Changeset 1060

Timestamp:

01/24/21 13:29:42 (4 years ago)

Author:

Maciej Komosinski

Message:

"Vectorized" mutation for better performance due to decreased time of communication py<->frams (mutate many genotypes in one call)

Location:

framspy

Files:

• FramsticksCLI.py (modified) (6 diffs)
• FramsticksEvolution.py (modified) (4 diffs)

framspy/FramsticksCLI.py

@@ -7 +7 @@
 import argparse
 import numpy as np
+import framsreader  # only needed for mutation: https://pypi.org/project/framsreader
 
 
@@ -24 +25 @@
         GENO_SAVE_FILE_FORMAT = Enum('GENO_SAVE_FILE_FORMAT', 'NATIVEFRAMS RAWGENO')  # how to save genotypes
         OUTPUT_DIR = "scripts_output"
+        GENOTYPE_INVALID = "/*invalid*/"  # this is how genotype invalidity is represented in Framsticks
         STDOUT_ENDOPER_MARKER = "FileObject.write"  # we look for this message on Framsticks CLI stdout to detect when Framsticks created a file with the result we expect
 
@@ -35 +37 @@
         EVALUATE_FILE = "genos_eval.json"
         CROSSOVER_CMD = "crossover"
-        CROSSOVER_FILE = "child.gen"
+        CROSSOVER_FILE = "crossover_child.gen"
         DISSIMIL_CMD = "dissimil"
         DISSIMIL_FILE = "dissimilarity_matrix.tsv"  # tab-separated values
-        ISVALID_CMD = "arevalid"
+        ISVALID_CMD = "isvalid_many"
         ISVALID_FILE = "validity.txt"
-        MUTATE_CMD = "mutate"
-        MUTATE_FILE = "mutant.gen"
+        MUTATE_CMD = "mutate_many"
+        MUTATE_FILE = "mutation_results.gen"
 
         CLI_INPUT_FILE = "genotypes.gen"
@@ -219 +221 @@
 
 
-        def mutate(self, genotype: str) -> str:
-                """
-                Returns:
-                        The genotype of the mutated individual. Empty string if the mutation failed.
-                """
-                files = self.__runCommand(self.MUTATE_CMD, [genotype], self.MUTATE_FILE, self.GENO_SAVE_FILE_FORMAT["RAWGENO"])
-                with open(files[-1]) as f:
-                        newgenotype = "".join(f.readlines())
-                self.__cleanUpCommandResults(files)
-                return newgenotype
+        def mutate(self, genotype_list: List[str]) -> List[str]:
+                """
+                Returns:
+                        The genotype(s) of the mutated source genotype(s). self.GENOTYPE_INVALID for genotypes whose mutation failed (for example because the source genotype was invalid).
+                """
+                assert isinstance(genotype_list, list)  # because in python str has similar capabilities as list and here it would pretend to work too, so to avoid any ambiguity
+                files = self.__runCommand(self.MUTATE_CMD, genotype_list, self.MUTATE_FILE, self.GENO_SAVE_FILE_FORMAT["NATIVEFRAMS"])
+                genos = framsreader.load(files[-1], "gen file")
+                self.__cleanUpCommandResults(files)
+                return [g["genotype"] for g in genos]
 
 
@@ -298 +300 @@
         # TODO ideas:
         # - check_validity with three levels (invalid, corrected, valid)
-        # - "vectorize" some operations (isvalid, evaluate) so that a number of genotypes is handled in one call
+        # - "vectorize" crossover so that many genotypes is handled in one call. Even better, use .so/.dll direct communication to CLI
         # - use threads for non-blocking reading from frams' stdout and thus not relying on specific strings printed by frams
         # - a pool of binaries run at the same time, balance load - in particular evaluation
-        # - if we read genotypes in "org:" format anywhere: import https://pypi.org/project/framsreader/0.1.2/ and use it if successful,
+        # - if we read genotypes in "org:" format anywhere: import https://pypi.org/project/framsreader and use it if successful,
         #    if not then print a message "framsreader not available, using simple internal method to save a genotype" and proceed as it is now.
-        #    So far we don't read, but we should use the proper writer to handle all special cases like quoting etc.
+        #    We should use the proper writer to handle all special cases like quoting special characters etc.
 
         parsed_args = parseArguments()
@@ -312 +314 @@
         simplest = framsCLI.getSimplest('1' if parsed_args.genformat is None else parsed_args.genformat)
         print("\tSimplest genotype:", simplest)
-        parent1 = framsCLI.mutate(simplest)
+        parent1 = framsCLI.mutate([simplest])[0]
         parent2 = parent1
         MUTATE_COUNT = 10
         for x in range(MUTATE_COUNT):  # example of a chain of 20 mutations
-                parent2 = framsCLI.mutate(parent2)
+                parent2 = framsCLI.mutate([parent2])[0]
         print("\tParent1 (mutated simplest):", parent1)
         print("\tParent2 (Parent1 mutated %d times):" % MUTATE_COUNT, parent2)

framspy/FramsticksEvolution.py

@@ -14 +14 @@
 
 def frams_evaluate(frams_cli, individual):
-        genotype = individual[0]  # [0] because we can't (?) have a simple str as a deap genotype/individual, only list of str.
+        genotype = individual[0]  # individual[0] because we can't (?) have a simple str as a deap genotype/individual, only list of str.
         data = frams_cli.evaluate([genotype])
         # print("Evaluated '%s'" % genotype, 'evaluation is:', data)
@@ -22 +22 @@
                 default_evaluation_data = evaluation_data[""]
                 fitness = [default_evaluation_data[crit] for crit in OPTIMIZATION_CRITERIA]
-        except (KeyError, TypeError) as e:  # the evaluation may have failed for invalid genotypes (or some other reason)
-                fitness = [-1] * len(OPTIMIZATION_CRITERIA)
+        except (KeyError, TypeError) as e:  # the evaluation may have failed for an invalid genotype (such as X[@][@] with "Don't simulate genotypes with warnings" option) or for some other reason
+                fitness = [-1] * len(OPTIMIZATION_CRITERIA)  # fitness of -1 is intended to discourage further propagation of this genotype via selection ("this one is very poor")
                 print("Error '%s': could not evaluate genotype '%s', returning fitness %s" % (str(e), genotype, fitness))
         return fitness
@@ -29 +29 @@
 
 def frams_crossover(frams_cli, individual1, individual2):
-        geno1 = individual1[0]  # [0] because we can't (?) have a simple str as a deap genotype/individual, only list of str.
-        geno2 = individual2[0]  # [0] because we can't (?) have a simple str as a deap genotype/individual, only list of str.
+        geno1 = individual1[0]  # individual[0] because we can't (?) have a simple str as a deap genotype/individual, only list of str.
+        geno2 = individual2[0]  # individual[0] because we can't (?) have a simple str as a deap genotype/individual, only list of str.
         individual1[0] = frams_cli.crossOver(geno1, geno2)
         individual2[0] = frams_cli.crossOver(geno1, geno2)
@@ -37 +37 @@
 
 def frams_mutate(frams_cli, individual):
-        individual[0] = frams_cli.mutate(individual[0])
+        individual[0] = frams_cli.mutate([individual[0]])[0]  # individual[0] because we can't (?) have a simple str as a deap genotype/individual, only list of str.
         return individual,