Phase F: register TABU_SEARCH and MIN_CONFLICTS, results

README.md

@@ -29,7 +29,10 @@ See the [releases page](https://github.com/bb4/bb4-optimization/releases) or [Ma
   **global simulated annealing** (same multistart pattern with SA locals via `GlobalSimulatedAnnealingConfig`), 
   simulated annealing (`SimulatedAnnealingConfig`: adaptive acceptance by default, budget-safe best-so-far), 
   **genetic / concurrent genetic** (`GeneticSearchConfig`: elitism, cull/crossover, neighbor modes, budget-safe evolve), 
-  **state-space search**, and brute force
+  **state-space search**, and brute force,
+  and for constraint problems **min-conflicts** (`MinConflictsConfig`; fixes a conflicted position by its best move, using `Optimizee.conflictedPositions`)
+  and **tabu search** (`TabuSearchConfig`; takes the best move that is not tabu). `RestartingStrategy` (`RestartingConfig`) re-runs any strategy with Luby budgets.
+  See [docs/solve-rate-results.md](docs/solve-rate-results.md) for how often each strategy solves the constraint benchmark problems.
 - **`SimulatedAnnealingConfig` / `SimulatedAnnealingStepMode`** — cooling schedule and proposal radius; 
   default `AdaptiveAcceptance` scales steps from Metropolis acceptance rate on continuous problems (discrete spaces automatically use Fixed radius); proposal radius is clamped to 1.5
 - **`GlobalSimulatedAnnealingConfig`** — global sample budget + multistart SA from diverse / top-fitness seeds (prefer over single-chain SA on multimodal discrete problems; 

docs/new-strategies-plan.md

@@ -216,3 +216,17 @@ These have defaults, so the plan can proceed without answers:
    bb4-puzzles later if one is needed.
 3. Is a 20 seed solve rate section acceptable for the length of a full benchmark run (about 10 minutes now)? Default:
    yes, with the section as its own `--solve-rate` run if it makes the full run too long.
+
+## Outcome (phases A to F, bb4-optimization)
+
+Measured with the 20 seed solve rate in [solve-rate-results.md](solve-rate-results.md):
+
+- **MIN_CONFLICTS** solves all 80 runs of the four problems, including coloring-200, which no earlier strategy solved,
+  and queens-400 in about 0.35 s against 0.76 s for annealing. It joins the enum. It needs `conflictedPositions` to be at
+  its best.
+- **TABU_SEARCH** solves all-interval-14 20/20 (the best earlier strategy: 10/20) and coloring-200 5/20 (earlier: 0/20),
+  but only 5/20 on queens-100 where annealing solves everything. It meets the rule on two problems, so it joins the enum,
+  with the note that it is not the first choice for queens.
+- **RestartingStrategy** stays off the enum: no combination was shown to beat the strategies above, and it can be built in
+  code for any strategy. It is available for the puzzles to try.
+- Not yet done: try min-conflicts and tabu on the Red puzzle in bb4-puzzles, and add planted 3-SAT and a Costas array.

docs/solve-rate-results.md

@@ -8,33 +8,41 @@ Java 21.0.12.1, 2 processors, 20 seeds, 2000 ms or 5000000 evaluations per run
 
