Adaptive Interpolation Project (BETA)

This is a pythonic method for creating an approximation for a function on a given domain with a specified error. It is done by adaptively interpolating said function on the domain until the allowed error is reached. C code is then generated and returned which evaluates said interpolant. This code can then be run in the library using a method that uses pyopencl to run the code in parallel.

Quick and Easy Demonstration

To see a demonstration of this code first clone this git repository to your machine using:

git clone https://www.github.com/jdoherty7/Adaptive_Interpolation

Then run the demonstration by entering the cloned directory and running:

python demo.py

The code should begin interpolating three functions. One is a bessel function, one is a wavy sin function, and another is a piecewise function. These are generated sequentially. THE MATPLOTLIB PLOTS MUST BE CLOSED TO GENERATE THE NEXT INTERPOLANT. The plots display the estimated and actual values as well as the absolute errors between the estimated values and the actual values. The allowed relative error is shown as a red line.

NOTE: When prompted for a processor by pyopencl please press ENTER, without typting anything to continue.