0

I'm tuning a function to data over 4-5 variables using Differential Evolution. The objective function calls frequently interpolation function. I came to a point, where compiling the objective function should make things much faster. But interpolation function cannot be compiled.

Is there compilable implementation of linear interpolation function over 2 variables that is faster or of similar speed with uncompilable internal interpolation function?

EDIT

looks like this is working solution:

How to reduce the InterpolatingFunction building overhead?

Community
  • 1
den.run.ai
  • 972
  • 5
  • 16
  • It's hard to give suggestions without a specific code sample, can you add one? – xzczd Apr 18 '14 at 04:04
  • no code is necessary, it is simple linear interpolation with 2 arguments on a regular grid – den.run.ai Aug 29 '15 at 02:47
  • All the functions in the thread you linked to are for the univariate case. I presume you want bilinear interpolation? Some more work will then be needed. – J. M.'s missing motivation Aug 29 '15 at 14:30
  • I meant that for your specific problem someone might found a way for speeding up which is not "writing a compilable interpolating function" :) – xzczd Sep 01 '15 at 08:20

0 Answers0