 | Problem | Strategy | Solved | Median ms (solved) | Median evals (solved) | Median broken (unsolved) |
 | --- | --- | ---: | ---: | ---: | ---: |
-| queens-100 | SIMULATED_ANNEALING | 20/20 | 60 | 17,517 | - |
-| queens-100 | GLOBAL_SIMULATED_ANNEALING | 20/20 | 62 | 18,800 | - |
+| queens-100 | SIMULATED_ANNEALING | 20/20 | 53 | 17,517 | - |
+| queens-100 | GLOBAL_SIMULATED_ANNEALING | 20/20 | 61 | 18,800 | - |
 | queens-100 | HILL_CLIMBING | 0/20 | - | - | 39.0 |
 | queens-100 | GLOBAL_HILL_CLIMBING | 0/20 | - | - | 36.0 |
 | queens-100 | GENETIC_SEARCH | 0/20 | - | - | 2.0 |
 | queens-100 | CONCURRENT_GENETIC_SEARCH | 0/20 | - | - | 2.0 |
 | queens-100 | ITERATED_LOCAL_SEARCH | 0/20 | - | - | 20.0 |
 | queens-100 | VARIABLE_NEIGHBORHOOD_SEARCH | 0/20 | - | - | 20.0 |
-| queens-400 | SIMULATED_ANNEALING | 20/20 | 818 | 97,017 | - |
-| queens-400 | GLOBAL_SIMULATED_ANNEALING | 20/20 | 993 | 94,068 | - |
+| queens-100 | TABU_SEARCH | 5/20 | 349 | 104,610 | 1.0 |
+| queens-100 | MIN_CONFLICTS | 20/20 | 8 | 3,178 | - |
+| queens-400 | SIMULATED_ANNEALING | 20/20 | 761 | 97,017 | - |
+| queens-400 | GLOBAL_SIMULATED_ANNEALING | 20/20 | 894 | 94,068 | - |
 | queens-400 | HILL_CLIMBING | 0/20 | - | - | 206.0 |
 | queens-400 | GLOBAL_HILL_CLIMBING | 0/20 | - | - | 199.0 |
-| queens-400 | GENETIC_SEARCH | 0/20 | - | - | 155.5 |
-| queens-400 | CONCURRENT_GENETIC_SEARCH | 0/20 | - | - | 155.0 |
+| queens-400 | GENETIC_SEARCH | 0/20 | - | - | 154.5 |
+| queens-400 | CONCURRENT_GENETIC_SEARCH | 0/20 | - | - | 154.0 |
 | queens-400 | ITERATED_LOCAL_SEARCH | 0/20 | - | - | 155.0 |
 | queens-400 | VARIABLE_NEIGHBORHOOD_SEARCH | 0/20 | - | - | 153.0 |
+| queens-400 | TABU_SEARCH | 0/20 | - | - | 12.0 |
+| queens-400 | MIN_CONFLICTS | 20/20 | 348 | 40,933 | - |
 | coloring-200 | SIMULATED_ANNEALING | 0/20 | - | - | 141.0 |
 | coloring-200 | HILL_CLIMBING | 0/20 | - | - | 107.5 |
-| coloring-200 | GENETIC_SEARCH | 0/20 | - | - | 116.0 |
-| coloring-200 | CONCURRENT_GENETIC_SEARCH | 0/20 | - | - | 115.0 |
+| coloring-200 | GENETIC_SEARCH | 0/20 | - | - | 106.5 |
+| coloring-200 | CONCURRENT_GENETIC_SEARCH | 0/20 | - | - | 105.5 |
 | coloring-200 | ITERATED_LOCAL_SEARCH | 0/20 | - | - | 131.0 |
 | coloring-200 | VARIABLE_NEIGHBORHOOD_SEARCH | 0/20 | - | - | 132.0 |
+| coloring-200 | TABU_SEARCH | 5/20 | 26 | 50,278 | 2.0 |
+| coloring-200 | MIN_CONFLICTS | 20/20 | 58 | 9,214 | - |
 | all-interval-14 | SIMULATED_ANNEALING | 1/20 | 2 | 1,771 | 1.0 |
-| all-interval-14 | GLOBAL_SIMULATED_ANNEALING | 1/20 | 9 | 9,764 | 1.0 |
+| all-interval-14 | GLOBAL_SIMULATED_ANNEALING | 1/20 | 16 | 9,764 | 1.0 |
 | all-interval-14 | HILL_CLIMBING | 0/20 | - | - | 2.0 |
 | all-interval-14 | GLOBAL_HILL_CLIMBING | 0/20 | - | - | 1.0 |
-| all-interval-14 | GENETIC_SEARCH | 10/20 | 205 | 58,450 | 1.0 |
-| all-interval-14 | CONCURRENT_GENETIC_SEARCH | 10/20 | 219 | 58,451 | 1.0 |
+| all-interval-14 | GENETIC_SEARCH | 10/20 | 252 | 58,450 | 1.0 |
+| all-interval-14 | CONCURRENT_GENETIC_SEARCH | 10/20 | 240 | 58,451 | 1.0 |
 | all-interval-14 | ITERATED_LOCAL_SEARCH | 5/20 | 3 | 2,436 | 1.0 |
-| all-interval-14 | VARIABLE_NEIGHBORHOOD_SEARCH | 4/20 | 17 | 9,609 | 1.0 |
+| all-interval-14 | VARIABLE_NEIGHBORHOOD_SEARCH | 4/20 | 13 | 9,609 | 1.0 |
+| all-interval-14 | TABU_SEARCH | 20/20 | 46 | 38,561 | - |
+| all-interval-14 | MIN_CONFLICTS | 20/20 | 9 | 7,762 | - |

