From 7d99bf7623c33a98f641be989d16c93793b1beed Mon Sep 17 00:00:00 2001
From: Winnus <winnus@posteo.de>
Date: Fri, 22 Sep 2023 15:52:59 +0200
Subject: [PATCH] Added processing time output to each epoch
---
README.md | 7 +++
src/PyGMA.py | 46 ++++++-------------
src/components/config_func_op.py | 0
.../config_logical_gate_construction.py | 14 +++---
...e_construction_no_explicit_input_coding.py | 0
src/components/evolutionary_phase.py | 8 ++++
src/components/genetic_operators.py | 1 -
src/config-Copy1.py | 0
src/core/controller.py | 15 ++++--
src/core/mpi_worker.py | 3 +-
10 files changed, 50 insertions(+), 44 deletions(-)
mode change 100644 => 100755 src/components/config_func_op.py
mode change 100644 => 100755 src/components/config_logical_gate_construction.py
mode change 100644 => 100755 src/components/config_logical_gate_construction_no_explicit_input_coding.py
mode change 100644 => 100755 src/config-Copy1.py
diff --git a/README.md b/README.md
index 5137f13..cc018a3 100755
--- a/README.md
+++ b/README.md
@@ -79,3 +79,10 @@ By conductiong the experiment the a fitness value is generated for the gene.
This should be returned.
+
+
+# Todos
+- implement a call to the experiment that handles how to save the fittest gene when evolution is finished.
+ - this has to be user defined and will save the gene into a users usable format for further use
+
+
diff --git a/src/PyGMA.py b/src/PyGMA.py
index 3096eb2..bf0e970 100755
--- a/src/PyGMA.py
+++ b/src/PyGMA.py
@@ -2,16 +2,6 @@ import numpy as np
from config import CONFIG, check_config
from core.controller import Controller
-
-# remove here
-def bool2int(x):
- r = 0
- for i, b in enumerate(x):
- if b:
- r += b << i
- return r
-
-
def mpi_worker(config):
# instantiate the worker
from core.mpi_worker import MPI_worker
@@ -28,16 +18,18 @@ def controller():
"""
# Init controller
GAController = Controller(CONFIG, rng_seed=9)
- print("Init controller")
- print(GAController.populations)
- print(GAController.populations[0].genetic_operator_stack)
- print(GAController.populations[0].individuals[0].genome)
+
+ # print information
+ print("Init Evolution controller:")
+ print(f'populations: {len(GAController.populations)}')
+ for i, pop in zip(range(len(GAController.populations)), GAController.populations):
+ print(f'Genetic Operators p{i}: {len(pop.genetic_operator_stack)}')
# evolution for all phases
GAController.evolve()
- print("Evolve finished")
+ print("Evolve finished (fitness or epoch criterion met)")
- print("First best genes")
+
# todo find the population with the best gene
# get max fitness values
# this is started inside the controller in get_fittest_individuals()
@@ -46,24 +38,14 @@ def controller():
m = max(fit_per_pop)
i = np.where(fit_per_pop == m)
- print(GAController.populations[0].individuals[0].fitness)
- print(GAController.populations[0].individuals[1].fitness)
+ print(f'Population fitness:{fit_per_pop}')
- print(GAController.populations[0].individuals[0].genome)
- gene = GAController.populations[0].individuals[0].genome
- numbers = np.split(gene, 3)
- # convert the numbers to ints
- x = bool2int(numbers[0])
- y = bool2int(numbers[1])
- z = bool2int(numbers[2])
- print(f"{x} , {y}, {z}")
- print(x * 2 + y * 3 + z)
+# print(GAController.populations[0].individuals[0].fitness)
+# print(GAController.populations[0].individuals[1].fitness)
- # epoch loop
- # GAController.epoch()
- # print('Epoch')
- # print(GAController.populations[0].individuals[0].genome)
+# print(GAController.populations[0].individuals[0].genome)
+# gene = GAController.populations[0].individuals[0].genome
return
@@ -89,7 +71,7 @@ def main():
rank = comm.Get_rank()
# if mpi worker switch to worker mode
- if rank != 0:
+ if rank > 0:
mpi_worker(config)
# rank 0 is the controller
diff --git a/src/components/config_func_op.py b/src/components/config_func_op.py
old mode 100644
new mode 100755
diff --git a/src/components/config_logical_gate_construction.py b/src/components/config_logical_gate_construction.py
old mode 100644
new mode 100755
index 9f9988b..d0d46c4
--- a/src/components/config_logical_gate_construction.py
+++ b/src/components/config_logical_gate_construction.py
@@ -64,9 +64,9 @@ def n_bit_multiplication_inout(n):
# Populations
# Genetic Phases
-OP_MUTATION = Mutation_operator(0.05, 10)
-OP_CROSSOVER = Single_point_crossover_operator(10)
-OP_REMOVAL = Removal_operator(20)
+OP_MUTATION = Mutation_operator(0.05, 45)
+OP_CROSSOVER = Single_point_crossover_operator(45)
+OP_REMOVAL = Removal_operator(90)
@@ -90,7 +90,7 @@ exp_inputs, exp_expected_outputs = n_bit_multiplication_inout(num_bits)
num_logical_inputs = num_bits+num_bits # we have two numbers each having n bits
num_logical_outputs = num_bits+num_bits# we need to save the result
-num_logical_gates = 96 # how many logic gates we want to use
+num_logical_gates = 128 # how many logic gates we want to use
PHASE0_EXPERIMENT = Logical_circuit_experiment(exp_inputs, exp_expected_outputs, num_logical_inputs, num_logical_outputs, num_logical_gates)
