Changeset 802 for cpp

Timestamp:

06/07/18 17:41:59 (6 years ago)

Author:

Maciej Komosinski

Message:

Crossing over with less bloat, but still biologically-inspired

Location:

cpp/frams/genetics/fB

Files:

• fB_conv.cpp (modified) (3 diffs)
• fB_conv.h (modified) (1 diff)
• fB_general.h (modified) (3 diffs)
• fB_oper.cpp (modified) (4 diffs)

cpp/frams/genetics/fB/fB_conv.cpp

@@ -170 +170 @@
 }
 
-bool GenoConv_fBH::getNextCharId(SString genotype, int &i)
+bool GenoConv_fBH::getNextCharId(const SString& genotype, int &i)
 {
         i++;
@@ -191 +191 @@
 }
 
-bool GenoConv_fBH::getNeuroClass(SString gene, SString &def, int nclassdefcount)
+bool GenoConv_fBH::getNeuroClass(const SString& gene, SString &def, int nclassdefcount)
 {
         SString lastdef = "N";
@@ -218 +218 @@
 }
 
-int GenoConv_fBH::processNextLetter(fH_Builder &creature, fH_Handle *&currhandle, Param &par, SString gene, int &propindex, int &i, std::vector<IRange> ranges[3], int &nclassdefcount)
+int GenoConv_fBH::processNextLetter(fH_Builder &creature, fH_Handle *&currhandle, Param &par, const SString& gene, int &propindex, int &i, std::vector<IRange> ranges[3], int &nclassdefcount)
 {
         if (propindex >= par.getPropCount())

cpp/frams/genetics/fB/fB_conv.h

@@ -12 +12 @@
 {
 private:
-        bool getNextCharId(SString genotype, int &i);
+        bool getNextCharId(const SString& genotype, int &i);
         double convertCharacterTo01(char c);
         double convertCharacterToWeight(char c);
         static fH_Handle* convertCharacterToHandle(char c, int dims, int start, int end, std::vector<IRange> ranges[3]);
-        int processNextLetter(fH_Builder &creature, fH_Handle *&currhandle, Param &par, SString gene, int &propindex, int &i, std::vector<IRange> ranges[3], int &nclassdefcount);
-        bool getNeuroClass(SString gene, SString &def, int nclassdefcount);
+        int processNextLetter(fH_Builder &creature, fH_Handle *&currhandle, Param &par, const SString& gene, int &propindex, int &i, std::vector<IRange> ranges[3], int &nclassdefcount);
+        bool getNeuroClass(const SString& gene, SString &def, int nclassdefcount);
 
 public:

cpp/frams/genetics/fB/fB_general.h

@@ -11 +11 @@
 {
 public:
-        static int geneCount(SString geno)
+        static int geneCount(const SString& geno)
         {
                 int start = 0;
@@ -27 +27 @@
         }
 
-        static SString getGene(int i, SString genotype, int &start, int &end)
+        static SString getNonNestedGene(int i, const SString& genotype, int &start, int &end)
+        {
+                int count = -1;
+                start = 0;
+                int tmp = 0;
+                SString result = "";
+                do {
+                        count++;
+                        if (start < genotype.len())
+                                result = getNextGene(start, genotype, tmp, start);
+                        else
+                                start = -1;
+                } while (start != -1 && count < i);
+                start = tmp;
+                end = start;
+                return result;
+        }
+
+        static int geneCountNoNested(const SString& geno)
+        {
+                int start = 0;
+                int count = -1;
+                int tmp = 0;
+                do {
+                        count++;
+                        if (start < geno.len())
+                                getNextGene(start, geno, tmp, start);
+                        else
+                                start = -1;
+                } while (start != -1);
+                return count;
+        }
+
+        static SString getGene(int i, const SString& genotype, int &start, int &end)
         {
                 start = 0;
@@ -62 +95 @@
         }
 
+        static SString getNextGene(int searchbegin, const SString& genotype, int &start, int &end)
+        {
+                start = searchbegin;
+                int count = -1;
+                do {
+                        count++;
+                        start = genotype.indexOf("aa", start) + 1;
+                        int quotecount = 0;
+                        for (int q = 0; q < start; q++) if (genotype[q] == '\"') quotecount++;
+                        if (quotecount % 2 != 0) count--; // 'aa' sequence is within quotes
+                } while (start - 1 != -1 && count != 0);
+                if (start - 1 == -1) // there is no gene with a given "i"
+                {
+                        start = -1;
+                        end = -1;
+                        return "";
+                }
+                end = start;
+                int quotecount = 0;
+                do {
+                        quotecount = 0;
+                        end = genotype.indexOf("zz", end);
+                        if (end != -1)
+                        {
+                                for (int q = start; q < end; q++) if (genotype[q] == '\"') quotecount++;
+                                if (quotecount % 2 != 0) end++;
+                        }
+                } while (quotecount % 2 != 0 && end != -1);
+
+                if (end == -1) end = genotype.len();
+                else end += 2;
+                start -= 1;
+                return genotype.substr(start, end - start);
+        }
 private:
         fB_GenoHelpers() {}

