Change Evaluator and features to minimize singular matrices

src/CheckersAI.java

@@ -52,7 +52,7 @@ public class CheckersAI{
         double v = Double.NEGATIVE_INFINITY;
         double check;
         Move max = null;
-        // System.out.println(s.actions().size());
+        System.out.println(s.actions().size());
         for(Move a: s.actions()){
             check = minValue(s.result(a), alpha, beta, depth + 1, a.isJump(), min_ply);
             if(check > v){

src/CheckersGameState3.java

@@ -374,13 +374,16 @@ public class CheckersGameState3 implements CheckersGameState{
    }
 
    /* computes feature vector:
-      piece-ratio, loners, safes, pawns,
-      moveable pawns, aggregate distance,
-      kings, moveable kings, promotion line
-      opening
+      [piece-ratio,
+      loners,
+      safes,
+      pawns,
+      moveable pawns,
+      aggregate distance,
+      promotion line opening]
    */
    public double[] getFeatures(int player){
-     double[] features = new double[9];
+     double[] features = new double[7];
      double total = 0.0;
      double mypieces = 0.0;
      for(int i = 0; i<this.board.length; i++){
@@ -396,14 +399,14 @@ public class CheckersGameState3 implements CheckersGameState{
               features[5] += getDistance(i);       //aggregate distance
            }
            else if(this.board[i] == player+2){     //king
-             features[6]+=1.0;
-             if(king_can_move(i)) features[7] += 1.0;
+             features[3]+=2.0;
+             if(king_can_move(i)) features[4] += 2.0;
          }
        }
      }
    }
      features[0] = mypieces/total;
-     features[8] = promotionLineOpenings(player);
+     features[6] = promotionLineOpenings(player);
      return features;
    }
 

src/Learn.java

@@ -4,9 +4,9 @@ public class Learn{
 
     public static void main(String[] args){
 
-        final int num_games = 1;
-        LearningEvaluator le =  new LearningEvaluator("weights/alpha.csv", .1);
-        BaseEvaluator be = new BaseEvaluator("weights/beta.csv");
+        final int num_games = 15;
+        LearningEvaluator le =  new LearningEvaluator("../src/weights/alpha.csv", .1);
+        BaseEvaluator be = new BaseEvaluator("../src/weights/beta.csv");
         CheckersAI alpha = new CheckersAI(le, 1);
         CheckersAI beta = new CheckersAI(be, 2);
         CheckersGameState s;
@@ -25,7 +25,7 @@ public class Learn{
             }
             if(played == 10){
                 if(won >= 7){ // if alpha wins 7 of every ten games, make beta use alpha's new evaluator
-                    le.commitWeights("weights/beta.csv");
+                    le.commitWeights("../src/weights/beta.csv");
                     be.refreshWeights();
                 }
                 played = 0;
@@ -58,12 +58,14 @@ public class Learn{
             moves++;
         }
         while(!current.isTerminal() && moves <= 50){
+            System.out.println("alphas moves:");
+            System.out.println(current.actions());
             Move next = alpha.minimax(current, 7); // get alpha's move
             le.addData(current.getFeatures(alpha.getPlayer()), next.getValue()); // add this moves data to the data set (the value of the state is stored in the move. there is probably a better way to do this)
             current = current.result(next); // make the move
             current.printState();
             moves++;
-            if(current.isTerminal()){ // if alpha won, then brea
+            if(current.isTerminal()){ // if alpha won, then break
                 break;
             }
             current = current.result(beta.minimax(current, 7)); // beta's move
@@ -74,5 +76,3 @@ public class Learn{
         return current.winner();
     }
 }
-
-

src/LearningEvaluator.java

@@ -1,12 +1,13 @@
 import java.util.ArrayList;
 import org.apache.commons.math3.stat.regression.OLSMultipleLinearRegression;
+import org.apache.commons.math3.linear.SingularMatrixException;
 import java.util.Arrays;
 
