Previous   Next   Version 4

Search: Minimax

Minimax

The search is done by the recursively called routine D(). This is the only subroutine in micro-Max, all other code was in-lined to save characters on calls and declarations. But for the recursive search this is not possible, of course. D() has a lot of arguments; the ones governing the search are: the side to move k which has to be toggled (as 24-k) for each ply, the alpha-beta window (q,l), and the 'depth-left' argument n. For the rest its code is rather conventional (this section supposes you are familiar with the principles of alpha-beta minimax, and just discusses the way it is implemented in Micro-Max. If not, an excellent explanation of general principles of computer chess can be found on Bruce Moreland's web site):


int D(int k, int q, int l, int n)
{
	int m;				/* score of best move so far */

	m = n>1 ? -I : EVAL();

	INITIALIZE_MOVE_GENERATION;
	DO{{{
		GENERATE_NEXT_MOVE;
		if(m>=l) goto C;
		h = n-1;		/* new depth left */
		if(h)
		{	DO_MOVE;
			v = -D(24-k, -l, q>m?-q:-m, h);
			UNDO_MOVE;
			if(v>m){m=v;}
		}
	}}} WHILE_MOVES_LEFT;

    C:	return m;
}

where the macros implied by the capital writing are not implemented as macros but just a condensed way of indicating the code sections discussed elsewhere. The three nested loops of the move generator are symbolyzed by triple braces.

Soft Fail

The search uses a 'soft fail', i.e. if the score of the state does not fall within the (q,l) window it returns the upper or lower bound it found, not merely q or l. Although this difference would not affect the operation of the bare alpha-beta search, this will be beneficial once we start using a hash table for storing the results: we might encounter the same position later in the search with a different window, where the original score bound might produce a cutoff. So always report the sharpest bound on the move value you know, even if for this node itself it does not matter.

A consequence is that the recursive call always has to look at both q and m to determine the upper limit of the new search window (q>m ? -q : -m). I don't want to tinker with q (upping it when a better maximum is found), which would be slightly more efficient, because when we implement iterative deepening the next iteration has to be aware of the original q. I did not feel that this was worth throwing a new variable at, that you have to declare, initialize, etc...

Depth Control

The meaning of the depth control argument n is such that if n==1 all moves are generated, but none is actually tried (so that this position is an end leaf of the search tree). This might seem strange at first, why generate moves that we are not going to look at? Well, in the first place, even for moves that are not looked at, King-captures are still detected. So you might view calling D() with n==1 just a code-efficient way of detecting if the move generated on the previous ply was truly a legal move, rather than just 'pseudo-legal'. But when we implement quiescence search we will allow trying out some moves in this situation, and in that case the moves have to be generated in order to judge if they are worth looking at. An end-leaf then occurs only for n==1 if none of the moves was found to be worth considering.

If n==1 many, if not all moves will stay unconsidered. Such moves receive the static evaluation score EVAL(), under the assumption that they will not significantly change the position. (If they did, they should have been considered!) To this end m is initialized to EVAL(), rather than -infinity, if n==1. In practice, EVAL() is not really a function, but is immediately available as the argument e, because the evaluation is so rudimentary and simplistic that it can be differentially updated along with doing the moves in the path that led to the current poition.

Fail-High Cutoffs

The detection of the 'fail high' cutoff (if(m>=l)) might seem to occur in an unusual place: a more conventional location would be to put it inside the if(v>m) update of the maximum, where it would be executed less frequently. In addition it would also immediately discover the fail high, rather than after generation of one more move (which subsequently is not looked at). In its current place it also acts for intercepting King captures, which become evident during the move generation, without first considering replies to such a King capture (which would be really costly). Also, the static evaluation score can now already causes the fail high, before any moves have to be considered at all. For opimum efficiency the test for fail high could be made in all these places, the chosen solution sacrifices a little efficiency to compactness of the code.

Below the code that implements the recursive alpha-beta search is highlighted: (note that d ends up as n.) The move generation is symbolized by its outer loop, although in fact it is three nested loops.

