newton_method

Introduction

This package contains functions to calculate the minimum of a mathematical function written as a lambda using Newton's gradient method (https://en.wikipedia.org/wiki/Newton%27s_method_in_optimization).

This package contains the following functions:

df_dx(func) returns a lambda function of the partial derivative on x direction
df_dy(func) returns a lambda function of the partial derivative on y direction
df_dxdx(func) returns a lambda function of the double partial derivative on x direction
df_dxdy(func) returns a lambda function of the mixed double partial derivative on x and y directions
df_dydy(func) returns a lambda function of the double partial derivative on y direction
df_dydx(func) returns a lambda function of the mixed double partial derivative on y and x directions
grad(func) returns an array of lambda functions (gradient of the function func)
hess(func) returns a matrix of lambda functions (hessian of the function func)
newton(func, x0=(0, 0), steps=2, delta=.05) returns the minimum point of the function from the point x0 with steps steps or until the delta is under delta
plot(func, start=-10, end=10, bins=100, savefig=False, plotfunc=True, plotgrad=True, filename='') plots the function func with its gradient if its a 2 variable function, in case it's a 2D function plots the function, its first and second derivative

func has to be a lambda function, for example:

newton_method.df_dx(lambda x,y: x ** 2 + y ** 2)

install this package using pip:

pip install newton_method

clone this repo and run setup.py:

git clone https://github.com/aXhyra/newton_method
cd newton_method
pip install .