Man-Machine game-generation of sub-game (III)

We can create a walking generator based on the rules of playing chess. The main logic is that if you have a pair of chess strings, you do not need to compress the other party, that is to say, you do not need to attach the opponent's pawns. In other cases

When the opponent's pawns fall into some special circumstances, when the other party can pick up the other party's pawns, the other party can go down to a place where there is no gas. In addition, the other party can take sub-games and win first, no competition is possible. So the algorithm generator is relatively non-

It is often simple.

You can use the algorithm described in the previous section to determine whether your opponent has a certain chess string.

Int cmovegenerator: createpossiblemove (byte position [grid_num] [grid_num], int nply, int nside) {m_nmovecount = 0; byte antiside = (nside + 1) % 2; cleanglobal (); setgo (position. For (INT I = 0; I <grid_num; I ++) for (Int J = 0; j <grid_num; j ++) {If (go [I] [J] = nside & g_gozi [I] [J] = 0) {str_lib (I, j, go [I] [J]); If (goqi = 1) {for (int K = 0; k <grid_num; k ++) for (INT W = 0; W <grid_num; W ++) {If (gokong [k] [W] = 1) {addmove (K, W, nply );}}}}} 64 // under normal circumstances, find the space around the enemy's pawns and compress them with gas for (INT I = 0; I <grid_num; I ++) for (Int J = 0; j <grid_num; j ++) {If (go [I] [J] = antiside) {if (I> 0 & go [I-1] [J] = nostone) {addmove (I-1, J, nply );} if (I <grid_num-1 & go [I + 1] [J] = nostone) {addmove (I + 1, J, nply );} if (j> 0 & go [I] [J-1] = nostone) {addmove (I, j-1, nply );} if (j <grid_num-1 & go [I] [J + 1] = nostone) {addmove (I, j + 1, nply );}}} return m_nmovecount ;}

This algorithm can be optimized to facilitate subsequent search engine pruning. Set a score for the walk method. If you can raise the score, set this step to 30 + number of raisins. The number of chess pieces that can be played is 20 +. The number of chess pieces that can grow up or down is 10 +. Other temporary designs are 0. You can use a priority queue with a rated length to retain the best scores. Or sort it if necessary.

Search with a higher score is given priority, which greatly improves the efficiency of search algorithm pruning.

This section is a code inspired by the method.

The pseudo code is as follows:

for(int i=0;i<Grid_Num;i++)for(int j=0;j<Grid_Num;j++){    cleanupGlobal();    if(go[i][j]==NOSTONE)    {    bool isVilid=false;    if(i>0&&go[i-1][j]==antiSide)    {       go[i][j]=nSide;       if(g_gozi[i-1][j]==0)       {          isVilid=true;          str_lib(i-1,j,antiSide);          switch(goqi){case 0:case 1:....}       }    }     if(i<Grid_Num-1&&go[i+1][j]==antiSide)     {       go[i][j]=nSide;       if(g_gozi[i-1][j]==0)       {          isValid=true;          str_lib(i-1,j,antiSide);          switch(goqi){case 0:case 1:....}       }    }...         if(isValid)         {          sumScore(i,j,nSide);          addMove(i,j,nSide);          }44     }}std::sort();

This algorithm is a method-based inspiration. Its learning name is static inspiration, rather than historical inspiration. Historical inspiration does not require the knowledge of chess and board games to be a dynamic inspiration, but the method-based inspiration is associated with the knowledge of board games.

Man-Machine game-generation of sub-game (III)