pijavski 0.0.5

Python CFFI bindings for Pijavski C++ function to calculate minimum of a given function.

Pijavski

This package uses CFFI to call the function Pijavski written in C++ that optimises a test function and returns the minimum.

Installation

To install simply type:

$ pip install pijavski

Usage

To test, open a python console and import the package. The function pijavski.get_minimum, with arguments lip, xl, xu, precision and maxiter, prints res, x0, f, prec, maxit as result.

>>> import pijavski
>>> pijavski.get_minimum()
0 314.1592651760589 -0.9999999991415484 8.584516431042744e-10 65533

Defining custom functions to optimise

The function to optimise needs to be declared following the conditions below:

• When writing the function, arguments f and x need to be declared as if they were pointers using the bracket notation f[] and x[].

• Use numpy math functions.

• Modify the definition of the function generic_fun to received as argument the name of your custom function.

Example:

>>> # Simple declaration of f = -cos^2(x) as callback function.
>>> import numpy as np
>>> from pijavski import get_minimum, ffi, lib
>>> def my_fun(x, f):
...     f[0] = (-1)*np.cos(x[0])**2
>>> def generic_fun(x, f, myfun=my_fun):
...     return myfun(x, f)
>>> # Call get_minimum
>>> get_minimum(func=lib.fun, xl=-100, xu=100)
0 -1.0 4.6838846e-317 4.6838846e-317 1