scala-source/com/barrybecker4/optimization/strategy/OptimizationStrategyType.scala

@@ -7,6 +7,8 @@ import com.barrybecker4.optimization.strategy.annealing.{GlobalSimulatedAnnealin
 import com.barrybecker4.optimization.strategy.genetic.{ConcurrentGeneticSearchStrategy, GeneticSearchStrategy}
 import com.barrybecker4.optimization.strategy.hillclimbing.{GlobalHillClimbingStrategy, HillClimbingStrategy}
 import com.barrybecker4.optimization.strategy.iteratedlocal.{IteratedLocalSearchConfig, IteratedLocalSearchStrategy}
+import com.barrybecker4.optimization.strategy.minconflicts.{MinConflictsConfig, MinConflictsStrategy}
+import com.barrybecker4.optimization.strategy.tabu.{TabuSearchConfig, TabuSearchStrategy}
 import com.barrybecker4.optimization.strategy.particleswarm.ParticleSwarmStrategy
 import com.barrybecker4.optimization.strategy.statespace.{StateSpaceSearchConfig, StateSpaceSearchStrategy}
 
@@ -33,7 +35,9 @@ enum OptimizationStrategyType {
        DIFFERENTIAL_EVOLUTION,
        PARTICLE_SWARM,
        ITERATED_LOCAL_SEARCH,
-       VARIABLE_NEIGHBORHOOD_SEARCH
+       VARIABLE_NEIGHBORHOOD_SEARCH,
+       TABU_SEARCH,
+       MIN_CONFLICTS
 
   def getDescription: String = this match
     case GLOBAL_SAMPLING =>
@@ -64,6 +68,11 @@ enum OptimizationStrategyType {
       "Iterated local search (ILS) for discrete problems such as permutations."
     case VARIABLE_NEIGHBORHOOD_SEARCH =>
       "Variable neighborhood search (VNS) for discrete problems such as permutations."
+    case TABU_SEARCH =>
+      "Tabu search for discrete problems: takes the best allowed move even when it is worse, and forbids undoing recent moves."
+    case MIN_CONFLICTS =>
+      "Min-conflicts for constraint problems: repeatedly fixes a conflicted position by its best move. " +
+        "Needs an optimizee that reports conflictedPositions to work well."
 
   /**
     * True when this strategy can run on `optimizee` with `params`.
@@ -81,6 +90,10 @@ enum OptimizationStrategyType {
         !optimizee.evaluateByComparison && params.isInstanceOf[NumericParameterArray]
       case HILL_CLIMBING | GLOBAL_HILL_CLIMBING =>
         locallySearchable(params)
+      case TABU_SEARCH =>
+        !optimizee.evaluateByComparison
+      case MIN_CONFLICTS =>
+        !optimizee.evaluateByComparison && params.supportsLocalMoves
       case _ =>
         true
 
@@ -123,6 +136,14 @@ enum OptimizationStrategyType {
         IteratedLocalSearchStrategy.strategyFor(optimizee, IteratedLocalSearchConfig.IteratedLocalSearch, rnd)
       case VARIABLE_NEIGHBORHOOD_SEARCH =>
         IteratedLocalSearchStrategy.strategyFor(optimizee, IteratedLocalSearchConfig.VariableNeighborhoodSearch, rnd)
+      case TABU_SEARCH =>
+        val strategy = new TabuSearchStrategy(optimizee, rnd)
+        strategy.setConfig(TabuSearchConfig.Default)
+        strategy
+      case MIN_CONFLICTS =>
+        val strategy = new MinConflictsStrategy(optimizee, rnd)
+        strategy.setConfig(MinConflictsConfig.Default)
+        strategy
 
   private def hasContinuousParameter(params: ParameterArray): Boolean =
     params match
@@ -153,7 +174,9 @@ object OptimizationStrategyType {
     DIFFERENTIAL_EVOLUTION,
     PARTICLE_SWARM,
     ITERATED_LOCAL_SEARCH, 
-    VARIABLE_NEIGHBORHOOD_SEARCH
+    VARIABLE_NEIGHBORHOOD_SEARCH,
+    TABU_SEARCH,
+    MIN_CONFLICTS
   )
 
