3COGmmc9GlMt06_VPLdtngnDbXEZxc,3770
		eigency/eigen/Eigen/src/Core/arch/Default/GenericPacketMathFunctions.h,sha256=4-8jZykCpqD-85ybbFlSup7WnmoOsy8OzBMAT8ZgBS0,67696
		eigency/eigen/Eigen/src/Core/arch/Default/ConjHelper.h,sha256=-tVUil_T8-S8sW_ihUnuc8Ht5fnin2yQIr1NitPqpSA,5251
		eigency/eigen/Eigen/src/Core/arch/AVX512/PacketMath.h,sha256=OUFH_dw0sH2Keii8gKXt0fpe8_ZRAbU1bOFo0H9xwx8,87891
		eigency/eigen/Eigen/src/Core/arch/AVX512/MathFunctions.h,sha256=05V1fuacBbf_IQEpvX9UIHv4q2RJCaKC9RhMjDVuMkc,13344
		eigency/eigen/Eigen/src/Core/arch/AVX512/TypeCasting.h,sha256=dSDpOtpAFcs4vLQTWSNai66TbSU53_hs1K-naDF1Gjs,2134
		eigency/eigen/Eigen/src/Core/arch/AVX512/Complex.h,sha256=ugr3JzEmI5-3tHJRvigqfAue6meW37D3xBG-XzsJI8Q,17160
		eigency/eigen/Eigen/src/Core/arch/SVE/PacketMath.h,sha256=ExnP6wxeTf-muQEl0kGqp9aFrG6ZB1WERaEkRJxX6v8,21200
		eigency/eigen/Eigen/src/Core/arch/SVE/MathFunctions.h,sha256=ArIHSTscu776ZEoith1MvV16BxFAhvV3iLYdC4zYHjI,1194
		eigency/eigen/Eigen/src/Core/arch/SVE/TypeCasting.h,sha256=euzd4LBaPcZvuqE4w2WMiFyysQSYE_No9MdqGfO7vjQ,1351
		eigency/eigen/Eigen/src/Core/arch/HIP/hcc/math_constants.h,sha256=TYakFU7yBjjXKrL1hexh9sCIGQjZhqK0_7RO-pf5ckY,691
		eigency/eigen/Eigen/src/Core/arch/AVX/PacketMath.h,sha256=I0SozIfkw1iRU0p6x93nPgEpNeDNZWmbGO53u-aX_kA,64608
		eigency/eigen/Eigen/src/Core/arch/AVX/MathFunctions.h,sha256=c4oN-JVUz9ohpSw7pG5Hvf-eNRmceI3Y3U2bo8FZgQU,8102
		eigency/eigen/Eigen/src/Core/arch/AVX/TypeCasting.h,sha256=LKu4896ma5FgFwWLY-VC0hih7Y9Z7Dlv8hrjV0tZcmw,2564
		eigency/eigen/Eigen/src/Core/arch/AVX/Complex.h,sha256=TReBJLZKy19ZFalvKel_Lr49xEIICFGZvQksN8hBbxA,15223
		eigency/eigen/Eigen/src/Core/arch/SYCL/PacketMath.h,sha256=dZkkF8ku0zF__pcYkyFNCekEhg3xBi9NDPTmUDiwpLU,27786
		eigency/eigen/Eigen/src/Core/arch/SYCL/MathFunctions.h,sha256=Szum19_ng_oX-Tt_oykOHuEHsUkF88gV5OE36aMupD8,12539
		eigency/eigen/Eigen/src/Core/arch/SYCL/SyclMemoryModel.h,sha256=IIOs2E1g5_iKaeuF-aC8ygJw6tb8KnDLzklMhkwGQ1k,21856
		eigency/eigen/Eigen/src/Core/arch/SYCL/InteropHeaders.h,sha256=2G5Noo9_d_MvJobezN9exX1zevfW_lo6lf63Y1j1OTQ,7428
		eigency/eigen/Eigen/src/Core/arch/SYCL/TypeCasting.h,sha256=B_oAog8UHkDn2lOO-DhVRoAaAWT2Wt5sBHWAIKC9_7o,2626
		eigency/eigen/Eigen/src/Core/arch/NEON/GeneralBlockPanelKernel.h,sha256=VNMQHhSfKRhI0RcKJhalOI8skbh4p08tMWPagiB-p6c,6815
