HP Forums

Full Version: Modified Artificial Cooperative Search (ACS) Algorithm for optimization
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
Hi All,

I have posted material for an evolutionary optimization algorithm, titled "Modified Artificial Cooperative Search (ACS) Algorithm for optimization." Click here to get to the page on my web site and located the last entry with the above title. There is a link to download the PDF file and another link to download a ZIP file containing Matlab files.

The Artificial Cooperative Search (ACS) Algorithm is robust and enjoys fast convergence and very good accuracy. The article presents the original ACS algorithm and introduces three variations to this algorithm that also exhibit fast convergence and very good accuracy.

Namir[/php][/code]
Nice Namir that you keep doing your searches (and sharing them). They are nice pointers.
Thank you for your comment. I have found a long list of evolutionary algorithms, some newer than others. I am going through the list and trying to find or write programs for them in Matlab and/or Excel VBA.

Namir
I don’t think that rand < rand will have a normal distribution as claimed. I think it will be a triangular distribution being the difference of two uniforms. Moreover, since the difference isn’t being looked at, rand < rand will be equivalent to rand < 1/2.

Rand < prob * rand should collapse similarly to rand < prob/2.

Assuming throughout that rand is producing IID U(0, 1) random variables.

Pauli
(03-01-2018 05:48 AM)Paul Dale Wrote: [ -> ]I don’t think that rand < rand will have a normal distribution as claimed. I think it will be a triangular distribution being the difference of two uniforms. Moreover, since the difference isn’t being looked at, rand < rand will be equivalent to rand < 1/2.

Rand < prob * rand should collapse similarly to rand < prob/2.

Assuming throughout that rand is producing IID U(0, 1) random variables.

Pauli

Correct. Of course, that may be OK. I've used rand-rand for a random walk as compounding this just acts like other random walks (they mostly look alike.)
(03-01-2018 05:48 AM)Paul Dale Wrote: [ -> ]I don’t think that rand < rand will have a normal distribution as claimed. I think it will be a triangular distribution being the difference of two uniforms. Moreover, since the difference isn’t being looked at, rand < rand will be equivalent to rand < 1/2.

Rand < prob * rand should collapse similarly to rand < prob/2.

Assuming throughout that rand is producing IID U(0, 1) random variables.

Pauli

I checked the Internet and it seems you are correct. The difference of two independent standard uniform random variables has the
standard triangular distribution.
Reference URL's