Package boxmin
[hide private]

Package boxmin

source code

Box constrained minimization functions.

Contains amoeba.

Fit functions may be implemented in python or in C.

Installation

You can download the latest version from svn:

svn co svn://danse.us/boxmin/trunk

If you are installing from source, you will need a python environment with numpy. You will need a C compiler to build c lib. On Windows this may require installing MinGW and adding distutils.cfg to your distutils directory:

[build]
compiler = mingw

Once you have the required supporting packages, use the following to build and install:

python setup.py install

Usage

Example using python function:

>>> from boxmin import amoeba
>>> def quadratic(vec):
>>>    return vec[0]**2+vec[1]**2+1
>>> F = amoeba.amoeba(quadratic,[1,1])
>>> p = F.fit(1e-7)

p should now contain the vector [x_0, x_1, f(x_0,x_1)] = [0, 0, 1]

See cfitfn_example in boxmin/python/tests for an example of how to define a C function which is directly callable from the fit.

Submodules [hide private]