		eigency/eigen/Eigen/src/Core/arch/NEON/PacketMath.h,sha256=PhMYCg3QG5ppvRfxeqhpw4xFFd8cyDBDuI9D2aExt18,189525
		eigency/eigen/Eigen/src/Core/arch/NEON/MathFunctions.h,sha256=jNj21p4bjGDqqIwPlR6rYZuOzpAzk9f5DXJqpIfagW4,3083
		eigency/eigen/Eigen/src/Core/arch/NEON/TypeCasting.h,sha256=c2yYt3iTKrpZFLQY_YEKE1V6LreGtdEec2goqfPucIE,51286
		eigency/eigen/Eigen/src/Core/arch/NEON/Complex.h,sha256=yyJ41YOVRLrIqws49Ddbi63_scdGqNavmXZv52Sv7fY,22503
		eigency/eigen/Eigen/src/Core/arch/AltiVec/PacketMath.h,sha256=oPYdaGjd2Dt3f1b3YAbzU5W73cgIz7_aDhzQ5-IeSi8,102394
		eigency/eigen/Eigen/src/Core/arch/AltiVec/MathFunctions.h,sha256=GkyOC20PLKzbUW2SJhPzqSvHf41jfZjcomaAZs4N10s,2323
		eigency/eigen/Eigen/src/Core/arch/AltiVec/MatrixProduct.h,sha256=qH4IawNh5kCC5e2EeYvbh8l1BfYlySMO4rDAqWtOqsY,119355
		eigency/eigen/Eigen/src/Core/arch/AltiVec/MatrixProductMMA.h,sha256=MRJTWUfUBVYsDs189H1oIBeoYwbK90anZxJ1w_Zqk9E,24820
		eigency/eigen/Eigen/src/Core/arch/AltiVec/MatrixProductCommon.h,sha256=BOcTnv97OFhluX8S-H-c40gQEoupm9BqXPOq_PoJ-04,9490
		eigency/eigen/Eigen/src/Core/arch/AltiVec/Complex.h,sha256=CUSoZRzKfCXxeQfELlIb5y879_nskuYPPpAa34ADUeE,16540
		eigency/eigen/Eigen/src/SparseLU/SparseLU_Utils.h,sha256=VfbLFtU8hQaZl3sijSs0KKJ1Ok4eH0kZwfSlAD1CchA,2049
		eigency/eigen/Eigen/src/SparseLU/SparseLU_kernel_bmod.h,sha256=wsTY9YnpE064Zp-tp79vhsn31leV4rkHmVVkjQaVWCU,5723
		eigency/eigen/Eigen/src/SparseLU/SparseLU_column_bmod.h,sha256=UEFzzPnZoMvp_q7hyaNJaMQaivnuKGgfU8OMcLkY9Eo,6712
		eigency/eigen/Eigen/src/SparseLU/SparseLU_gemm_kernel.h,sha256=E0ypMty0CtJ9jyiQAlbsBrUdgU3t1IHhLXoq_29kP6I,10217
		eigency/eigen/Eigen/src/SparseLU/SparseLU.h,sha256=0jr8FVLqbXlYeSoQFJCA8z9Wp1gUZj-rM_8Zz7Dd4FM,33316
		eigency/eigen/Eigen/src/SparseLU/SparseLU_column_dfs.h,sha256=_1ukmKpCiPxqLHP_sW7wwJF0f58LNzJmb12dl1duO3c,6584
		eigency/eigen/Eigen/src/SparseLU/SparseLU_panel_dfs.h,sha256=HHxwbCV4bDeDVwo6Pl77jhTnCVAinCQ10m2ozi0_xDw,9028
		eigency/eigen/Eigen/src/SparseLU/SparseLU_relax_snode.h,sha256=xmWkJ3HrCeYx_o2b7V1NtTwoskgt84Fzmg5OHO4ZJNY,2889
		eigency/eigen/Eigen/src/SparseLU/SparseLU_heap_relax_snode.h,sha256=6H52qNW19OTAgaGACscVS30r9ZZjktHYzS4aEIxFPLk,4181
		eigency/eigen/Eigen/src/SparseLU/SparseLU_panel_bmod.h,sha256=E0ZMPmU59QNrxZLBFDQafZpbW30yduzwYIJux419A10,8485
		eigency/eigen/Eigen/src/SparseLU/SparseLU_Structs.h,sha256=iiBKapQJWGE1_1IYQg3G4Tifza4c_gcnUP1EeKg5RhQ,4974