 public class LearningEvaluator extends BaseEvaluator{
 
     ArrayList<double[]> params;
     ArrayList<Double> values;
-    // need to download jar and set classpath to import and run 
+    // need to download jar and set classpath to import and run
     OLSMultipleLinearRegression reg; // performs linear regression (ordinary least squares)
     double alpha; // learning parameter, higher alpha means weights are closer to the regression output
                   // alpha of 1 is directly setting weights to be regression weights
@@ -39,33 +40,38 @@ public class LearningEvaluator extends BaseEvaluator{
         // get a lot of singular matrices
         // we could do samuel's method or come up with another function to modify the coefficients
         int curr_in = 0;
-        while(params.size() - curr_in > 10){ // need to do regression with data sets of size 10, so each iteration of loop uses 10 lines of data
-            double[] vals = new double [10]; //converting arraylist to array
+        int data_sz = 8;   // need to do regression with data sets of size 10, so each iteration of loop uses 10 lines of data
+        while(params.size() - curr_in > data_sz){
+            double[] vals = new double [data_sz]; //converting arraylist to array
             System.out.println("printing values");
             int j = 0;
-            for(int i = curr_in; i < curr_in + 10; i++){
+            for(int i = curr_in; i < curr_in + data_sz; i++){
                 vals[j] = values.get(i);
                 System.out.println(values.get(i));
                 j++;
             }
             System.out.println(vals);
             System.out.println("printing params");
-            double[][] pars = new double[10][]; //converting 2d arraylist to array
+            double[][] pars = new double[data_sz][]; //converting 2d arraylist to array
             j=0;
-            for(int i=curr_in; i < curr_in + 10; i++){
+            for(int i=curr_in; i < curr_in + data_sz; i++){
                 pars[j] = params.get(i);
                 System.out.println(Arrays.toString(params.get(i)));
                 j++;
             }
             System.out.println(pars);
             reg.newSampleData(vals, pars); //add data
             reg.setNoIntercept(true);
-            double[] new_weights = reg.estimateRegressionParameters(); //get parameters
-            for(int i = 0; i < this.weights.length; i++){
-                this.weights[i] = this.weights[i] + alpha * (new_weights[i] - this.weights[i]);
+            try {
+              double[] new_weights = reg.estimateRegressionParameters(); //get parameters
+              for(int i = 0; i < this.weights.length; i++){
+                  this.weights[i] = this.weights[i] + alpha * (new_weights[i] - this.weights[i]);
+              }
+              commitWeights(this.file);
+            } catch(SingularMatrixException e) {
+              System.out.println("Matrix was singular, not updating weights");
             }
-            commitWeights(this.file);
-            curr_in += 10;
+            curr_in += data_sz;
         }
 
         values.clear();

