Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Search algorithms

Author: Robert Hyatt

Date: 09:41:18 11/08/03

Go up one level in this thread


On November 07, 2003 at 09:22:45, Gian-Carlo Pascutto wrote:

>On November 06, 2003 at 19:48:33, Robert Hyatt wrote:
>
>>On November 06, 2003 at 10:18:37, Gian-Carlo Pascutto wrote:
>>
>>>On November 06, 2003 at 09:47:32, Robert Hyatt wrote:
>>>
>>>>I'm not sure how you could say that the distinction is very hazy.  They
>>>>are as different as night and day...
>>>
>>>Read the thesis, you will understand.
>>>
>>>--
>>>GCP
>>
>>This is one of those cases where I don't have to "read the thesis to
>>understand".  There's a _ton_ of books that discuss the differences
>>between breadth-first and depth-first search strategies.  They are
>>_never_ mentioned in the same paragraph as having similar properties.
>>Because they don't.
>
>They search the same (amount of) nodes. For game tree search, that's a pretty
>fucking important performance criterium.

No it isn't.  Suppose one is 100x _slower_ than the other?


>
>Discovering the similarity between SSS and AlpahBeta+TT allowed MTD(n,f) to
>be developed, which is one of the best tree searching algorithms out
>there.

MTD(f) is pure alpha/beta depth-first.  Nothing else.


>
>You're saying that's all irrelevant because it's best first and depth
>first and hence they have 'automatically' nothing to do with each other
>and anyone suggesting otherwhise is 'automatically' wrong even though
>you've not read the articles explaining it?

Why do you conclude I haven't read his thesis?  Do you recall that _I_
played with MTD(f) for several months?  Where do you think I got the
info???


>
>Geez, maybe tomorrow you'll start giving 'proof by induction' too.

That is what you are doing, not me.  Just because two searches happen to
search the same tree does _not_ mean they are equivalent algorithms.  To
suggest such is lunacy...

>
>--
>GCP



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.