cpp/frams/genetics/fB/fB_oper.cpp

@@ -19 +19 @@
         { "Genetics: fB: Crossover", },
         { "fB_mut_substitution", 1, 0, "Substitution", "f 0 1 0.6", FIELD(mutationprobs[FB_SUBSTITUTION]), "Probability of mutation by changing single random letter in genotype", },
-        { "fB_mut_insertion", 1, 0, "Insertion", "f 0 1 0.95", FIELD(mutationprobs[FB_INSERTION]), "Probability of mutation by inserting characters in random place of genotype", },
-        { "fB_mut_nclassins", 1, 0, "Insertion of neuron class definition", "f 0 1 0.05", FIELD(mutationprobs[FB_NCLASSINS]), "Probability of mutation by inserting neuron class definition in random place of genotype", },
+        { "fB_mut_insertion", 1, 0, "Insertion", "f 0 1 0.095", FIELD(mutationprobs[FB_INSERTION]), "Probability of mutation by inserting characters in random place of genotype", },
+        { "fB_mut_nclassins", 1, 0, "Insertion of neuron class definition", "f 0 1 0.005", FIELD(mutationprobs[FB_NCLASSINS]), "Probability of mutation by inserting neuron class definition in random place of genotype", },
         { "fB_mut_deletion", 1, 0, "Deletion", "f 0 1 0.1", FIELD(mutationprobs[FB_DELETION]), "Probability of mutation by deleting random characters in genotype", },
-        { "fB_mut_duplication", 1, 0, "Duplication", "f 0 1 0.05", FIELD(mutationprobs[FB_DUPLICATION]), "Probability of mutation by copying single *gene* of genotype and appending it to the beginning of this genotype", },
+        { "fB_mut_duplication", 1, 0, "Duplication", "f 0 1 0.0", FIELD(mutationprobs[FB_DUPLICATION]), "Probability of mutation by copying single *gene* of genotype and appending it to the beginning of this genotype", },
         { "fB_mut_translocation", 1, 0, "Translocation", "f 0 1 0.15", FIELD(mutationprobs[FB_TRANSLOCATION]), "Probability of mutation by replacing two substrings in genotype", },
-        { "fB_cross_gene_transfer", 2, 0, "Horizontal gene transfer", "f 0 1 0.8", FIELD(crossoverprobs[FB_GENE_TRANSFER]), "Probability of crossing over by transferring single genes from both parents to beginning of each other", },
-        { "fB_cross_crossover", 2, 0, "Crossing over", "f 0 1 0.2", FIELD(crossoverprobs[FB_CROSSING_OVER]), "Probability of crossing over by random distribution of genes from both parents to both children", },
+        { "fB_cross_gene_transfer", 2, 0, "Horizontal gene transfer", "f 0 1 0.0", FIELD(crossoverprobs[FB_GENE_TRANSFER]), "Probability of crossing over by transferring single genes from both parents to beginning of each other", },
+        { "fB_cross_crossover", 2, 0, "Crossing over", "f 0 1 1.0", FIELD(crossoverprobs[FB_CROSSING_OVER]), "Probability of crossing over by random distribution of genes from both parents to both children", },
         { 0, },
 };
@@ -281 +281 @@
                 break;
         }
+        case FB_NCLASSINS:
+        {
+                std::list<SString> tokenized = tokenizeSequence(line);
+                std::list<SString>::iterator it = tokenized.begin();
+                int rndid = randomN(tokenized.size()); // select random insertion point
+                std::advance(it, rndid);
+                NeuroClass *cls = getRandomNeuroClass();
+                if (cls)
+                {
+                        SString classdef = cls->getName();
+                        Geno_fH::mutateNeuronProperties(classdef);
+                        SString res = "\"";
+                        res += classdef;
+                        res += "\"";
+                        tokenized.insert(it, res);
+                        chg = (double)classdef.len() / line.len();
+                        line = detokenizeSequence(tokenized);
+                        break;
+                }
+        }
+        /* no break */
         case FB_INSERTION:
         {
@@ -291 +312 @@
                 letter.directWrite()[0] = 'a' + randomN(26);
                 tokenized.insert(it, letter);
-                line = detokenizeSequence(tokenized);
-                break;
-        }
-        case FB_NCLASSINS:
-        {
-                std::list<SString> tokenized = tokenizeSequence(line);
-                std::list<SString>::iterator it = tokenized.begin();
-                int rndid = randomN(tokenized.size()); // select random insertion point
-                std::advance(it, rndid);
-                NeuroClass *cls = getRandomNeuroClass();
-                SString classdef = cls->getName();
-                Geno_fH::mutateNeuronProperties(classdef);
-                SString res = "\"";
-                res += classdef;
-                res += "\"";
-                tokenized.insert(it, res);
-                chg = (double)classdef.len() / line.len();
                 line = detokenizeSequence(tokenized);
                 break;
@@ -416 +420 @@
                 break;
         }
+//      case FB_CROSSING_OVER:
+//      {
+//              // iterate through all genes of the first parent and assign them
+//              // randomly to children
+//              for (int i = 0; i < fB_GenoHelpers::geneCount(parent1); i++)
+//              {
+//                      int start, end;
+//                      SString gene = fB_GenoHelpers::getGene(i, parent1, start, end);
+//                      if (randomN(2) == 0)
+//                      {
+//                              child1 += gene;
+//                              chg1 += 1.0f;
+//                      }
+//                      else
