RomiChess useing Dann Corbit's smooth formula for 'R'

[Michael Sherwin]

Olithink 5.xx seems to be a nemeses to RomiChess--doing much better than it should. Recent changes have brought Romi's score up from about 55% to about 70%. A recent test had Romi at 75% when it was noticed that Arena 2.0.1 was stuck on position 32 of my test set for 6 games in a row. The 75% was before the last 4 games were played--Romi won all 6. Wowie's always come with a price! I had to reinstall Arena to fix this problem.

With Dann's formula unchanged Romi is doing quite well even though she has gotten of to a very slow start.

RomiChess96 - Olithinkwin32 : 21.0/29 17-4-8 (=10===11111101==10=1111110=11) 72% +164

Romi is usually up 5/6 or 4.5/6 in the first 6 games, not even. But, the overall percentage is better at this stage of 29 games. The percentage should climb from here. The last 25 games of the match usually see a small decline.

Dann's slightly tuned formula for Stockfish should not be very optimal for RomiChess so there is hope for a better result.

Now:
RomiChess96 - Olithinkwin32 : 23.0/31 19-4-8 (=10===11111101==10=1111110=1111) 74% +182

More later!

[Michael Sherwin]

RomiChess96 - Olithinkwin32 : 26.0/35 21-4-10 (=10===11111101==10=1111110=1111=1=1) 74% +182

At this stage of the match the number of draws is up, HOWEVER the number of losses is way way down. If it wasn't for the slow start Romi would be way ahead at this point. Hopefully it wont take much tweaking to start Romi flirting with the 80% level!

[Michael Sherwin]

Romi at 75%! If this keeps up, Romi will be flirting with 80% before this match is over!!

RomiChess96 - Olithinkwin32 : 30.0/40 25-5-10 (=10===11111101==10=1111110=1111=1=101111) 75% +191

[Michael Sherwin]

I had to discontinue the tournament because Arena 2.0.1 was repeating position 32 again every 4th game when Romi had the white pieces. I really love the brown and tan canvas playing board. Oh well back to Arena 1.1 for me and its ugly board.

I am changing the formula to this for the next run.

Code: Select all

    double delta;
    double ddepth;
    int r;

    if&#40;depth < 6&#41; r = 3; else &#123;
      delta = max&#40;h->eval - beta, 1.0&#41;; 
      ddepth = &#40;double&#41;depth; 
      r = &#40;int&#41;&#40;0.18 * ddepth + 3.1 + log&#40;delta&#41;/5.0&#41;; 
    &#125;

[michiguel]

I had to discontinue the tournament because Arena 2.0.1 was repeating position 32 again every 4th game when Romi had the white pieces. I really love the brown and tan canvas playing board. Oh well back to Arena 1.1 for me and its ugly board.

I am changing the formula to this for the next run.

Code: Select all

    double delta;
    double ddepth;
    int r;

    if&#40;depth < 6&#41; r = 3; else &#123;
      delta = max&#40;h->eval - beta, 1.0&#41;; 
      ddepth = &#40;double&#41;depth; 
      r = &#40;int&#41;&#40;0.18 * ddepth + 3.1 + log&#40;delta&#41;/5.0&#41;; 
    &#125;

Don't you like Winboard?

What did you mean before that RC should perform better as the time progresses? How do you know?

Miguel

[George Tsavdaris]

Olithink 5.xx seems to be a nemeses to RomiChess--doing much better than it should. Recent changes have brought Romi's score up from about 55% to about 70%. A recent test had Romi at 75% when it was noticed that Arena 2.0.1 was stuck on position 32 of my test set for 6 games in a row. The 75% was before the last 4 games were played--Romi won all 6.

Do you want to say that the normal "unchanged" Romi's score is about 55% and with the Dann's changed has gone to 70% and 75% in your last test?
Because you've described things in a bit complicated way for me, i must say.

[Michael Sherwin]

I had to discontinue the tournament because Arena 2.0.1 was repeating position 32 again every 4th game when Romi had the white pieces. I really love the brown and tan canvas playing board. Oh well back to Arena 1.1 for me and its ugly board.

