numphp is a library that provides objects for computing large sets of numbers in PHP.
Install Numphp into your project with Composer:
$ composer require ghostjat/numphp##Sample Code
require __DIR__ . '/../vendor/autoload.php';
use numphp\matrix;
matrix::time();
matrix::getMemory();
$ta = matrix::randn(1000, 1000);
$tb = matrix::randn(1000, 1000); // to generate random 2d matrix
$ta->dotMatrix($tb); // do a dot operation on given matrix
matrix::getMemory(); // get memory use
matrix::time(); // get time
/**
* Memory-Consumed 7.7mb
* Time-Consumed:- 0.19370794296265
*/WARNING:
This module is in its early stages and should be considered a Work in Progress.
The interface is not final and may change in the future.
- PHP 7.4+ with ffi & #libblas, #liblapacke
Make sure you have all the necessary tools installed such as FFI, libblas, liblapacke.
System-Conf:- CPU:- Intel(R) Core(TM) i3-2370M CPU @ 2.40GHz 64bit MEM:- 8GB PhpBench @git_tag@. Running benchmarks. Using configuration file: /home/ghost/projects/git/numphp/phpbench.json
| benchmark | subject | set | revs | its | mem_peak | mode | rstdev |
|---|---|---|---|---|---|---|---|
| eignBench | eign | 0 | 1 | 5 | 2.699mb | 0.309s | ±4.51% |
| svdBench | svd | 0 | 1 | 5 | 3.604mb | 0.148s | ±3.60% |
| poissonMatrixBench | poisson | 0 | 1 | 5 | 11.738mb | 0.105s | ±7.07% |
| gaussianMatrixBench | gaussian | 0 | 1 | 5 | 11.738mb | 0.112s | ±17.12% |
| randMatrixBench | randn | 0 | 1 | 5 | 1.429mb | 0.048s | ±2.37% |
| uniformMatrixBench | uniform | 0 | 1 | 5 | 1.429mb | 0.063s | ±8.16% |
| matrixTransposeBench | transpose | 0 | 1 | 5 | 8.431mb | 0.120s | ±1.32% |
| rrefBench | rref | 0 | 1 | 5 | 1.501mb | 28.513s | ±1.90% |
| refBench | ref | 0 | 1 | 5 | 1.731mb | 0.023s | ±7.24% |
| sumMatrixBench | sum | 0 | 1 | 5 | 2.434mb | 0.051s | ±3.59% |
| matrixPseudoInverseBench | inverse | 0 | 1 | 5 | 4.775mb | 0.222s | ±13.76% |
| matrixInverseBench | inverse | 0 | 1 | 5 | 1.731mb | 0.032s | ±127.50% |
| dotMatrixBench | dotMatrix | 0 | 1 | 5 | 3.656mb | 0.013s | ±27.94% |
| matrixL1NormBench | normL1 | 0 | 1 | 10 | 1.525mb | 0.001s | ±0.80% |
| matrixL2NormBench | normL2 | 0 | 1 | 10 | 1.525mb | 0.003s | ±1.63% |
The code is licensed MIT and the documentation is licensed CC BY-NC 4.0.
Shubham Chaudhary ghost.jat@gmail.com