		eigency/eigen/Eigen/src/SparseLU/SparseLU_pruneL.h,sha256=xbkTSkO6Y2EGGHWy8zHDVbEgOUJjm_bQLbHV7OOhWbw,4545
		eigency/eigen/Eigen/src/SparseLU/SparseLU_copy_to_ucol.h,sha256=xuejAH5T973lZoxXrfTfZMrp5d_GWWOEUzeSsqRaFjk,3681
		eigency/eigen/Eigen/src/SparseLU/SparseLU_Memory.h,sha256=0RtpqZwv2VeoYnL1E_2W55Zf-Q8GJBfCGWKnJVSMNKo,7602
		eigency/eigen/Eigen/src/SparseLU/SparseLU_pivotL.h,sha256=WtGmsoGD9gUhPImH2f4V3RfVeHbbyVghDsPtAhMwbFs,4979
		eigency/eigen/Eigen/src/SparseLU/SparseLUImpl.h,sha256=xqFVMyyht4lOAPP36lz-OLTKDLLV4_abAlxAE358DkA,4303
		eigency/eigen/Eigen/src/SparseLU/SparseLU_SupernodalMatrix.h,sha256=XEsGCbVbRuN7n3frLwrGupWG9vKhkYHBzFVmbY6Vflo,12837
		eigency/eigen/Eigen/src/OrderingMethods/Ordering.h,sha256=zQaPZlp5hpbFNKmxRQx42zI6Y3E9uTa7ddt5tP_1LJM,5248
		eigency/eigen/Eigen/src/OrderingMethods/Eigen_Colamd.h,sha256=aGVdr2uuSAuOnywk06Ox5iRwLxWsHoxIHuOpuiYJgOs,61681
		eigency/eigen/Eigen/src/OrderingMethods/Amd.h,sha256=c7g7u_79UNDMwQhqLLF-io6d843uZM7zEBkL-BrNMZw,16105
		eigency/eigen/Eigen/src/plugins/ArrayCwiseUnaryOps.h,sha256=Lw8DNEdrrQmtr1uj64wlKMK2qzrkchCDLL0nY4hRP0w,21431
		eigency/eigen/Eigen/src/plugins/CommonCwiseUnaryOps.h,sha256=KounaHKSxrMBzP7sUcUSdSELN0QhjCSXsYZbU3DYMlY,6089
		eigency/eigen/Eigen/src/plugins/BlockMethods.h,sha256=VP5GDS9pFosNtDF1bpa0pUZ9uqpCas3CG4elB6fCaw8,59020
		eigency/eigen/Eigen/src/plugins/ArrayCwiseBinaryOps.h,sha256=3J685GySQ-w9ATdG_yU5K2_enofrPBBjhYzBaHXBnUk,14060
		eigency/eigen/Eigen/src/plugins/MatrixCwiseBinaryOps.h,sha256=kOqEkZJMkAgDl_sePzPQ4ahC9aO4C5n8UqIGUZWMaYI,6387
		eigency/eigen/Eigen/src/plugins/IndexedViewMethods.h,sha256=XpWD8sYQXjvwE93Br-KgaFtKugYO9hEERuk3l1QEZUg,12283
		eigency/eigen/Eigen/src/plugins/CommonCwiseBinaryOps.h,sha256=H4qb7Ef81ff1KdWVbhXaLQ0yURh5vfYT05-yL7ybRJg,4828
		eigency/eigen/Eigen/src/plugins/ReshapedMethods.h,sha256=OYyO_yoZ3TL7X8GFrIW-IKzqs_1LbM0M96HJyZAVOi8,6915
		eigency/eigen/Eigen/src/plugins/MatrixCwiseUnaryOps.h,sha256=hBC2LK2aHti-DqelsfGCGw5zMMxqaFjIyd_DJ-Sd17M,3350
		eigency/eigen/Eigen/src/Householder/HouseholderSequence.h,sha256=qJx7vJqOfwuk2bOCwT_qSA4uQRA5uDK07lhMOmJNTOg,23611
		eigency/eigen/Eigen/src/Householder/Householder.h,sha256=Sm81woK9lvrX8YcXqLStZqdoHyWUxO9VlLOzE2iOiy0,5365
		eigency/eigen/Eigen/src/Householder/BlockHouseholder.h,sha256=gT-lKuZc5n9FylQ3dARF5S0a7M4J93mT7oRzI87Koa4,4784