I am changing the formula to this for the next run.

Code: Select all

    double delta;
    double ddepth;
    int r;

    if&#40;depth < 6&#41; r = 3; else &#123;
      delta = max&#40;h->eval - beta, 1.0&#41;; 
      ddepth = &#40;double&#41;depth; 
      r = &#40;int&#41;&#40;0.18 * ddepth + 3.1 + log&#40;delta&#41;/5.0&#41;; 
    &#125;

Don't you like Winboard?

What did you mean before that RC should perform better as the time progresses? How do you know?

Miguel

Does the newer Winboard do tournaments with out a helper program?

I expected Romi to increase her percentage because it normally does during the middle section of games from my test set.

[Michael Sherwin]

Olithink 5.xx seems to be a nemeses to RomiChess--doing much better than it should. Recent changes have brought Romi's score up from about 55% to about 70%. A recent test had Romi at 75% when it was noticed that Arena 2.0.1 was stuck on position 32 of my test set for 6 games in a row. The 75% was before the last 4 games were played--Romi won all 6.

Do you want to say that the normal "unchanged" Romi's score is about 55% and with the Dann's changed has gone to 70% and 75% in your last test?
Because you've described things in a bit complicated way for me, i must say.

I had made progress on Romi before Dann's formula was published. I am just trying to prove it with the scant very little time I have. I was hoping that using Dann's formula would help. I was just giving some background to define where I was with Romi when Dann's formula was introduced.

[Michael Sherwin]

I had to discontinue the tournament because Arena 2.0.1 was repeating position 32 again every 4th game when Romi had the white pieces. I really love the brown and tan canvas playing board. Oh well back to Arena 1.1 for me and its ugly board.

I am changing the formula to this for the next run.

Code: Select all

    double delta;
    double ddepth;
    int r;

    if&#40;depth < 6&#41; r = 3; else &#123;
      delta = max&#40;h->eval - beta, 1.0&#41;; 
      ddepth = &#40;double&#41;depth; 
      r = &#40;int&#41;&#40;0.18 * ddepth + 3.1 + log&#40;delta&#41;/5.0&#41;; 
    &#125;

Well, the above one was terrible.

The following one is doing fantastic!

Code: Select all

    double delta = max(h->eval - beta, 1.0); 
    double ddepth = (double)depth; 
    int r = (int)(0.25 * ddepth + 2.5 + log(delta)/5.0); 

RomiChess96 - Olithinkwin32 : 11.5/13 11-1-1 (101=111111111) 88% +346

Don't anyone dare say 'WOWIE!'!

[George Tsavdaris]

I am changing the formula to this for the next run.

Code: Select all

    double delta;
    double ddepth;
    int r;

    if&#40;depth < 6&#41; r = 3; else &#123;
      delta = max&#40;h->eval - beta, 1.0&#41;; 
      ddepth = &#40;double&#41;depth; 
      r = &#40;int&#41;&#40;0.18 * ddepth + 3.1 + log&#40;delta&#41;/5.0&#41;; 
    &#125;

Well, the above one was terrible.

The following one is doing fantastic!

Code: Select all

    double delta = max(h->eval - beta, 1.0); 
    double ddepth = (double)depth; 
    int r = (int)(0.25 * ddepth + 2.5 + log(delta)/5.0); 

RomiChess96 - Olithinkwin32 : 11.5/13 11-1-1 (101=111111111) 88% +346

Don't anyone dare say 'WOWIE!'!

OK can i dare to ask for an explanation of all these?

I mean what exactly these 2 lines do?
double delta = max(h->eval - beta, 1.0);
int r = (int)(0.25 * ddepth + 2.5 + log(delta)/5.0);

And how did you find them?

Also Dann mentioned he predicted almost by hand his formula for Stockfish. How did he do that exactly?
And what one should do to do it in a more scientific way(regression analysis, interpolation etc, but how exactly)?

Also what Dann's method does more or less? How it works i mean. What different thing it does. (Note that i'm not a programmer but i'm trying- with a turtle speed- to get to that side.)