Add rank based bonus for blocked pawns.

[locutus2]

This is a rewrite of my original idea which uses only two parameters. Thanks to rocky640 for pointing this out.

STC:
LLR: 2.94 (-2.94,2.94) {-0.50,1.50}
Total: 50800 W: 9707 L: 9446 D: 31647
Ptnml(0-2): 831, 5851, 11822, 6018, 878
https://tests.stockfishchess.org/tests/view/5f00b4f359f6f03532894304

LTC:
LLR: 2.93 (-2.94,2.94) {0.25,1.75}
Total: 52064 W: 6477 L: 6167 D: 39420
Ptnml(0-2): 331, 4628, 15834, 4878, 361
https://tests.stockfishchess.org/tests/view/5f0115fe59f6f03532894345

Bench: 4882833

[mstembera]

Congrats! Just curious why it makes sense to stop at rank 3?

[Rocky640]

example 1; white e4, black e5. Both pawn on relative rank r = RANK_4. So the bonus will cancel out.

example 2: white e3 black e4. White pawn on r=RANK_3, Black pawn on r=RANK_5
let's say you give bonus S(a,b) or white pawn, and bonus S(c, d) for black pawn.
at the end of the day, you can simply give S(a-c, b-d) for white pawn, and S(0,0) for black

example 3: white e2 black e3. White pawn on r=RANK_2, Black pawn on r=RANK_6
let's say you give bonus S(a,b) or white pawn, and bonus S(c, d) for black pawn.
at the end of the day, you can simply give S(a-c, b-d) for white pawn, and S(0,0) for black

Therefore it is useless to score blocked pawn on r>=RANK_4 so we score only for r < RANK_4

[locutus2]

if it feels more familiar i could do a non-functional rewrite so that rank 5 and 6 are get the bonus.

constexpr Score BlockedPawn[2] = { S(-10, -3), S(-3, 3) };

if (blocked && r > RANK_4)
            score += BlockedPawn[r-4];

[snicolet]

if it feels more familiar i could do a non-functional rewrite so that rank 5 and 6 are get the bonus.

Yes, that would be an incentive for developers to work on the subject or tweak the bonuses, as we think of fawn pawns as advanced pawns in the 5th and 6th ranks.

[locutus2]

@snicolet
Ok, i have pushed the rewritten version and add a comment at the bonus that its clear that this counts for 5th and 6th rank.

[locutus2]

Correct grammatical error in commit and PR message (insert is)

[xoto10]

The only thing that puzzles me is that the bonus is almost always negative, so it's really a penalty for advanced pawns :(

On the upside, it gives us more flexibility/detail in scoring these pawns, which should be a good thing, and perhaps future tuning will refine the various values that affect these pawns :)

[locutus2]

@xoto10
This values simply say that we currently overevaluated blocked pawns on the 5th and 6th rank the most time.

[vondele]

Thanks!