Merge branch 'best_per_generation'

blakeaw · Jun 22, 2020 · cb16b4b · cb16b4b
2 parents b91a523 + c2224f9
commit cb16b4b
Show file tree

Hide file tree

Showing 7 changed files with 83 additions and 51 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -0,0 +1,30 @@
+# Change Log
+All notable changes to this project will be documented in this file.
+
+The format is based on [Keep a Changelog](http://keepachangelog.com/)
+and this project adheres to [Semantic Versioning](http://semver.org/).
+
+## [Unreleased] - yyyy-mm-dd
+
+N/A
+
+### Added
+
+### Changed
+
+### Fixed
+
+## [0.6.0] - 2020-06-21
+
+### Added
+- core GA now returns an array with fitness value of the fittest individual from each generation which can be accessed from the GAO property `GAO.best_fitness_per_generation`.
+
+### Fixed
+- Bug fix in core GA for sorting the population before selection and mating.
+
+## [0.5.0] - 2020-06-20
+
+### Added
+- Optional progress bar to monitor passage of generations during GAO run that is only displayed if [tqdm](https://github.com/tqdm/tqdm) is installed.
+- Optional [multiprocessing](https://docs.python.org/2/library/multiprocessing.html) based parallelism when evaluating the fitness function over the population during a GAO run.
+
diff --git a/README.md b/README.md
@@ -2,8 +2,8 @@
 
 ![Python version badge](https://img.shields.io/badge/python-3.6,3.7-blue.svg)
 [![license](https://img.shields.io/github/license/blakeaw/GAlibrate.svg)](LICENSE)
-![version](https://img.shields.io/badge/version-0.5.0-orange.svg)
-[![release](https://img.shields.io/github/release-pre/blakeaw/GAlibrate.svg)](https://github.com/blakeaw/GAlibrate/releases/tag/v0.5.0)
+![version](https://img.shields.io/badge/version-0.6.0-orange.svg)
+[![release](https://img.shields.io/github/release-pre/blakeaw/GAlibrate.svg)](https://github.com/blakeaw/GAlibrate/releases/tag/v0.6.0)
 [![anaconda cloud](https://anaconda.org/blakeaw/galibrate/badges/version.svg)](https://anaconda.org/blakeaw/galibrate)
 [![Codacy Badge](https://api.codacy.com/project/badge/Grade/6cdd91c06b11458384becb85db9adb15)](https://www.codacy.com/app/blakeaw1102/GAlibrate?utm_source=github.com&amp;utm_medium=referral&amp;utm_content=blakeaw/GAlibrate&amp;utm_campaign=Badge_Grade)
 [![DOI](https://zenodo.org/badge/197295657.svg)](https://zenodo.org/badge/latestdoi/197295657)
@@ -17,24 +17,29 @@
 Although **GAlibrate** provides a general framework for running continuous
 genetic algorithm-based optimizations, it was created with systems biology models in mind. It therefore supplies additional tools for working with biological models in the [PySB](http://pysb.org/) format.
 
-### What's new in version 0.5.0
+### What's new in
 
+#### version 0.6.0
+ * core GA now returns an array with fitness value of the fittest individual from each generation which can be accessed from the GAO property `GAO.best_fitness_per_generation`.
+ * Bug fix in core GA for sorting the population before selection and mating.
+
+#### version 0.5.0
  * Optional progress bar to monitor passage of generations during GAO run that is only displayed if [tqdm](https://github.com/tqdm/tqdm) is installed  
  * Optional [multiprocessing](https://docs.python.org/2/library/multiprocessing.html) based parallelism when evaluating the fitness function over the population during a GAO run.  
 
