quantagonia-api-client

Quantagonia API Client

This package contains Quantagonia's CLI client and client-side APIs for using HybridSolver in the cloud.

Installation

Python >= 3.8 is required. Then, do

python -m pip install quantagonia-api-client

In order to use HybridSolver through Quantagonia's cloud system, you'll need to sign up for an API key and set it system-wide by

export QUANTAGONIA_API_KEY=<your API key here>

Quick start: using the CLI client

This package provides the script hybridsolver to interact with the cloud solver. In order to solve a MIP (supported formats: .mps, .lp, .mps.gz, .lp.gz) or QUBO (spported formats: .qubo, .qubo.gz), just use

hybridsolver solve path/to/example.mps

which will automatically set all parameters and solver controls to their default value, which is a good compromise for many middle-sized optimization problems out there.

The solve call has a few options (try hybridsolver solve --help) and will, after submitting the problem, stream the logs and output the solution vector if one is found.

To start a job without attaching to its log stream and leave it running in the background, use

hybridsolver submit path/to/example.mps

which supports the same parameters as solve and returns a job_id. using this ID, logs, results and job status as well as time billed can be retrieved using

hybridsolver (logs | solution | status | time_billed) job_id

Lastly, in order to cancel a runnning job and avoid unnecessary costs, use hybridsolver cancel job_id. To list k active and/or past jobs, use hybridsolver list --n=k.

Using the API client in your own code

The capabilities of the API client include, but are not limited to:

• a plugin for PuLP to solve PuLP-generated models locally and in the client,
• a simple API to model QUBOs locally and solve them in the cloud and
• an importer to convert PyQUBO models into our format - allowing to solve PyQUBO models in QUantagonia's cloud.

For starters, you can run the examples in examples/ through, e.g.,

QUANTAGONIA_API_KEY="<redacted>" python examples/submit_qubo.py

with a valid Quantagonia API key.

For more details, we refer to our API reference.