ghostjat / np

A Lite & Memory Efficient PHP Library for Scientific Computing

Description

---

Lite, Fast & Memory Efficient Mathematical PHP library for scientific computing

Np(numphp) is a library that provides objects for computing large sets of numbers in PHP.

Installation

Install Np into your project with Composer:

$ composer require ghostjat/np

##Sample Code

require __DIR__ . '/../vendor/autoload.php';
use Np\matrix;

$ta = matrix::randn(1000, 1000);    
$tb = matrix::randn(1000, 1000); // to generate random 2d matrix
$ta->dot($tb);                  // do a dot operation on given matrix
$ta->getMemory();              // get memory use
$ta->time();                  // get time
/**
 * 7.7mb
 * Time-Consumed:- 0.18390893936157
 */

Synopsis

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.

Requirements

• PHP 8+ 64bit with ffi & #libblas, #liblapacke

Make sure you have all the necessary tools installed such as FFI, libblas, liblapacke.

Performance

System Conf:- Intel(R) Core(TM) i3-2370M CPU @ 2.40GHz 64bit Memory:- 8GB php:- 8.0.5 64bit

Current Benchmarks of this library

Data Size :- [500x500] Revolutions:- 5 Iterations:- 5

subject	mem_peak	best	mode	mean	worst	stdev
sum	3.606mb	0.014s	0.014s	0.015s	0.015s	0.000s
multiply	8.589mb	0.070s	0.071s	0.071s	0.071s	0.000s
lu	4.648mb	0.064s	0.065s	0.065s	0.068s	0.001s
eign	2.801mb	0.085s	0.086s	0.086s	0.088s	0.001s
cholesky	1.621mb	0.001s	0.001s	0.001s	0.001s	0.000s
svd	3.706mb	0.126s	0.126s	0.127s	0.133s	0.002s
normL2	1.621mb	0.003s	0.003s	0.003s	0.003s	0.000s
Pinverse	4.903mb	0.156s	0.156s	0.158s	0.163s	0.003s
inverse	1.819mb	0.016s	0.016s	0.016s	0.017s	0.000s
normL1	1.621mb	0.001s	0.001s	0.001s	0.001s	0.000s
dotMatrix	3.769mb	0.006s	0.006s	0.006s	0.006s	0.000s
det	4.662mb	0.066s	0.066s	0.067s	0.067s	0.000s
rref	1.529mb	9.227s	9.271s	9.309s	9.427s	0.072s
ref	1.818mb	0.007s	0.008s	0.008s	0.008s	0.000s
clip	8.516mb	0.073s	0.076s	0.075s	0.077s	0.002s
clipUpper	8.516mb	0.055s	0.056s	0.057s	0.059s	0.002s
clipLower	8.516mb	0.055s	0.058s	0.057s	0.059s	0.002s
joinBelow	4.517mb	0.027s	0.027s	0.027s	0.028s	0.000s
transpose	8.504mb	0.057s	0.057s	0.058s	0.059s	0.001s
joinLeft	4.511mb	0.025s	0.025s	0.026s	0.027s	0.001s
poisson	1.590mb	0.029s	0.029s	0.029s	0.030s	0.000s
gaussian	20.203mb	0.056s	0.056s	0.056s	0.056s	0.000s
randn	1.528mb	0.017s	0.017s	0.017s	0.017s	0.000s
uniform	1.528mb	0.021s	0.021s	0.021s	0.022s	0.000s
multiply	4.507mb	0.042s	0.042s	0.043s	0.045s	0.001s

Previous BenchMark

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%

License

The code is licensed MIT and the documentation is licensed CC BY-NC 4.0.

Author

Shubham Chaudhary [email protected]