-
 ## Table of Contents
 
  1. [Install](#install)
      1. [pip install](#pip-install)
      2. [conda install](#conda-install)
      3. [Recomended additional software](#recomended-additional-software)
  2. [License](#license)
- 3. [Documentation and Usage](#documentation-and-usage)
+ 3. [Change Log](#change-log)
+ 4. [Documentation and Usage](#documentation-and-usage)
      1. [Quick Overview](#quick-overview)
      2. [Examples](#examples)
- 4. [Contact](#contact)
- 5. [Citing](#citing)  
+ 5. [Contact](#contact)
+ 6. [Citing](#citing)  
 
 ------
 
@@ -53,7 +58,7 @@ Note that `galibrate` has the following core dependencies:
 ### pip install
 You can install the latest release of the `galibrate` package using `pip` sourced from the GitHub repo:
 ```
-pip install -e git+https://github.com/blakeaw/GAlibrate@v0.5.0#egg=galibrate
+pip install -e git+https://github.com/blakeaw/GAlibrate@v0.6.0#egg=galibrate
 ```
 However, this will not automatically install the core dependencies. You will have to do that separately:
 ```
@@ -97,6 +102,12 @@ This project is licensed under the MIT License - see the [LICENSE](LICENSE) file
 
 ------
 
+# Change Log
+
+See: [CHANGELOG](CHANGELOG.md)
+
+------
+
 # Documentation and Usage
 
 ### Quick Overview

diff --git a/galibrate/gao.py b/galibrate/gao.py
@@ -57,6 +57,9 @@ def __init__(self, sampled_parameters, fitness_function, population_size,
         if population_size < 4:
             population_size = 4
         self.population_size = int(population_size/4.) * 4
+        if self.population_size != population_size:
+            warn_string = "--Population size was adjusted from {} to {} to make it a multiple of four--".format(population_size, self.population_size)
+            warnings.warn(warn_string)
         self.generations = generations
         self.mutation_rate = mutation_rate
         #self.survival_rate = survival_rate
@@ -66,6 +69,7 @@ def __init__(self, sampled_parameters, fitness_function, population_size,
         self._last_generation = None
         self._fittest_chromosome = None
         self._fittest_fitness = None
+        self._best_fitness_per_generation = None
 
         return
 
@@ -79,7 +83,7 @@ def run(self, verbose=False, nprocs=1):
         """
         sp_locs = np.array([sampled_parameter.loc for sampled_parameter in self.sampled_parameters])
         sp_widths = np.array([sampled_parameter.width for sampled_parameter in self.sampled_parameters])
-        last_gen_chromosomes = run_gao.run_gao(self.population_size, self._n_sp,
+        last_gen_chromosomes, best_fitness_per_generation = run_gao.run_gao(self.population_size, self._n_sp,
                                                sp_locs, sp_widths,
                                                self.generations, self.mutation_rate,
                                                self.fitness_function, nprocs)
@@ -93,4 +97,14 @@ def run(self, verbose=False, nprocs=1):
         fittest_fitness = fitnesses[fittest_idx]
         self._fittest_chromosome = fittest_chromosome
         self._fittest_fitness = fittest_fitness
+        best_fitness_per_generation[-1] = fittest_fitness
+        self._best_fitness_per_generation = best_fitness_per_generation
         return fittest_chromosome, fittest_fitness
+
+    @property
+    def best(self):
+        return self._fittest_chromosome, self._fittest_fitness
+
+    @property
+    def best_fitness_per_generation(self):
+        return self._best_fitness_per_generation
diff --git a/galibrate/run_gao_cython.pyx b/galibrate/run_gao_cython.pyx
@@ -27,6 +27,7 @@ def run_gao(int pop_size, int n_sp, np.ndarray[np.double_t, ndim=1] locs,
     # Initialize
     cdef np.ndarray[np.double_t, ndim=2, mode='c'] chromosomes = random_population(pop_size, n_sp, locs, widths)
     new_chromosome = np.zeros([pop_size, n_sp], dtype=np.double)