/***************************************************************************/
/*                               micro-Max,                                */
/* A chess program smaller than 2KB (of non-blank source), by H.G. Muller  */
/***************************************************************************/
/* version 3.2 (2000 characters) features:                                 */
/* - recursive negamax search                                              */
/* - quiescence search with recaptures                                     */
/* - recapture extensions                                                  */
/* - (internal) iterative deepening                                        */
/* - best-move-first 'sorting'                                             */
/* - a hash table storing score and best move                              */
/* - full FIDE rules (expt minor ptomotion) and move-legality checking     */

#define F(I,S,N) for(I=S;I<N;I++)
#define W(A) while(A)
#define K(A,B) *(int*)(T+A+(B&8)+S*(B&7))
#define J(A) K(y+A,b[y])-K(x+A,u)-K(H+A,t)

#define U 16777224
struct _ {int K,V;char X,Y,D;} A[U];           /* hash table, 16M+8 entries*/

int V=112,M=136,S=128,I=8e3,C=799,Q,N,i;       /* V=0x70=rank mask, M=0x88 */

char O,K,L,
w[]={0,1,1,3,-1,3,5,9},                        /* relative piece values    */
o[]={-16,-15,-17,0,1,16,0,1,16,15,17,0,14,18,31,33,0, /* step-vector lists */
     7,-1,11,6,8,3,6,                          /* 1st dir. in o[] per piece*/
     6,3,5,7,4,5,3,6},                         /* initial piece setup      */
b[129],                                        /* board: half of 16x8+dummy*/
T[1035],                                       /* hash translation table   */

n[]=".?+nkbrq?*?NKBRQ";                        /* piece symbols on printout*/

D(k,q,l,e,J,Z,E,z,n)    /* recursive minimax search, k=moving side, n=depth*/
int k,q,l,e,J,Z,E,z,n;  /* (q,l)=window, e=current eval. score, E=e.p. sqr.*/
{                       /* e=score, z=prev.dest; J,Z=hashkeys; return score*/
 int j,r,m,v,d,h,i=9,F,G;
 char t,p,u,x,y,X,Y,H,B;
 struct _*a=A;
                                               /* lookup pos. in hash table*/
 j=(k*E^J)&U-9;                                /* try 8 consec. locations  */
 W((h=A[++j].K)&&h-Z&&--i);                    /* first empty or match     */
 a+=i?j:0;                                     /* dummy A[0] if miss & full*/
 if(a->K)                                      /* hit: pos. is in hash tab */
 {d=a->D;v=a->V;X=a->X;                        /* examine stored data      */
  if(d>=n)                                     /* if depth sufficient:     */
  {if(v>=l|X&S&&v<=q|X&8)return v;             /* use if window compatible */
   d=n-1;                                      /* or use as iter. start    */
  }X&=~M;Y=a->Y;                               /*      with best-move hint */
  Y=d?Y:0;                                     /* don't try best at d=0    */
 }else d=X=Y=0;                                /* start iter., no best yet */
 N++;                                          /* node count (for timing)  */
 W(d++<n|z==8&N<1e7&d<98)                      /* iterative deepening loop */
 {x=B=X;                                       /* start scan at prev. best */
  Y|=8&Y>>4;                                   /* request try noncastl. 1st*/
  m=d>1?-I:e;                                  /* unconsidered:static eval */
  do{u=b[x];                                   /* scan board looking for   */
   if(u&k)                                     /*  own piece (inefficient!)*/
   {r=p=u&7;                                   /* p = piece type (set r>0) */
    j=o[p+16];                                 /* first step vector f.piece*/
    W(r=p>2&r<0?-r:-o[++j])                    /* loop over directions o[] */
    {A:                                        /* resume normal after best */
     y=x;F=G=S;                                /* (x,y)=move, (F,G)=castl.R*/
     do{H=y+=r;                                /* y traverses ray          */
      if(Y&8)H=y=Y&~M;                         /* sneak in prev. best move */
      if(y&M)break;                            /* board edge hit           */
      if(p<3&y==E)H=y^16;                      /* shift capt.sqr. H if e.p.*/
      t=b[H];if(t&k|p<3&!(r&7)!=!t)break;      /* capt. own, bad pawn mode */
      i=99*w[t&7];                             /* value of capt. piece t   */
      if(i<0||E-S&&b[E]&&y-E<2&E-y<2)m=I;      /* K capt. or bad castling  */
      if(m>=l)goto C;                          /* abort on fail high       */
    
