exactnumber

ExactNumber

Arbitrary-precision decimals. Enables making math calculations with rational numbers, without precision loss.

Quick comparision with JS floating-point arithmetic

console.log(1 + 0.36);
// returns 1.3599999999999999

console.log(ExactNumber(1).add('0.36').toString());
// returns 1.36

console.log(1 / 3);
// returns 0.3333333333333333

console.log(ExactNumber(1).div(3).toString());
// returns 0.(3)

console.log((1 / 49) * 49);
// returns 0.9999999999999999

console.log(ExactNumber(1).div(49).mul(49).toString());
// returns 1

console.log(10e16 + 5);
// returns 100000000000000000

console.log(ExactNumber('10e16').add(5).toString());
// returns 100000000000000005

Features

• Works with arbitrary large numbers without precision loss
• There is no automatic rounding or truncation. Rounding can be archived by explicit calls to rounding functions
• Works with all number bases between 2 and 16
• There are no special values like NaN, Infinity or -0.
• No silent errors: it throws errors immediatelly when a confusing parameter is received.
• All fractions can be represented as repeating decimals like 1.23(45)
• Repeating decimal format can also be parsed back
• Supports bitwise operators (and, or, xor, shiftLeft, shiftRight)
• Approximation algorithms for irrational numbers (experimental)
• Supports all modern browsers, web workers, Node.js and Deno
• Includes TypeScript type definitions: documentation
• Zero external dependencies
• Uses BigInt type under the hood. It automatically switches back and forth between fixed-precision and fractional representations.
• Tries to deliver the best possible performance
• 100% open source + MIT license

Installation

npm i exactnumber

It can also be used directly from HTML (via jsDelivr):

<!-- loads the full, minified library into the global `exactnumber` variable -->
<script src="https://cdn.jsdelivr.net/npm/exactnumber"></script>

<!-- or loads the non-minified library -->
<script src="https://cdn.jsdelivr.net/npm/exactnumber/dist/index.umd.js"></script>

Usage

import { ExactNumber as N } from 'exactnumber';

N(1).add('3').toString(); // 4

N('1/7').add('1/10').toFraction(); // 17/70

N('1/7').toString(); // 0.(142857)
N('1/7').toFixed(3); // 0.142
N('1/7').trunc(3).toString(); // 0.142
N('0.(3)').add('0.(6)').toString(); // 1

N('0b1100').bitwiseAnd('0b1010').toString(2); // 1000

N.max('1/1', '10/2', 3).toString(); // 5
N.fromBase('123', 4).toString(); // 27

import { PI, sin } from 'exactnumber';
PI(10); // 3.1415926535

const PI_OVER_2 = N(PI(10)).div(2);
sin(PI_OVER_2, 5); // 1.00000

Supported approximations

• Roots: sqrt(), cbrt(), nthroot()
• Exponentials: pow(), exp()
• Logarithms: log(), logn(), log10(), log2(),
• Constants: PI()
• Trigonometric functions: sin(), cos(), tan()
• Inverse trigonometric functions: asin(), acos(), atan()
• Hyperbolic functions: sinh(), cosh(), tanh()
• Inverse hyperbolic functions: asinh(), acosh(), atanh()

Copyright

License: MIT

Copyright © 2022 Dani Biró

Changelog

0.11.0 (October 29, 2022)

• Faster rounding and more reliable normalization of fractions
• ExactNumber.gcd() and ExactNumber.lcm()
• Limit precision of input binaries at approximations