Aspiration Windows: Rubbish!

[Desperado]

Hi,

have you tried to disable the transposition table. If the research gets confused it might encounter some rubbish there.

Thomas...

@Thomas
Good idea.

Transposition tables in context with mates scores are
always worth a _first_ look. Especially the relative mate score computation.

@Matthew
Another thing should be to localize where the "phantom" occures in the
search for the first time, maybe identified by nodecount or hashcode.

Many things can be spotted following the search, like buggy mate distance
pruning, wrong initialization of data and so on, and just following with the
debugger, if you know where to start.

If you catch a return value, you have to identify the successor of course, and start debugging from there.

The harder the bug is too find, the more important it gets to switch
of complete code parts, like Thomas mentioned already with the TT.

A simple point to check is of course if you are always returning valid scores in the search, and not an uninitialized best Value maybe, which will be returned because all moves are pruned somehow.
Better, have a closer look at bestValue handling immediatelly.

Just my 2 cents, hopefully it guides you in the right direction.

Michael

[ZirconiumX]

This is Fruit though. All of that should be fine.

Matthew:out

[lucasart]

So no-one pays any attention to what I have written. Fine.

It's not like anyone normally pays attention. Goodbye.

Lucas, that was fixed ages ago.

No wonder computer chess is dying. Because when they try to join, they get attacked. Marek, author of Sting had a great idea. But he has given up already. I will at least wait it out until the attack of the clones culture is gone.

Matthew:out

You really think computer chess will die because newbies like you are losing interest ? This is hilarious

[gleperlier]

You really think computer chess will die because newbies like you are losing interest ? This is hilarious

Please Lucas, no matter to argue.

Cheers,

Gab

[Desperado]

This is Fruit though. All of that should be fine.

Matthew:out

Well, that is Fruit modified by you. Obviously you have introduced a bug
somewhere in another stage of your development, or simple not adapted the code enough that it will run properly with aspiration windows.

Your statement that it is Fruit doesnt play a role, because you get the quittance in form of a phantom value. That seems to be the fact.

Now it is up to you to hunt this bug, and to make the aspiration window
technique work at least in a proper way. (independent if it will give additional strength). By the way, you may find a bug somewhere in
the original source, which was undiscovered so far, do you know without
examining it ?

You will not only debug the source, but you may discover a lot of other details ? Why do you think even much more experienced programmers have a look at Fruit again and again .... and again, from time to time !?!

As I said, it is only up to yourself. The points pointed out in my last post just should help to start somewhere.

Best

Michael

[tpetzke]

No, it is not Fruit. It is a modified something similar to chess engine formerly known as Fruit.

It is very difficult to improve a strong chess engine, especially if you are fairly new in this area. Chances are that whatever you change weakens the engine. Despite of its name I don't think there are low hanging fruits to be picked.

Just a small recommendation from my side. If you interested in chess programming, start implementing your own engine. Start simple, only add code you understand. It sounds boring, but it isn't. You learn a lot. You will not have any strength success stories to tell for the next time but just watching your engine getting slowly stronger is quite rewarding.

My first move generator took 90 seconds to generate perft 6, but I understood every bit of it.

Thomas...

[Richard Allbert]

Agreed, the most enjoyable part for me is writing something that passes perft.

Tuning and testing later is not so much fun.

[Richard Allbert]

Have a look at the code for Fruit 1.0, it does aspiration search, and is a UCI option. You can test it with that.

Richard

[michiguel]

I know this sounds like a search bug

Right.

Worth nomination for post of the year isn't it?

Also for nomination
http://talkchess.com/forum/viewtopic.php?t=44306

The technical forum is the proper place for newbies to request technical help with the chance of being helped (or ignored in the worst case). Ganging up on them is not really the purpose. If the request sounds outrageous, the best strategy is to let others help.

Miguel

[bob]

So no-one pays any attention to what I have written. Fine.

It's not like anyone normally pays attention. Goodbye.

Lucas, that was fixed ages ago.

No wonder computer chess is dying. Because when they try to join, they get attacked. Marek, author of Sting had a great idea. But he has given up already. I will at least wait it out until the attack of the clones culture is gone.

Matthew:out

What are you talking about? The fact that aspiration window works for everyone but you, so you consider it not useful, and the rest of us that use it successfully suspect you have one or more bugs? A 5 second 1 ply search certainly is not normal and suggests a bug. The only way aspiration search hurts is if you set the window too narrow so that you are continually having to re-search and widen the window. It doesn't always help, but it helps some. It doesn't always hurt, but it hurts some. Overall it helps more than it hurts. How much depends on several things.

No need to get your feelings hurt when someone suggests you have a bug. It is a near-certainty in this case...