      if(h=d-(y!=z))                           /* remaining depth(-recapt.)*/
      {v=p<6?b[x+8]-b[y+8]:0;                  /* center positional pts.   */
       b[G]=b[H]=b[x]=0;b[y]=u&31;             /* do move, strip virgin-bit*/
       if(!(G&M)){b[F]=k+6;v+=30;}             /* castling: put R & score  */
       if(p<3)                                 /* pawns:                   */
       {v-=9*(((x-2)&M||b[x-2]!=u)+            /* structure, undefended    */
              ((x+2)&M||b[x+2]!=u)-1);         /*        squares plus bias */
        if(y+r+1&S){b[y]|=7;i+=C;}             /* promote p to Q, add score*/
       }
       v=-D(24-k,-l-(l>e),m>q?-m:-q,-e-v-i,    /* recursive eval. of reply */
            J+J(0),Z+J(8)+G-S,F,y,h);          /* J,Z: hash keys           */
       v-=v>e;                                 /* delayed-gain penalty     */
       if(z==9)                                /* called as move-legality  */
       {if(v!=-I&x==K&y==L)                    /*   checker: if move found */
        {Q=-e-i;O=F;return l;}                 /*   & not in check, signal */
        v=m;                                   /* (prevent fail-lows on    */
       }                                       /*   K-capt. replies)       */
       b[G]=k+38;b[F]=b[y]=0;b[x]=u;b[H]=t;    /* undo move,G can be dummy */
       if(Y&8){m=v;Y&=~8;goto A;}              /* best=1st done,redo normal*/
       if(v>m){m=v;X=x;Y=y|S&G;}               /* update max, mark with S  */
      }                                        /*          if non castling */
      t+=p<5;                                  /* fake capt. for nonsliding*/
      if(p<3&6*k+(y&V)==S                      /* pawn on 3rd/6th, or      */
          ||(u&~24)==36&j==7&&                 /* virgin K moving sideways,*/
          G&M&&b[G=(x|7)-(r>>1&7)]&32          /* 1st, virgin R in corner G*/
          &&!(b[G^1]|b[G^2])                   /* 2 empty sqrs. next to R  */
      ){F=y;t--;}                              /* unfake capt., enable e.p.*/
     }W(!t);                                   /* if not capt. continue ray*/
  }}}W((x=x+9&~M)-B);                          /* next sqr. of board, wrap */
C:if(m>I/4|m<-I/4)d=99;                        /* mate is indep. of depth  */
  m=m+I?m:-D(24-k,-I,I,0,J,K,S,z,1)/2;         /* best loses K: (stale)mate*/
  if(!a->K|(a->X&M)!=M|a->D<=d)                /* if new/better type/depth:*/
  {a->K=Z;a->V=m;a->D=d;A->K=0;                /* store in hash,dummy stays*/
   a->X=X|8*(m>q)|S*(m<l);a->Y=Y;              /* empty, type (limit/exact)*/
  }                                            /*    encoded in X S,8 bits */
/*if(z==8)printf("%2d ply, %9d searched, %6d by (%2x,%2x)\n",d-1,N,m,X,Y&0x77);*/
 }
 if(z&8){K=X;L=Y&~M;}
 return m;                                     
}

main()
{
 int j,k=8,*p,c[9];

 F(i,0,8)
 {b[i]=(b[i+V]=o[i+24]+40)+8;b[i+16]=18;b[i+96]=9;   /* initial board setup*/
  F(j,0,8)b[16*j+i+8]=(i-4)*(i-4)+(j-3.5)*(j-3.5);   /* center-pts table   */
 }                                                   /*(in unused half b[])*/
 F(i,M,1035)T[i]=random()>>9;

 W(1)                                                /* play loop          */
 {F(i,0,121)printf(" %c",i&8&&(i+=7)?10:n[b[i]&15]); /* print board        */
  p=c;W((*p++=getchar())>10);                        /* read input line    */
  N=0;
  if(*c-10){K=c[0]-16*c[1]+C;L=c[2]-16*c[3]+C;}else  /* parse entered move */
   D(k,-I,I,Q,1,1,O,8,0);                            /* or think up one    */
  F(i,0,U)A[i].K=0;                                  /* clear hash table   */
  if(D(k,-I,I,Q,1,1,O,9,2)==I)k^=24;                 /* check legality & do*/
 }
}
Previous   Next   Version 4