src/weights/alpha.csv

@@ -1 +1,30 @@
-50, 10, 10, 5, 5, 5, 30, 15, 10
+50, 10, 10, 5, 5, 5
+-9.3685899134567718E17, 7.794026190160416E16, -6.049540124358089E14, -6.932782101834882E14, 7.463767693764232E16, 1.1512441608154629
+-2.8738797198358354E18, 5.3535252570681231E18, 3.3336822929004636E16, 3.433162512891778E16, -4.0433107676171347E17, 7.819856042285374E15
+1.817671195817515E32, -5.438334910319976E30, 3.7970448614125307E30, 5.999603288154183E30, 4.2657208866270885E30, -3.420682188700641E30
+1.7206054178773225E32, -1.2372315866627587E31, 4.135155496534389E30, 4.954032483591002E30, 5.800958050966378E30, -3.0930635559914327E30
+2.2189863047375575E32, -1.1200084316563596E31, 2.795880090367318E30, 5.808933256464445E30, 2.5411383173703824E30, -3.0954239523221725E30
+1.4570594680388968E32, -8.866159458937068E30, -4.184354587177827E45, 4.184354587177812E45, -4.184354587177801E45, -9.203700236115977E29
+4.828745193602742E46, -2.117283421111761E47, -3.431170761485821E45, 3.849606220203588E45, 1.8411160183577473E45, -2.510612752306685E44
+2.6886269228363255E46, -1.9092532105319082E47, -2.5877150925926487E45, 5.4585742201173775E45, 1.9360832367781485E45, -4.176681793472993E44
+9.68056166753188E47, -1.822791958582696E47, 3.2799103164625757E46, 6.982451188800286E46, -1.2759822630112968E46, -2.1251547085596302E46
+6.607747772113344E47, -1.5775834620045355E47, 1.9082744084210428E46, 4.943658822412271E46, -1.0928504000796732E46, -1.268366597892222E46
+-6.640413286411508E63, 6.191527539777473E61, 8.048985801710929E61, 5.262798408810957E62, -1.8574582619332455E62, -6.12794061383161E45
+-9.975387123273426E76, 1.9950774246545522E77, -4.750184344415589E75, 5.842609164879715E62, -1.0648576764909763E62, -2.3269824336997295E61
+-8.977848410946235E76, 1.795569682189097E77, -4.275165909974025E75, 6.109277826946604E62, -8.213410456570652E61, -2.5089001685635713E61
+-1.395983161991158E77, 1.6160127139701874E77, 9.721046181162213E75, 5.25986567097873E62, -9.42642845969764E61, -4.2967917201625334E61
+-2.783659649264689E77, 7.993733678169044E77, 2.3936964084631586E76, -1.9442092362320297E75, -9.098576084558526E76, 1.3443043569096783E76
+1.1278410857435816E78, 7.301806896008975E77, 2.500785282308765E76, 4.311516318358936E75, -1.0440491512512438E77, 8.264007014739985E75
+4.3825255215879347E77, 6.46912413373827E77, 3.3368415505215776E77, 5.44692780818499E76, -8.380612099682645E76, -8.224821683358007E75
+6.536493354945876E76, 5.90016774220315E77, 3.033792609868331E77, 1.9930898294085956E77, -1.0093881611537739E77, -1.996258945193199E76
+-6.484760054376507E77, 5.693556733979261E77, 3.52941302481836E77, 4.0495478599741917E77, -1.2805312734164537E77, -4.6965237247001146E76
+-2.9395502408946725E78, 5.165544365100268E77, 9.744982653143925E77, -1.2426575673168948E77, 4.5261748871156896E77, -6.567082327835312E76
+-8.200718081043308E78, 5.423910896161437E77, 1.2845513616993203E78, -1.375451023113857E77, 4.2791673573848214E77, -3.2178738680038086E76
+8.373699459108173E78, 4.596571219901508E77, 1.11227296311433E78, -1.3219280792060426E77, 9.137064349203897E76, -2.2307875336690484E76
+5.722730436152141E78, 3.266170201871765E77, 1.1356081942713786E78, -2.924947663746416E77, 3.799050720906769E77, -7.641060717501331E75
+3.8630225969823536E78, 2.8181958605463597E77, 1.709325061397922E78, 7.731448514518302E76, 3.2094710923023195E77, -7.463884758456114E76
+3.621525694225168E78, 2.577641971114625E77, 1.5178164811966045E78, 2.33441167062367E77, 2.810616951734669E77, -7.504402810380263E76
+-3.9246295090853187E80, 2.5992773231027646E77, -1.3579310902003569E93, -5.248570840331341E93, 5.248570840331358E93, -3.2362800212151406E76
+-2.1067562907318428E94, -4.468385728885715E92, -2.5123998504920498E93, -9.180443867939546E93, 6.345104944836925E93, 7.980054086800604E92
+-7.81284835659837E94, -1.6166995865070447E93, -5.128336274216761E93, -1.9307596009148744E94, 9.615561351884508E93, 2.7628625498248885E93
+-2.258434065448233E95, -4.623187033028576E93, -1.148760484999136E94, -4.5123076330934E94, 1.837401892192431E94, 7.685804031253954E93

src/weights/beta.csv

@@ -1 +1 @@
-50, 10, 10, 5, 5, 5, 30, 15, 10
+50, 10, 10, 5, 5, 5