+    best_fitness_per_generation = np.zeros(n_gen+1, dtype=np.double)
     if nprocs > 1:
         def evaluate_fitnesses(fitness_func, chromosomes, pop_size, i_n_new, fitnesses, nprocs):
             new_fitnesses = par_fitness_eval(fitness_func, chromosomes, i_n_new, nprocs)
@@ -47,40 +48,25 @@ def run_gao(int pop_size, int n_sp, np.ndarray[np.double_t, ndim=1] locs,
 
         fitnesses_idxs = np.zeros([pop_size, 2], dtype=np.double)
         _fill_fitness_idxs(pop_size, fitnesses, fitnesses_idxs)
-        #for i_mp in range(pop_size):
-        #    fitnesses_idxs[i_mp][0] = fitnesses[i_mp]
-        #    fitnesses_idxs[i_mp][1] = i_mp
+
         # Selection
-        fitnesses_idxs_sort = np.sort(fitnesses_idxs, axis=0)
+        ind = np.argsort(fitnesses_idxs[:,0])
+        fitnesses_idxs_sort = fitnesses_idxs[ind]
+        best_fitness_per_generation[i_gen] = fitnesses_idxs_sort[-1,0]
         survivors = fitnesses_idxs_sort[pop_size/2:]
         # Move over the survivors
         _move_over_survivors(pop_size, survivors, chromosomes, new_chromosome)
-        #for i_mp in range(pop_size/2):
-        #    new_chromosome[i_mp] = chromosomes[int(survivors[i_mp][1])][:]
         mating_pairs = choose_mating_pairs(survivors, pop_size)
         # Generate children
         _generate_children(pop_size, n_sp, i_n_new, mating_pairs, chromosomes, new_chromosome)
-#        for i_mp in range(pop_size/4):
-#            i_mate1_idx = mating_pairs[i_mp][0]
-#            i_mate2_idx = mating_pairs[i_mp][1]
-#            chromosome1 = chromosomes[i_mate1_idx,:]
-#            chromosome2 = chromosomes[i_mate2_idx,:]
-#            # Crossover and update the chromosomes
-#            children = crossover(chromosome1, chromosome2, n_sp)
-#            child1 = children[0,:]
-#            child2 = children[1, :]
-#            new_chromosome[i_n_new] = child1
-#            i_n_new = i_n_new + 1
-#            new_chromosome[i_n_new] = child2
-#            i_n_new = i_n_new + 1
         # Replace the old population with the new one
         #chromosomes = new_chromosome.copy()
         double_deepcopy_2d(chromosomes, new_chromosome, pop_size, n_sp)
         # Mutation
         if i_gen < (n_gen-1):
             mutation(chromosomes, locs, widths, pop_size, n_sp, mutation_rate)
         _copy_survivor_fitnesses(pop_size, survivors, fitnesses)
-    return chromosomes
+    return chromosomes, best_fitness_per_generation
 
 @cython.boundscheck(False)
 @cython.wraparound(False)

diff --git a/galibrate/run_gao_numba.py b/galibrate/run_gao_numba.py
@@ -19,6 +19,7 @@ def run_gao(pop_size, n_sp, locs, widths, n_gen,
     # Initialize
     chromosomes = random_population(pop_size, n_sp, locs, widths)
     new_chromosome = np.zeros((pop_size, n_sp))
+    best_fitness_per_generation = np.zeros(n_gen+1)
     if nprocs > 1:
         def evaluate_fitnesses(fitness_func, chromosomes, pop_size, i_n_new, fitnesses, nprocs):
             new_fitnesses = par_fitness_eval(fitness_func, chromosomes, i_n_new, nprocs)
@@ -36,42 +37,30 @@ def evaluate_fitnesses(fitness_func, chromosomes, pop_size, i_n_new, fitnesses,
             fitnesses = np.zeros(pop_size)
             fitnesses = evaluate_fitnesses(fitness_func, chromosomes, pop_size, 0, fitnesses, nprocs)
         else:
-            fitnesses = evaluate_fitnesses(fitness_func, chromosomes, pop_size, i_n_new, fitnesses, nprocs)
+            fitnesses = evaluate_fitnesses(fitness_func, chromosomes, pop_size, 0, fitnesses, nprocs)
 
 
         fitnesses_idxs = np.zeros((pop_size, 2), dtype=np.double)
         fitnesses_idxs = _fill_fitness_idxs(pop_size, fitnesses, fitnesses_idxs)
 
         # Selection
-        fitnesses_idxs_sort = np.sort(fitnesses_idxs, axis=0)
+        ind = np.argsort(fitnesses_idxs[:,0])
