Open
Description
I was checking the implementation of the NedlerMeadSimplex
algorithm while looking at a few different code-implementations in order to better understand the algorithm and noticed a few discrepancies with how the algorithm is described both here and here
Namely
- The expansion step uses a chi-value of -2 (2 as the function parameter but the function actually flips the sign in comparison with the paper)
- All
TryToScaleSimplex
functions recalculate the centroid instead of calculating it only once per iteration
Is there any reason for those changes or a source paper? Or are those bugs?
Metadata
Metadata
Assignees
Labels
No labels