// This file is a part of Framsticks SDK. http://www.framsticks.com/ // Copyright (C) 1999-2018 Maciej Komosinski and Szymon Ulatowski. // See LICENSE.txt for details. #include "fn_oper.h" #include "fn_conv.h" #include //randomN, rnd01 /** \class GenoOper_fn This genetic representation only stores a vector of real numbers. A fitness function must be provided for the gene pool, for example the "Booth function" would be: var X = String.deserialize(this.geno.rawgenotype); //a vector of real values var result = Math.pow(X[0]+2*X[1]-7,2) + Math.pow(2*X[0]+X[1]-5,2); return -result; //negation because Framsticks assumes maximization, and the original function needs to be minimized */ #define FIELDSTRUCT GenoOper_fn static ParamEntry GENOfnparam_tab[] = { { "Genetics: fn", 1, 6, }, { "fn_xover", 0, 0, "Fraction inherited in linear mix crossover", "f 0.5 1.0 0.9", FIELD(xover_proportion), "0.5 => children are averaged parents.\n0.8 => children are only 20% different from parents.\n1.0 => each child is identical to one parent (no crossover).", }, { "fn_xover_random", 0, 0, "Random fraction inherited in crossover", "d 0 1 1", FIELD(xover_proportion_random), "If active, the amount of linear mix is random in each crossover operation, so the \"Fraction inherited in linear mix crossover\" parameter is ignored.", }, { "fn_mut_bound_low", 1, 0, "Lower bounds for mutation", "s 0 0 [-10.0, -10.0]", FIELD(mut_bound_low), "A vector of lower bounds (one real value for each variable)", }, { "fn_mut_bound_high", 1, 0, "Higher bounds for mutation", "s 0 0 [10.0, 10.0]", FIELD(mut_bound_high), "A vector of higher bounds (one real value for each variable)", }, { "fn_mut_stddev", 1, 0, "Standard deviations for mutation", "s 0 0 [0.1, 0.1]", FIELD(mut_stddev), "A vector of standard deviations (one real value for each variable)", }, { "fn_mut_single_var", 0, 0, "Mutate only a single variable", "d 0 1 0", FIELD(mut_single_var), "If active, only a single randomly selected variable will be mutated in each mutation operation. Otherwise all variables will be mutated.", }, { 0, }, }; #undef FIELDSTRUCT GenoOper_fn::GenoOper_fn() { par.setParamTab(GENOfnparam_tab); par.select(this); par.setDefault(); supported_format = 'n'; } int GenoOper_fn::checkValidity(const char* gene, const char *genoname) { vector values = GenoConv_fn0::stringToVector(gene); return values.size() > 0 ? GENOPER_OK : 1; } int GenoOper_fn::validate(char *&gene, const char *genoname) { vector values = GenoConv_fn0::stringToVector(gene); if (values.size() == 0) values.push_back(0.0); string validated = GenoConv_fn0::vectorToString(values); free(gene); gene = strdup(validated.c_str()); //reallocate return GENOPER_OK; } //Creep-mutate variable(s) int GenoOper_fn::mutate(char *&gene, float &chg, int &method) { method = 0; vector values = GenoConv_fn0::stringToVector(gene); if (values.size() == 0) return GENOPER_OPFAIL; vector bound_low = GenoConv_fn0::stringToVector(mut_bound_low.c_str()); vector bound_high = GenoConv_fn0::stringToVector(mut_bound_high.c_str()); vector stddev = GenoConv_fn0::stringToVector(mut_stddev.c_str()); if (bound_low.size() != bound_high.size() || bound_high.size() != stddev.size() || stddev.size() != values.size()) { logPrintf("GenoOper_fn", "mutate", LOG_ERROR, "The solution vector, bound vectors, and standard deviation vectors must all have the same number of values"); return GENOPER_OPFAIL; } if (mut_single_var) //mutate only one, randomly selected variable { int which = randomN(values.size()); values[which] = GenoOperators::mutateCreep('f', values[which], bound_low[which], bound_high[which], stddev[which], false); chg = 1.0f / values.size(); } else //mutate all variables { for (int which = 0; which < (int)values.size(); which++) values[which] = GenoOperators::mutateCreep('f', values[which], bound_low[which], bound_high[which], stddev[which], false); chg = 1.0f; } string saved = GenoConv_fn0::vectorToString(values); free(gene); gene = strdup(saved.c_str()); //reallocate return GENOPER_OK; } //Averaging crossover int GenoOper_fn::crossOver(char *&g1, char *&g2, float& chg1, float& chg2) { //g1 = strdup("[1,0.5,0.5,0.5,0.5,1,1]"); //testing... //g2 = strdup("[4,1, 1, 1, 1, 2,2]"); //testing... //xover_proportion = 0.1; //testing... double proportion = xover_proportion_random ? 0.5 + rnd0N(0.5) : xover_proportion; chg1 = proportion; chg2 = 1 - proportion; vector v1 = GenoConv_fn0::stringToVector(g1); vector v2 = GenoConv_fn0::stringToVector(g2); if (v1.size() != v2.size()) { logPrintf("GenoOper_fn", "crossOver", LOG_ERROR, "Tried to cross over solutions with a differing number of variables (%d and %d)", v1.size(), v2.size()); return GENOPER_OPFAIL; } GenoOperators::linearMix(v1, v2, proportion); string saved = GenoConv_fn0::vectorToString(v1); free(g1); g1 = strdup(saved.c_str()); //reallocate saved = GenoConv_fn0::vectorToString(v2); free(g2); g2 = strdup(saved.c_str()); //reallocate return GENOPER_OK; } //Applying some colors and font styles... uint32_t GenoOper_fn::style(const char *g, int pos) { char ch = g[pos]; uint32_t style = GENSTYLE_CS(0, GENSTYLE_INVALID); //default, should be changed below if (strchr("-.e 0123456789", ch) != NULL) style = GENSTYLE_CS(GENCOLOR_NUMBER, GENSTYLE_NONE); else if (strchr("[,]", ch) != NULL) style = GENSTYLE_RGBS(0, 0, 0, GENSTYLE_BOLD); return style; }