Pseudo versus True Randomness (PRNGs versus TRNGs)
Jaap de Jonge, Management Consultant, Netherlands
Random numbers can be used for several different purposes, such as generating data encryption keys and explaining / simulating / modeling complex phenomena like evolution and many phenomena in the management domain.
When you want use a sequence of random numbers, each number drawn has to be statistically independent of the others.
According to Dr Mads Haahr at random.org, in randomness one has to distinguish pseudo from true (=chaotic) randomness and also pseudo from true (=chaotic) random number generators (PRNGs from TRNGs).
The characteristics of TRNGs have been found to be quite different from PRNGs:
||Pseudo-Random Number Generators
||True Random Number Generators
Without being an expert in this field, I suspect the scientific answer to your question (Why chaos improves evolutionary algorithms?) results from the above table.
On a more day to day level, it appears obvious that using chaotic approaches to simulate evolution is superior, since it is common knowledge that evolution also took place in chaotic circumstances.