+//                      {
+//                              child2 += gene;
+//                      }
+//              }
+//              chg1 /= fB_GenoHelpers::geneCount(parent1);
+//
+//              // do the same with second parent
+//              for (int i = 0; i < fB_GenoHelpers::geneCount(parent2); i++)
+//              {
+//                      int start, end;
+//                      SString gene = fB_GenoHelpers::getGene(i, parent2, start, end);
+//                      if (randomN(2) == 0)
+//                      {
+//                              child1 += gene;
+//                      }
+//                      else
+//                      {
+//                              child2 += gene;
+//                              chg2 += 1.0f;
+//                      }
+//              }
+//              chg2 /= fB_GenoHelpers::geneCount(parent2);
+//              break;
+//      }
         case FB_CROSSING_OVER:
         {
-                // iterate through all genes of the first parent and assign them
-                // randomly to children
-                for (int i = 0; i < fB_GenoHelpers::geneCount(parent1); i++)
-                {
-                        int start, end;
-                        SString gene = fB_GenoHelpers::getGene(i, parent1, start, end);
-                        if (randomN(2) == 0)
-                        {
-                                child1 += gene;
-                                chg1 += 1.0f;
-                        }
-                        else
-                        {
-                                child2 += gene;
-                        }
-                }
-                chg1 /= fB_GenoHelpers::geneCount(parent1);
-
-                // do the same with second parent
-                for (int i = 0; i < fB_GenoHelpers::geneCount(parent2); i++)
-                {
-                        int start, end;
-                        SString gene = fB_GenoHelpers::getGene(i, parent2, start, end);
-                        if (randomN(2) == 0)
-                        {
-                                child1 += gene;
-                        }
-                        else
-                        {
-                                child2 += gene;
-                                chg2 += 1.0f;
-                        }
-                }
-                chg2 /= fB_GenoHelpers::geneCount(parent2);
+                // get maximal count of genes from both parents
+                int maxgenecount = max(fB_GenoHelpers::geneCountNoNested(parent1),
+                                fB_GenoHelpers::geneCountNoNested(parent2));
+
+                // while there are genes in at least one genotype
+                for (int i = 0; i < maxgenecount; i++)
+                {
+                        SString to1 = "", to2 = "";
+                        int start = 0, end = 0;
+
+                        // if both parents have genes available, then distribute them
+                        if (i < fB_GenoHelpers::geneCountNoNested(parent1) &&
+                                        i < fB_GenoHelpers::geneCountNoNested(parent2))
+                        {
+                                if (randomN(2) == 0)
+                                {
+                                        to1 = fB_GenoHelpers::getNonNestedGene(i, parent1, start, end);
+                                        to2 = fB_GenoHelpers::getNonNestedGene(i, parent2, start, end);
+                                        chg1 += 1.0f;
+                                        chg2 += 1.0f;
+                                }
+                                else
+                                {
+                                        to1 = fB_GenoHelpers::getNonNestedGene(i, parent2, start, end);
+                                        to2 = fB_GenoHelpers::getNonNestedGene(i, parent1, start, end);
+                                }
+                        }
+                        else if (i < fB_GenoHelpers::geneCountNoNested(parent1))
+                        {
+                                if (randomN(2) == 0)
+                                {
+                                        to1 = fB_GenoHelpers::getNonNestedGene(i, parent1, start, end);
+                                        chg1 += 1.0f;
+                                }
+                                else
+                                {
+                                        to2 = fB_GenoHelpers::getNonNestedGene(i, parent1, start, end);
+                                }
+                        }
+                        else // if (i < fB_GenoHelpers::geneCountNoNested(parent2))
+                        {
+                                if (randomN(2) == 0)
+                                {
+                                        to1 = fB_GenoHelpers::getNonNestedGene(i, parent2, start, end);
+                                }
+                                else
+                                {
+                                        to2 = fB_GenoHelpers::getNonNestedGene(i, parent2, start, end);
+                                        chg2 += 1.0f;
+                                }
+                        }
+                        child1 += to1;
+                        child2 += to2;
+                }
+
+                chg1 /= fB_GenoHelpers::geneCountNoNested(parent1);
+                chg2 /= fB_GenoHelpers::geneCountNoNested(parent2);
                 break;
         }