		eigency/eigen/Eigen/src/PardisoSupport/PardisoSupport.h,sha256=wsVkKSSLM_PyT6okkE-68CY-i9uw5JQK-J_49b53K0w,20092
		eigency/eigen/Eigen/src/SparseCore/SparseSelfAdjointView.h,sha256=pKMa9HSZVf4TDrSs3WQPHm81-yvYWN_X3QbDGXZhChs,25889
		eigency/eigen/Eigen/src/SparseCore/SparseMatrix.h,sha256=Uahcnyjjk9enq1aUvlnqI1d5zBeHSv8rHDjVr8BKufs,57475
		eigency/eigen/Eigen/src/SparseCore/CompressedStorage.h,sha256=KbsmaOECLX-_dCrw4l5OH8qSYthcNl51lHEUrOAElfU,8743
		eigency/eigen/Eigen/src/SparseCore/SparseCwiseBinaryOp.h,sha256=UM8IsTqnqOIVvcVPNMI74lkGu0TC7c1-3OHD4E00U7U,25524
		eigency/eigen/Eigen/src/SparseCore/SparseView.h,sha256=tt7YUvbYq4FC7o8OjMRAE5ENMAyDgBjJQw23fkFxKYI,8127
		eigency/eigen/Eigen/src/SparseCore/SparseColEtree.h,sha256=hxvhX86IU5AEN0Re5JE7ugycPWrwjmMT8965y5Fuk0Q,6485
		eigency/eigen/Eigen/src/SparseCore/SparseMatrixBase.h,sha256=vKuTKTtOY6UAAogaE7yVSPElvanBCqJB4og5B7kI21g,17451
		eigency/eigen/Eigen/src/SparseCore/SparseDiagonalProduct.h,sha256=ql1gDrRmdDxurRXKYKt3558VV56rwkkg39q1rmNjUZ8,5808
		eigency/eigen/Eigen/src/SparseCore/SparseTriangularView.h,sha256=wLzim0CxFIHthPPnf-BETneIxDpr5os7WMmm22Ihps8,6437
		eigency/eigen/Eigen/src/SparseCore/SparseAssign.h,sha256=V9_BJwEgsy433vZPaZkNcz7VHaUtxFty8DrG2y7Bby0,11368
		eigency/eigen/Eigen/src/SparseCore/SparsePermutation.h,sha256=MOCjCgQY25pYtPMQhvGvkRk8koWuGO8dk0QRMqStiM8,7329
		eigency/eigen/Eigen/src/SparseCore/SparseCwiseUnaryOp.h,sha256=Gt77DWmbFD21-2tgm3UCHDFB3vFqg07ZzPOmK1sYdU8,4757
		eigency/eigen/Eigen/src/SparseCore/SparseMap.h,sha256=LENinGDgzP-0fSLxe9IC2QmTGpvw6-WmYZouy98GH6A,12589
		eigency/eigen/Eigen/src/SparseCore/SparseDenseProduct.h,sha256=S-SIciqeQ8Cq0eRnjdwO0W-nJTdWrJ9cDSyUpTKDnwQ,13256
		eigency/eigen/Eigen/src/SparseCore/SparseRef.h,sha256=KpWEyz-BSDQr2jkFY41qpl81k1xobd09MroJeAet90A,15600
		eigency/eigen/Eigen/src/SparseCore/SparseFuzzy.h,sha256=-HvMRkGwLuoeAVRUzaTxJOZerK1LKoF7zIhUntzpE3A,1107
		eigency/eigen/Eigen/src/SparseCore/MappedSparseMatrix.h,sha256=LqSrqm_EN6KXQXTFIExr2w5AzGtZ39M70CDZwt1mDto,2191
		eigency/eigen/Eigen/src/SparseCore/SparseSparseProductWithPruning.h,sha256=oJ95Mr8PoxBlVcuyIk83xAnA9B8BTF8SLFFr8JLzMcg,8704
		eigency/eigen/Eigen/src/SparseCore/SparseSolverBase.h,sha256=_FJeIwvbZvS6TXnQI4bqcW07AeultzRTOhcfBDogpHk,4424
		eigency/eigen/Eigen/src/SparseCore/SparseBlock.h,sha256=B_Rou_EU1DXsnaBNbKC5QPH9fWO2JUjH-dgTBVC0JFE,24360