   /** @deprecated use [[valuesForUi]] */

scala-test/com/barrybecker4/optimization/benchmark/BenchmarkConfig.scala

@@ -23,6 +23,9 @@ object BenchmarkConfig {
     SIMULATED_ANNEALING, GLOBAL_SIMULATED_ANNEALING, HILL_CLIMBING, GLOBAL_HILL_CLIMBING,
     GENETIC_SEARCH, CONCURRENT_GENETIC_SEARCH, ITERATED_LOCAL_SEARCH, VARIABLE_NEIGHBORHOOD_SEARCH)
 
+  /** The solve rate run adds the strategies made for constraint problems. */
+  val CONSTRAINT_STRATEGIES: Seq[OptimizationStrategyType] = DEFAULT_STRATEGIES ++ Seq(TABU_SEARCH, MIN_CONFLICTS)
+
   /**
     * Reads options of the form `--name=value`. The names are time-ms, max-evals, seeds, problems, strategies
     * (the last two are comma separated lists), report (the Markdown summary file, or `none`) and out (a file to
@@ -41,8 +44,8 @@ object BenchmarkConfig {
     val solveRate = options.get("solve-rate").contains("true")
     val problemNames = options.get("problems").map(_.split(",").toSeq).getOrElse(Seq.empty)
     val strategies = options.get("strategies").map(_.split(",").toSeq.map(OptimizationStrategyType.valueOf))
-      .getOrElse(DEFAULT_STRATEGIES)
-    val isFullRun = problemNames.isEmpty && strategies == DEFAULT_STRATEGIES
+      .getOrElse(if (solveRate) CONSTRAINT_STRATEGIES else DEFAULT_STRATEGIES)
+    val isFullRun = problemNames.isEmpty && strategies == (if (solveRate) CONSTRAINT_STRATEGIES else DEFAULT_STRATEGIES)
     BenchmarkConfig(
       timeLimitMillis = options.get("time-ms").map(_.toLong).getOrElse(defaultTimeLimit(solveRate)),
       maxEvaluations = options.get("max-evals").map(_.toLong).getOrElse(DEFAULT_MAX_EVALUATIONS),

scala-test/com/barrybecker4/optimization/benchmark/BenchmarkConfigSuite.scala

@@ -70,4 +70,11 @@ class BenchmarkConfigSuite extends AnyFunSuite {
   test("an ordinary run is not a solve rate run") {
     assert(!BenchmarkConfig.parse(Seq.empty).solveRate)
   }
+
+  test("the solve rate run includes the strategies for constraint problems, and a full one writes its report") {
+    val config = BenchmarkConfig.parse(Seq("--solve-rate=true"))
+    assertResult(BenchmarkConfig.CONSTRAINT_STRATEGIES)(config.strategies)
+    assert(config.strategies.containsSlice(Seq(TABU_SEARCH, MIN_CONFLICTS)))
+    assertResult(Some(BenchmarkConfig.DEFAULT_SOLVE_RATE_REPORT_FILE))(config.reportFile)
+  }
 }

scala-test/com/barrybecker4/optimization/strategy/OptimizationStrategyTypeSuite.scala

@@ -83,7 +83,8 @@ class OptimizationStrategyTypeSuite extends AnyFunSuite {
       "DIFFERENTIAL_EVOLUTION",
       "PARTICLE_SWARM",
       "ITERATED_LOCAL_SEARCH",
-      "VARIABLE_NEIGHBORHOOD_SEARCH"
+      "VARIABLE_NEIGHBORHOOD_SEARCH",
+      "TABU_SEARCH"
     ))
   }