Gather data

src/main/java/puzzle/Main.java

@@ -340,7 +340,7 @@ public class Main {
    static PuzzleResult attempt(Rng rng, Dict dict, Opts opts) {
       TOTAL_ATTEMPTS.incrementAndGet();
       var swe    = new SwedishGenerator();
-      var mask   = swe.generateMask(rng, dict.lenCounts(), opts.pop, opts.gens);
+      var mask   = swe.generateMask(rng, opts.pop, opts.gens);
       var filled = new CSP(rng).fillMask(mask, dict.index(), 200, opts.fillTimeout);
       
       TOTAL_NODES.addAndGet(filled.stats().nodes);

src/main/java/puzzle/SwedishGenerator.java

@@ -281,16 +281,14 @@ public record SwedishGenerator() {
    
    public static record Dict(
            DictEntry[] index,
-           int[] lenCounts, int length) {
+           int length) {
       
       static final Gson GSON = new Gson();
       public Dict(Lemma[] wordz) {
-         var lenCounts = new int[MAX_WORD_LENGTH_PLUS_ONE];
          var index     = new DictEntry[MAX_WORD_LENGTH_PLUS_ONE];
          Arrays.setAll(index, i -> new DictEntry(i));
          for (var lemma : wordz) {
             var L = lemma.word.length;
-            lenCounts[L]++;
             
             var entry = index[L];
             var idx   = entry.words.size();
@@ -302,13 +300,13 @@ public record SwedishGenerator() {
                entry.pos[i][letter].add(idx);
             }
          }
-         for (var len : lenCounts) {
+         for (int i = MIN_LEN; i < index.length; i++) {
+            var len = index[i].words.size();
             if (len <= 0) {
-               System.out.println("No words for length " + len);
-               //throw new RuntimeException("Invalid word length: " + len);
+               throw new RuntimeException("No words for length " + i);
             }
          }
-         this(index, lenCounts, Arrays.stream(lenCounts).sum());
+         this(index, Arrays.stream(index).mapToInt(i -> i.words.size()).sum());
       }
       static Dict loadDict(String wordsPath) {
          String raw;
@@ -355,7 +353,6 @@ public record SwedishGenerator() {
          
          return new Dict(map.toArray(Lemma[]::new));
       }
-      public int dictLength() { return Arrays.stream(lenCounts).sum(); }
    }
    
    static int intersectSorted(int[] a, int aLen, int[] b, int bLen, int[] out) {
@@ -424,7 +421,7 @@ public record SwedishGenerator() {
       return false;
    }
    
-   long maskFitness(Grid grid, int[] lenCounts) {
+   long maskFitness(Grid grid) {
       final long[] penalty = { 0 };
       
       var clueCount = grid.clueCount();
@@ -460,9 +457,9 @@ public record SwedishGenerator() {
          hasSlots[0] = true;
          if (n < MIN_LEN) {
             penalty[0] += 8000;
-         } else if (lenCounts[n] <= 0) {
+         } /*else if (lenCounts[n] <= 0) {
             penalty[0] += 12000;
-         }
+         }*/
          var horiz = Slot.horiz(d) ? covH : covV;
          for (var i = 0; i < n; i++) horiz[Slot.offset(packedPos, i)] += 1;
       }
@@ -488,9 +485,9 @@ public record SwedishGenerator() {
          hasSlots[0] = true;
          if (n < MIN_LEN) {
             penalty[0] += 8000;
-         } else if (lenCounts[n] <= 0) {
+         } /*else if (lenCounts[n] <= 0) {
             penalty[0] += 12000;
-         }
+         }*/
          var horiz = Slot.horiz(d) ? covH : covV;
          for (var i = 0; i < n; i++) horiz[Slot.offset(packedPos, i)] += 1;
       }
@@ -595,14 +592,14 @@ public record SwedishGenerator() {
       return out;
    }
    
-   Grid hillclimb(Rng rng, Grid start, int[] lenCounts, int limit) {
+   Grid hillclimb(Rng rng, Grid start, int limit) {
       var best  = start;
-      var bestF = maskFitness(best, lenCounts);
+      var bestF = maskFitness(best);
       var fails = 0;
       
       while (fails < limit) {
          var cand = mutate(rng, best);
-         var f    = maskFitness(cand, lenCounts);
+         var f    = maskFitness(cand);
          if (f < bestF) {
             best  = cand;
             bestF = f;
@@ -614,21 +611,21 @@ public record SwedishGenerator() {
       return best;
    }
    
-   public Grid generateMask(Rng rng, int[] lenCounts, int popSize, int gens) {
+   public Grid generateMask(Rng rng, int popSize, int gens) {
       class GridAndFit {
          
          Grid grid;
          Long fite;
          GridAndFit(Grid grid) { this.grid = grid; }
          long fit() {
-            if (fite == null) this.fite = maskFitness(grid, lenCounts);
+            if (fite == null) this.fite = maskFitness(grid);
             return this.fite;
          }
       }
       if (Main.VERBOSE) System.out.println("generateMask init pop: " + popSize);
       var pop = new ArrayList<GridAndFit>();
       for (var i = 0; i < popSize; i++) {
-         pop.add(new GridAndFit(hillclimb(rng, randomMask(rng), lenCounts, 180)));
+         pop.add(new GridAndFit(hillclimb(rng, randomMask(rng), 180)));
       }
       
       for (var gen = 0; gen < gens; gen++) {
@@ -640,7 +637,7 @@ public record SwedishGenerator() {
             var p1    = pop.get(rng.randint(0, pop.size() - 1));
             var p2    = pop.get(rng.randint(0, pop.size() - 1));
             var child = crossover(rng, p1.grid, p2.grid);
-            children.add(new GridAndFit(hillclimb(rng, child, lenCounts, 70)));
+            children.add(new GridAndFit(hillclimb(rng, child, 70)));
          }
          
          pop.addAll(children);