Author: Uri Blass
Date: 13:21:43 09/29/04
Go up one level in this thread
On September 28, 2004 at 22:45:47, Stuart Cracraft wrote:
>On September 28, 2004 at 13:39:44, Henk Bossinade wrote:
>
>>On September 28, 2004 at 01:04:52, Stuart Cracraft wrote:
>>
>>>On September 26, 2004 at 15:53:53, David B Weller wrote:
>>>
>>>> 2 -223 0.01 375 {12} Kf1 Re2
>>>> 3 -205 0.02 1061 {13} Kf1 Re2 Kg1
>>>>4r1k1/p1qr1p2/2pb1Bp1/1p5p/3P1n1R/1B3P2/PP3PK1/2Q4R w - - bm Qxf4; id "WAC.141";
>>>>
>>>> 4 -239 0.05 3815 {18} Kf1 Re2 Qc3 b4
>>>> 5+ -199 0.18 16983 {23} Kf1
>>>> 5 -185 0.35 36630 {23} Kf1 a5 Qb1 a4 Bc2
>>>> 6 -206 0.86 91286 {25} Kf1 Re2 Qb1 Rd2 Kg1 Nd3
>>>> 7 -175 2.48 248962 {27} Kf1 Re2 Qb1 Rd2 Qc1 Rd3 Be5 Bxe5 dxe5
>>>> 8+ -135 7.29 832695 {29} Qxf4
>>>> 8 299 9.88 1150509 {29} Qxf4 Bxf4 Rxh5 gxh5 Rxh5 Bh6 Rxh6 Qh2+ Rxh2
>>>> 9 299 13.58 1503202 {29} Qxf4 Bxf4 Rxh5 gxh5 Rxh5 Bh6 Rxh6 Qh2+ Rxh2
>>>>10+ 339 25.34 2971184 {30} Qxf4
>>>>Knps: 94[r=745471 q=4971821(86%)]Qxf4
>>>>fh=83% bf=3.17
>>>>ext: ck=107333 p7=1950 mt=1051 1r=1488 re=2311 delay=0 good=0
>>>>red: nm=93529 fc=16253558 zg=0
>>>>ht=88%
>>>>move c1f4
>>>>
>>>David,
>>>
>>>Hey that's great -- only 7.29 seconds. Congratulations.
>>>
>>>Okay, so what are the goods? How did you do it?
>>>
>>>Can you repost your web URL for the GES that did the above and
>>>cite the relevant portions of code you believe are responsible for
>>>your superior time for Qxf4!
>>>
>>>Stuart
>>
>>Here's my wac141 with mate threat:
>>
>> time score nodes pv
>> 5. 0.89 -70 84104 Kg2f1 Re8e2 Qc1b1 Re2d2 Qb1c1 Rd2e2 Qc1b1
>> 6 4.51 -70 470874 Kg2f1 Re8f8 Qc1b1 Nf4d5 Bf6g5 Qc7c8 Qb1e1
>> 6 7.27 -69 762159 Qc1xNf4 Bd6xQf4 Rh4xPh5 Pg6xRh5 Rh1xPh5 Bf4h6
>> 6. 7.27 -69 762159 Qc1xNf4 Bd6xQf4 Rh4xPh5 Pg6xRh5 Rh1xPh5 Bf4h6
>> 7 9.21 -69 1026574 Qc1xNf4 Bd6xQf4 Rh4xPh5 Pg6xRh5 Rh1xPh5 Bf4h6
>> 7. 11.86 -69 1259450 Qc1xNf4 Bd6xQf4 Rh4xPh5 Pg6xRh5 Rh1xPh5 Bf4h6
>> 8 32.27 ++ 3422740 Qc1xNf4 Bd6xQf4 Rh4xPh5 Pg6xRh5 Rh1xPh5 Bf4h6
>> 8 83.38 260 9293559 Qc1xNf4 Bd6xQf4 Rh4xPh5 Pg6xRh5
>> 8. 87.74 260 9778848 Qc1xNf4 Bd6xQf4 Rh4xPh5 Pg6xRh5
>>
>>here without:
>>
>> 7. 13.49 -70 1458658 Kg2f1 Re8f8 Qc1b1
>> 8 90.45 -70 9788445 Kg2f1 Pa7a5 Rh1g1
>> 8 112.56 ++ 11842418 Qc1xNf4 Bd6xQf4 Rh4xPh5 Pg6xRh5 Rh1xPh5
>> 8 157.69 260 16856578 Qc1xNf4 Bd6xQf4 Rh4xPh5 Pg6xRh5
>> 8. 160.32 260 17120337 Qc1xNf4 Bd6xQf4 Rh4xPh5 Pg6xRh5
>>
>>I noticed in the search() code you posted that you return a mate score like
>>this:
>>
>> if (legals == 0) {
>> if (checked) {
>> best = -MATE+depth;
>> } else {
>> best = STALEMATE;
>> }
>>
>>and your mate threat code looks like this:
>>
>> if (!extended && value == -MATE+ply+2) {
>>// printf("mate threat:\n");listmvs(hist);pbd(bd);getchar();
>> extend=1;
>> extended=1;
>> depth++;
>> }
>>
>>I think you should use either depth/depth or ply/ply but not like it is.
>
>I've chosen ply/ply. I was told ply is better since it returns a lower
>value for a mate discovered higher in the tree. Not sure if depth/depth
>or ply/ply minimizes the moves to mate. Thought depth/depth did that...
>
>>Also instead of 'value == -MATE+ply+2' you can try 'value <= -MATE+maxply'
>>
>
>I've tried that in the past. It blows up my search.
I think that it is clear that extending every nate threat by a full ply will
blow your search.
If you do not use partial extensions then I suggest that you do not extend mate
threats unless you have some conditions to extend them only near the root and
not every where.
Uri
This page took 0.19 seconds to execute
Last modified: Thu, 07 Jul 11 08:48:38 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.