		eigency/eigen/Eigen/src/SparseCore/TriangularSolver.h,sha256=6crxaL_epxj6kAKDRIOV3leRsjdoPfEMrwtFPPTtMU8,9657
		eigency/eigen/Eigen/src/SparseCore/SparseVector.h,sha256=FXR2BBlYjN74_Y7HGt_xK14qGY93lmZOLi7FEZNW4h8,14832
		eigency/eigen/Eigen/src/SparseCore/SparseCompressedBase.h,sha256=i_QDAUmYXh_o_or8Ooc6-OHwzl88PFIjK1gK1uTxENc,13606
		eigency/eigen/Eigen/src/SparseCore/SparseTranspose.h,sha256=vzGAkYMV3rfcQz-g1POI16Qgn6QweDzJ2ANzq4i_1y4,3175
		eigency/eigen/Eigen/src/SparseCore/SparseProduct.h,sha256=XL278sDrscm5H1ClbMyWtp9zMF-KoM-vecd-fXEmZZ0,7593
		eigency/eigen/Eigen/src/SparseCore/AmbiVector.h,sha256=5rbOU6LmuocubflDlpHJW-LV8RbLVt52PxlNwAD-24Q,10670
		eigency/eigen/Eigen/src/SparseCore/SparseRedux.h,sha256=8-aCkMrV-7ug7PgoT6I4pSByEI4zaMBjL4yQgWtB3JE,1699
		eigency/eigen/Eigen/src/SparseCore/ConservativeSparseSparseProduct.h,sha256=NSTC00y1pePE2y11NaWiKkV-cZ7iQDu4D0uQdlX1aI4,13166
		eigency/eigen/Eigen/src/SparseCore/SparseUtil.h,sha256=QbJdHLBXVilPHMYRlJUpu242IYXCyljuRWU6Gg0CHyE,6827
		eigency/eigen/Eigen/src/SparseCore/SparseDot.h,sha256=qNMFtSwdxuZrJALftuwc1Uln8dXPHn7O-B6HYnfXh2c,3080
		eigency/eigen/Eigen/src/Jacobi/Jacobi.h,sha256=eqr6CeolwC9s94yeCFlGF_MdSvztGqDr8OLnT6XDp08,16383
		eigency/eigen/Eigen/src/SPQRSupport/SuiteSparseQRSupport.h,sha256=aBRbczGluCFCW0z6zj-VPv3fPnseGS5Y_LWtCTPvugE,11826
		eigency/eigen/Eigen/src/QR/FullPivHouseholderQR.h,sha256=0mO6wR2Oz4LkO-63diixti5mRpEwlK1j6JQpXTjknh4,26768
		eigency/eigen/Eigen/src/QR/HouseholderQR.h,sha256=buDUUKKkR7GtuG59nGmnZOHPuPOfnWqZ-omhYq7e7t8,14641
		eigency/eigen/Eigen/src/QR/CompleteOrthogonalDecomposition.h,sha256=K80XQdqtt7GtaO6trSe9d-6SvwF_dmYkTK-aip00P9Y,23429
		eigency/eigen/Eigen/src/QR/ColPivHouseholderQR_LAPACKE.h,sha256=Pg4g_uASg01GUvKv85HZ10mcLIb2TsBAe8TxUIUzMbs,4662
		eigency/eigen/Eigen/src/QR/ColPivHouseholderQR.h,sha256=GRplSpNfH9u1hGj-UO_0ASrJsqUw1X873zHylf98Msg,25498
		eigency/eigen/Eigen/src/QR/HouseholderQR_LAPACKE.h,sha256=ioSGFFyQnCEIqvG71bRI-RwLslUwcNpeThs4Zht2qoc,2993
		eigency/eigen/Eigen/src/SparseQR/SparseQR.h,sha256=b7kIX0lfsWOEXscFjviaUOKMkieFnGnPdkHNOIC07ms,29167
		eigency/eigen/Eigen/src/SVD/JacobiSVD_LAPACKE.h,sha256=ka623NBfDraF1xca6Y-VNZlJPWUcPuw7DI35DlnSLpo,5099
		eigency/eigen/Eigen/src/SVD/SVDBase.h,sha256=7BZYx_gFqsiiYhobHmP6hKj-0hx_JOF27IYqlFViMAo,14743
