add generic interface for switching selection/operator policies

src/common.odin

@@ -12,9 +12,14 @@ SKEW :: 0
 TOURNAMENT_SIZE :: 5
 CROSSOVER_RATE :: 0.7
 MUTATION_RATE :: 0.01
+
+SURVIVOR_SELECTION_POLICY :: probabilistic_crowding
+PARENT_SELECTION_POLICY :: roulette_selection
+CROSSOVER_POLICY :: uniform_crossover
+MUTATION_POLICY :: bit_flip_mutation
+
 RANDOM_SEED :: u64(42)
 OUTPUT_FILE :: "output/data.csv"
-SURVIVOR_SELECTION_POLICY :: probabilistic_crowding
 
 Problem :: struct {
 	name:            string,

src/ga.odin

@@ -15,7 +15,17 @@ Survivor_Selection :: proc(
 	maximize: bool,
 ) -> Population
 
-run_ga :: proc(problem: Problem, survivor_selection: Survivor_Selection) {
+Parent_Selection :: proc(pop: ^Population, fitnesses: []f64, maximize: bool) -> Chromosome
+Crossover :: proc(parent1, parent2: Chromosome) -> (Chromosome, Chromosome)
+Mutation :: proc(chromosome: Chromosome)
+
+run_ga :: proc(
+	problem: Problem,
+	survivor_selection: Survivor_Selection,
+	parent_selection: Parent_Selection,
+	crossover: Crossover,
+	mutation: Mutation,
+) {
 	fmt.printfln("=== Running GA: %s ===", problem.name)
 
 	population := generate_population(problem.chromosome_size)
@@ -38,7 +48,14 @@ run_ga :: proc(problem: Problem, survivor_selection: Survivor_Selection) {
 			stats.entropy,
 		)
 
-		offspring := create_offspring_simple(&population)
+		offspring := create_offspring(
+			&population,
+			pop_fitnesses[:],
+			problem.maximize,
+			parent_selection,
+			crossover,
+			mutation,
+		)
 		offspring_fitnesses := evaluate_population(&offspring, problem.fitness_proc)
 
 		next_gen := survivor_selection(
@@ -112,7 +129,6 @@ compute_population_entropy :: proc(pop: ^Population, chromosome_size: int) -> f6
 	return entropy
 }
 
-// Modified compute_stats to include entropy
 compute_stats :: proc(pop: ^Population, fitnesses: []f64, chromosome_size: int, maximize: bool) -> Stats {
 	best := maximize ? -math.F64_MAX : math.F64_MAX
 	worst := maximize ? math.F64_MAX : -math.F64_MAX
@@ -251,16 +267,23 @@ clone_chromosome :: proc(chrom: Chromosome) -> Chromosome {
     return clone
 }
 
-create_offspring_simple :: proc(pop: ^Population) -> Population {
+create_offspring :: proc(
+	pop: ^Population,
+	fitnesses: []f64,
+	maximize: bool,
+	parent_selection: Parent_Selection,
+	crossover: Crossover,
+	mutation: Mutation,
+) -> Population {
 	offspring: Population
 
 	for i := 0; i < POPULATION_SIZE; i += 2 {
-		p1_idx := rand.int_max(POPULATION_SIZE)
-		p2_idx := rand.int_max(POPULATION_SIZE)
+		parent1 := parent_selection(pop, fitnesses, maximize)
+		parent2 := parent_selection(pop, fitnesses, maximize)
 
-		child1, child2 := uniform_crossover(pop[p1_idx], pop[p2_idx])
-		bit_flip_mutation(child1)
-		bit_flip_mutation(child2)
+		child1, child2 := crossover(parent1, parent2)
+		mutation(child1)
+		mutation(child2)
 
 		offspring[i] = child1
 		if i + 1 < POPULATION_SIZE {
@@ -273,6 +296,10 @@ create_offspring_simple :: proc(pop: ^Population) -> Population {
 	return offspring
 }
 
+random_selection :: proc(pop: ^Population, fitnesses: []f64, maximize: bool) -> Chromosome {
+	return pop[rand.int_max(POPULATION_SIZE)]
+}
+
 deterministic_crowding :: proc(
 	pop: ^Population,
 	offspring: ^Population,

src/main.odin

@@ -7,8 +7,10 @@ import "base:runtime"
 main :: proc() {
 	problem_type := "feature_selection"
 
-	// state := rand.create(RANDOM_SEED)
-	// context.random_generator = runtime.default_random_generator(&state)
+	when RANDOM_SEED != u64(0){
+		state := rand.create(RANDOM_SEED)
+		context.random_generator = runtime.default_random_generator(&state)
+	}
 
 	problem: Problem
 	switch problem_type {
@@ -33,5 +35,11 @@ main :: proc() {
 		fmt.printfln("RMSE: %.4f\n", baseline)
 	}
 
-	run_ga(problem, SURVIVOR_SELECTION_POLICY)
+	run_ga(
+		problem,
+		SURVIVOR_SELECTION_POLICY,
+		PARENT_SELECTION_POLICY,
+		CROSSOVER_POLICY,
+		MUTATION_POLICY
+	)
 }