+        fitnesses_idxs_sort = fitnesses_idxs[ind]
+        best_fitness_per_generation[i_gen] = fitnesses_idxs_sort[-1,0]
         survivors = fitnesses_idxs_sort[int(pop_size/2):]
         # Move over the survivors
         new_chromosome = _move_over_survivors(pop_size, survivors, chromosomes, new_chromosome)
-        #for i_mp in range(int(pop_size/2)):
-        #    new_chromosome[i_mp] = chromosomes[int(survivors[i_mp][1])][:]
+        # Choose the mating pairs
         mating_pairs = choose_mating_pairs(survivors, pop_size)
         # Generate children
         new_chromosome = _generate_children(pop_size, n_sp, i_n_new, mating_pairs, chromosomes, new_chromosome)
-#        for i_mp in range(int(pop_size/4)):
-#            i_mate1_idx = mating_pairs[i_mp][0]
-#            i_mate2_idx = mating_pairs[i_mp][1]
-#            chromosome1 = chromosomes[i_mate1_idx,:]
-#            chromosome2 = chromosomes[i_mate2_idx,:]
-#            # Crossover and update the chromosomes
-#            children = crossover(chromosome1, chromosome2, n_sp)
-#            child1 = children[0,:]
-#            child2 = children[1, :]
-#            new_chromosome[i_n_new] = child1
-#            i_n_new = i_n_new + 1
-#            new_chromosome[i_n_new] = child2
-#            i_n_new = i_n_new + 1
         # Replace the old population with the new one
         chromosomes = new_chromosome.copy()
         # Mutation
         if i_gen < (n_gen-1):
             mutation(chromosomes, locs, widths, pop_size, n_sp, mutation_rate)
         fitnesses = _copy_survivor_fitnesses(pop_size, survivors, fitnesses)
-    return chromosomes
+    return chromosomes, best_fitness_per_generation
 
 @numba.njit(cache=True)
 def _fill_fitness_idxs(pop_size, fitnesses, fitnesses_idxs):
@@ -152,7 +141,6 @@ def crossover(c1, c2, n_sp):
 
     crossover_point = int(n_sp * np.random.random())
     crossover_beta = np.random.random()
-    #cdef np.ndarray[np.double_t, ndim=2] children = np.zeros([2, n_sp], dtype=np.double)
     children = np.zeros((2, n_sp), dtype=np.double)
     x1 = c1[crossover_point]
     x2 = c2[crossover_point]

diff --git a/galibrate/run_gao_py.py b/galibrate/run_gao_py.py
@@ -16,6 +16,7 @@ def run_gao(pop_size, n_sp, locs, widths, n_gen,
     # Initialize
     chromosomes = random_population(pop_size, n_sp, locs, widths)
     new_chromosome = np.zeros([pop_size, n_sp], dtype=np.double)
+    best_fitness_per_generation = np.zeros(n_gen+1)
     if nprocs > 1:
         def evaluate_fitnesses(fitness_func, chromosomes, nprocs):
             return  par_fitness_eval(fitness_func, chromosomes, 0, nprocs)
@@ -33,7 +34,9 @@ def evaluate_fitnesses(fitness_func, chromosomes, nprocs):
             fitnesses_idxs[i_mp][0] = fitnesses[i_mp]
             fitnesses_idxs[i_mp][1] = i_mp
         # Selection
-        fitnesses_idxs_sort = np.sort(fitnesses_idxs, axis=0)
+        ind = np.argsort(fitnesses_idxs[:,0])
+        fitnesses_idxs_sort = fitnesses_idxs[ind]
+        best_fitness_per_generation[i_gen] = fitnesses_idxs_sort[-1,0]
         survivors = fitnesses_idxs_sort[int(pop_size/2):]
         # Move over the survivors
         for i_mp in range(int(pop_size/2)):
@@ -59,7 +62,7 @@ def evaluate_fitnesses(fitness_func, chromosomes, nprocs):
         if i_gen < (n_gen-1):
             mutation(chromosomes, locs, widths, pop_size, n_sp, mutation_rate)
 
-    return chromosomes
+    return chromosomes, best_fitness_per_generation
 
 def random_population(pop_size, n_sp,
                       locs, widths):

diff --git a/setup.py b/setup.py
@@ -1,7 +1,7 @@
 from distutils.core import setup
 
 setup(name='galibrate',
-      version='0.5.0',
+      version='0.6.0',
       description='Python toolkit for continuous Genetic Algorithm optimization.',
       author='Blake A. Wilson',
       author_email='[email protected]',