		eigency/eigen/Eigen/src/SVD/JacobiSVD.h,sha256=hAnEE_IZPVHqlmnC8LfSgDGdqVJlo1wyP6vGX4pAcQg,32988
		eigency/eigen/Eigen/src/SVD/BDCSVD.h,sha256=_W5LHUpTg7pE48jCCqoZf6JbGTuUBQ76x4EEknP9Er8,54214
		eigency/eigen/Eigen/src/SVD/UpperBidiagonalization.h,sha256=6IpmJCGMm2kzzhJeLAPQNt-cSmdDm-UceOm5QNDSH18,15957
		eigency/eigen/Eigen/src/Cholesky/LLT_LAPACKE.h,sha256=Hvw3IINOyWug30oGIqogw6gK5I2Q1gz6rV4L-cHPoEs,3974
		eigency/eigen/Eigen/src/Cholesky/LDLT.h,sha256=zGNi3c3eoUA-qewcvqLKbq8cWHRjDxsTCE9Rx48znXg,24934
		eigency/eigen/Eigen/src/Cholesky/LLT.h,sha256=Ni_msIDJtn0REG-CH5JvqE7SSyEg-9uioJ77CGnu8ew,18760
		eigency/eigen/Eigen/src/UmfPackSupport/UmfPackSupport.h,sha256=to6gjstZ3u7FA3DBEekczSMwUP5inY9UoSRyeSP-6nY,24456
		eigency/eigen/Eigen/src/IterativeLinearSolvers/BasicPreconditioners.h,sha256=YMCQ6x4_-d79glE5ruPlzad6znhQFY02gqV0ymmGdsY,6771
		eigency/eigen/Eigen/src/IterativeLinearSolvers/IncompleteCholesky.h,sha256=l_qee65b_AyF54NQTEhqiQn4wpxF0EOgQfrjzKM-Z-o,15036
		eigency/eigen/Eigen/src/IterativeLinearSolvers/SolveWithGuess.h,sha256=skUBIA1GTm3pchMagwn-Q0vElsbBKyOXqVfZ6Ig9gp0,4212
		eigency/eigen/Eigen/src/IterativeLinearSolvers/IncompleteLUT.h,sha256=9N6ZL4g7dpQ1-C_AYBr2jV4R1nrToYgg1CJSstT8VIQ,14940
		eigency/eigen/Eigen/src/IterativeLinearSolvers/LeastSquareConjugateGradient.h,sha256=hxTZyH0vfNtGucFrc_JnlmaBnvnbYwpV6lnJVrQZDB8,7349
		eigency/eigen/Eigen/src/IterativeLinearSolvers/ConjugateGradient.h,sha256=bAe7AXzOohoeR4hASXMS6QKHt2GIAwvEXN1zwa32EdM,8887
		eigency/eigen/Eigen/src/IterativeLinearSolvers/BiCGSTAB.h,sha256=JAMwkJDDKrvMpaK4iCZQ3vhXaqbDleMsAafEft8Aay0,6850
		eigency/eigen/Eigen/src/IterativeLinearSolvers/IterativeSolverBase.h,sha256=RVUk3D9zTiShzR68o8N4qlsrbnu3xHAY-xCHP79Ixso,13379
		eigency/eigen/Eigen/src/LU/PartialPivLU_LAPACKE.h,sha256=405eiTTFwaWRyj6v0acEWJjEBKlbk-6zdEZqEnjERiU,3555
		eigency/eigen/Eigen/src/LU/InverseImpl.h,sha256=YcMuvxXWRaOKJJgf6YzAL6UanhW7sVAwgyyWOSahVeQ,15727
		eigency/eigen/Eigen/src/LU/Determinant.h,sha256=Ej8IFYHDjgeiBM8MsooiCpbZMFH8fADI0ZuaGG_C-yc,3439
		eigency/eigen/Eigen/src/LU/PartialPivLU.h,sha256=Lb6kmlOg7MA18WeAQAj0y0ZS5J-hiZ-kiIzX0fbApks,22069
		eigency/eigen/Eigen/src/LU/FullPivLU.h,sha256=cC7S7j5wQlYzGYKS6YnPMIVZ5k_aFyKE2vR1UZMSpHE,32383
		eigency/eigen/Eigen/src/LU/arch/InverseSize4.h,sha256=jEo1e85De1QKK5gpSB3wUciHnvmr4QgS4YThDNIgNQI,13693
