It is motivated by Gianluca Dorini's isParetoSetMember program. The new m-file version is much faster than the C version because of the more elegant algorithm. The efficiency is significantly improved in version 3. By implementing a new sorting scheme and recoding to reduce overhead, the code is even faster than the mex version, paretomember, where sorting is not adopted. However, it puzzles me that the performance of mex code does not affected by sorting. Follow the links bellow to download these two codes for comparison.
A more efficient version available in
Tried on 2011a 64bit and didn't get corret results (only two points). Seems that default dimensions changed the fix is adding the ',2' for the max direction.
Corrected: line 23 of pretoset.m