@@ -121,7 +121,7 @@ CONFIG_CIRCUIT_EVOLVE = {
# mpiexec -n 3 python PyGMA.py
# or use threads if not specified via slurm
# mpiexec -n 3 --use-hwthread-cpus python PyGMA.py
- 'use_mpi': False,
+ 'use_mpi': True,
# use mpi4py.futures
# this will dynamically spawn workers.
@@ -149,7 +149,7 @@ CONFIG_CIRCUIT_EVOLVE = {
# pop 0
[OP_REMOVAL, OP_CROSSOVER, OP_CROSSOVER],
# pop 1
- [OP_REMOVAL, OP_CROSSOVER, OP_MUTATION],
+ [OP_REMOVAL, OP_MUTATION, OP_MUTATION],
# pop 2
[OP_REMOVAL, OP_CROSSOVER, OP_MUTATION]
],
@@ -158,7 +158,7 @@ CONFIG_CIRCUIT_EVOLVE = {
# 'num_populations': 3,
# how many individuals per population
- 'num_individuals': 24,
+ 'num_individuals': 94,
# start genome lenght for each individual
'genome_length': PHASE0_EXPERIMENT.needed_gene_length()[0],
diff --git a/src/components/config_logical_gate_construction_no_explicit_input_coding.py b/src/components/config_logical_gate_construction_no_explicit_input_coding.py
old mode 100644
new mode 100755
diff --git a/src/components/evolutionary_phase.py b/src/components/evolutionary_phase.py
index 693fb93..020196f 100755
--- a/src/components/evolutionary_phase.py
+++ b/src/components/evolutionary_phase.py
@@ -23,6 +23,14 @@ class Evolutionary_phase:
def completed(self, population_fitness: list, epochs: int) -> bool:
"""
Called to check if the phase has reached its end.
+ You have to return true or false depending on if the pase is finished or not.
+ For example:
+ max_fitness = max(population_fitness)
+ if max_fitness > 90.9:
+ return True
+ if epochs > 900:
+ return True
+ return False
"""
raise NotImplementedError
diff --git a/src/components/genetic_operators.py b/src/components/genetic_operators.py
index 44581bd..58a6498 100755
--- a/src/components/genetic_operators.py
+++ b/src/components/genetic_operators.py
@@ -54,7 +54,6 @@ class Set_zeros(Genetic_operator):
for gene in old_population:
gene[self.indexes] = 0
new_population.append(gene)
- print(len(new_population))
return new_population
class Mutation_all_operator(Genetic_operator):
diff --git a/src/config-Copy1.py b/src/config-Copy1.py
old mode 100644
new mode 100755
diff --git a/src/core/controller.py b/src/core/controller.py
index eb4b74e..5b98dec 100755
--- a/src/core/controller.py
+++ b/src/core/controller.py
@@ -1,6 +1,7 @@
from core.population import Population
from concurrent.futures import ProcessPoolExecutor
import numpy as np
+import time as time
# FIXME:
# implement correct print functions in the classes
# such that print(object) can be called
@@ -304,7 +305,7 @@ class Controller:
# ----------
# init phase
# ----------
- print(f'Evolutionary phase {evolutionary_phase}')
+ print(f'Evolutionary phase: {evolutionary_phase}')
# reset epochs
self.epochs = 0
@@ -328,13 +329,17 @@ class Controller:
# apply genetic operator stack on each population
# NOTE: after applying the stack the fitness values
# will be zero again for all individuals
+ t_op_stack = time.time()
self.__apply_operator_stack()
+ t_op_stack = time.time() - t_op_stack
# instantiate individuals (get them a fitness value)
# tell them the phase index becaus the worker processes
# (mpi and local) will have their instance of the
# phase list and with it know the experiments.
+ t_evaluate = time.time()
self.__instantiate_individuals(phase_index=p_index)
+ t_evaluate = time.time() - t_evaluate
# check if algorithm stagnates
# IMP
@@ -346,8 +351,12 @@ class Controller:
# epoch done
# ----------
# information
- print(
- f'epoch {self.epochs}, Pop max fitness {self.get_population_fitness()}')
+ print(f'epoch {self.epochs}')
+ print('--------------')
+ print(f't_op({t_op_stack})')
+ print(f't_eval({t_evaluate})')
+ print(f'Pop max fitness {self.get_population_fitness()}')
+ print()
# increment counter
self.epochs += 1
diff --git a/src/core/mpi_worker.py b/src/core/mpi_worker.py
index ba3b5a2..3c453da 100755
--- a/src/core/mpi_worker.py
+++ b/src/core/mpi_worker.py
@@ -9,12 +9,13 @@ class MPI_worker():
self.comm = MPI.COMM_WORLD
self.rank = self.comm.Get_rank()
self.sleep_time = sleep_time
+ print(f'initialized MPI worker with rank {self.rank}')
def start(self):
# worker
# loop until stop is signalled
while True:
- # print(f'worker {rank} loop')
+ print(f'worker {self.rank} loop')
# signaling that we are waiting for new work
self.comm.isend(None, dest=0, tag=mpi_tags.IDLE)
--
GitLab