		eigency/eigen/Eigen/src/Geometry/AngleAxis.h,sha256=hWEB6P7KkVOjUDyiPkdVZ7v1PofDUIQFQDVguHGydCU,8403
		eigency/eigen/Eigen/src/Geometry/RotationBase.h,sha256=EiowLy81k6w5UeRAjH0UBDYKAxuqA4Bhe6HdbdZFGMw,8063
		eigency/eigen/Eigen/src/Geometry/EulerAngles.h,sha256=MvmHhZbH0Nf9YzZG8B8YjXvHXQsWz3W8HY-bbCXoX6k,3624
		eigency/eigen/Eigen/src/Geometry/Hyperplane.h,sha256=L5qiKx7H8mhpvf3eXhPCpcudz_jw4XJ7nvaEty82qtc,11962
		eigency/eigen/Eigen/src/Geometry/Homogeneous.h,sha256=GubxVxEyxSFiZiGMTfln7dmXw3ayq55IrdJ5R-oxb98,20726
		eigency/eigen/Eigen/src/Geometry/AlignedBox.h,sha256=LJX63IVJ-iNtb4qjgCK635GG1jrtEqECu4Brzw9tUi4,18939
		eigency/eigen/Eigen/src/Geometry/Quaternion.h,sha256=jigf4p9JcYKOeQ8zRlP8qq_jSnhTsO7OQiSmOhzPH6g,34367
		eigency/eigen/Eigen/src/Geometry/Umeyama.h,sha256=8WBq8Bq-eOCwCPpCEJvCV4AdG9i--MYPuIQzekjltHA,6190
		eigency/eigen/Eigen/src/Geometry/Translation.h,sha256=MQelKqMEPMYDxBx_o_Bin1j4nlL5aQrunz3_TYrcRjk,7664
		eigency/eigen/Eigen/src/Geometry/Scaling.h,sha256=VP1L8yU9-kjXgRpJRDQ_Mobm4F6c47WBQu2qF75qbOw,6724
		eigency/eigen/Eigen/src/Geometry/OrthoMethods.h,sha256=PUA_fyzkfUoG16Tlvi82LCsO2Lx8LARN56CfsjGVbEQ,8955
		eigency/eigen/Eigen/src/Geometry/Rotation2D.h,sha256=ZJF39LieDyLVc82wCBjcv-89XUmjgiP8_FGtW4lhvOM,6862
		eigency/eigen/Eigen/src/Geometry/Transform.h,sha256=_LyXBb70PoLUGeas7TkZRADBH8gHKJp01ui7RmJmesM,61930
		eigency/eigen/Eigen/src/Geometry/ParametrizedLine.h,sha256=_XBm00cMu2GvSxc_Vq0Lw3CY6gPQPmLreuj3-ZhGkSE,9812
		eigency/eigen/Eigen/src/Geometry/arch/Geometry_SIMD.h,sha256=K0pUdNdQMSUnnx3iuujvVvmaxS3-zTIY__XXkCd3mno,5945
		eigency/eigen/Eigen/src/SuperLUSupport/SuperLUSupport.h,sha256=wmnrpnMFQ2KKgZJ2YZ0sGwtnORWfNB6zVx7yV_-mIzM,34324
		eigency/eigen/Eigen/src/MetisSupport/MetisSupport.h,sha256=x7X2CnuvlYp6C0a_SEUYueSdVryY74llxUM718ZPJk8,4588
		eigency/eigen/Eigen/src/StlSupport/StdList.h,sha256=_PoP9JjpMxFHyRt2ierD07kKYw5YCaQ-m2uff_9bv9A,4155
		eigency/eigen/Eigen/src/StlSupport/StdDeque.h,sha256=nwcLxBznnCK-XXIVzLj6m3TE9vNf6hCB5rEwVglEMtA,4730
		eigency/eigen/Eigen/src/StlSupport/details.h,sha256=N4XAZ1MVBkj5Wqqf0purzw8BN2oBMqkenPyCC7ZUF4E,2809
		eigency/eigen/Eigen/src/StlSupport/StdVector.h,sha256=ZMFTM2irx-ZrmNDMIYCyoRRWvfNdab18XXVN7WYgh4g,5338
		eigency/eigen/Eigen/src/SparseCholesky/SimplicialCholesky_impl.h,sha256=dFdpa-FOJlaOirwy9MZadsnMjesv7vwHpXWVky5bSPM,5830
		eigency/eigen/Eigen/src/SparseCholesky/SimplicialCholesky.h,sha256=oJz3HAOgW9nflw7znky97aP_SJz7S9vwhOhelWfXe_A,24216
		eigency/eigen/Eigen/src/Eigenvalues/GeneralizedEigenSolver.h,sha256=gjhXg_YIsizsBpnfbljjuOn7zmtAkz0levvRZRzXT4c,17176
		eigency/eigen/Eigen/src/Eigenvalues/EigenSolver.h,sha256=dk5LGEjuFvvUJFChALix6WXx-S9Buki-srtgELoLhEA,22970
		eigency/eigen/Eigen/src/Eigenvalues/RealSchur_LAPACKE.h,sha256=MKr7VFn4KX1oUapqi2USdVZdn6pFEzLQSvUHE8BDVis,3650
		eigency/eigen/Eigen/src/Eigenvalues/RealSchur.h,sha256=YIcKltQsAZXPrfkWnM44lWAviB4EDQ8x_1ZKlaPqZy8,21078
		eigency/eigen/Eigen/src/Eigenvalues/ComplexSchur_LAPACKE.h,sha256=_WVTQ369kknRbjGuKu7XPJg8r6PKuDXMmCk0PlzO-k8,4178
		eigency/eigen/Eigen/src/Eigenvalues/Tridiagonalization.h,sha256=mnlGNd83WlBplAPiJBD2ptx8z-yFtqasH7oEygEFEEc,22764
		eigency/eigen/Eigen/src/Eigenvalues/HessenbergDecomposition.h,sha256=O5VrdbpZMSheCsU_bUsr7VH_sJUo5Gh8Rg-D0nEt6Vs,14349
		eigency/eigen/Eigen/src/Eigenvalues/ComplexEigenSolver.h,sha256=slDfS5K-VOr0aXn4M4QT2IfQEcHoH-W3HcqYqFqKY1c,12559
		eigency/eigen/Eigen/src/Eigenvalues/GeneralizedSelfAdjointEigenSolver.h,sha256=4_rDPfwSRhuobLvKWLsS2dR1C_1O6ofWPX81JKXTsIQ,9716
		eigency/eigen/Eigen/src/Eigenvalues/SelfAdjointEigenSolver_LAPACKE.h,sha256=WVrgd6o4MYqGf6ZUP-qftealRVMvJ9Z2lYzB-ubvb6Q,4104
		eigency/eigen/Eigen/src/Eigenvalues/ComplexSchur.h,sha256=ZxIzjl0HXd09pS0vKLopsZLqR5IRwJcWGETc0Bgup5o,17274
		eigency/eigen/Eigen/src/Eigenvalues/SelfAdjointEigenSolver.h,sha256=C9q8ME35ACSMx57vfhud9XprJYazudk9QhNEuAbCbuY,35182
		eigency/eigen/Eigen/src/Eigenvalues/RealQZ.h,sha256=872Nd9lFrTCQwhtLWCyAtnZeG1mnaGWi1Q60IqVWGYo,23640
		eigency/eigen/Eigen/src/Eigenvalues/MatrixBaseEigenvalues.h,sha256=uxKK8knBdzSTGVJCn8qt07J_SqGDtNfT7e2z4wq1Jeg,5575
		eigency/eigen/Eigen/src/PaStiXSupport/PaStiXSupport.h,sha256=guI9MhWY5GTjrYXHWWRI6mXbUfaB23ooNEgTM35fLvA,22249
		eigency/eigen/Eigen/src/KLUSupport/KLUSupport.h,sha256=txjmMFEbp2X2xCmPMDheZGtR-wA_AnQWSCSnn7sAVRQ,11555

-1

top_level.txt

eigency

-6

WHEEL

		Wheel-Version: 1.0
		Generator: setuptools (82.0.0)
		Root-Is-Purelib: false
		Tag: cp310-cp310-macosx_11_0_arm64
		Generator: delocate 0.13.0

eigency/conversions.cpp

Sorry, the diff of this file is too big to display

eigency/core.cpp

Sorry, the diff of this file is too big to display

eigency - npm Package Compare versions

Improved metrics