Review Multiobjective compatibility #91
Conversation
Examples fix
# Conflicts: # experiments/karines_experiments/baseline.py # experiments/karines_experiments/plastic.py # pyrevolve/config.py # pyrevolve/evolution/population.py # pyrevolve/experiment_management.py # pyrevolve/revolve_bot/revolve_bot.py fix in mutaion and multi objective
There was a problem hiding this comment.
Environment involved in many of the functions might not be the best approach when environment is just a dictionary of one element, if it is more than one, we can also do this from the experiment file instead of pushing this information deeper into the Revolve system.
| measure, value = line.strip().split(' ') | ||
| behavior_headers.append(measure) | ||
| file_summary.write(measure+'\t') | ||
| behavior_headers.append('velocity') |
There was a problem hiding this comment.
Revert Changes to put the criteria in the behavior_desc_robot_1 text file.
| @@ -56,8 +67,12 @@ def build_headers(path): | |||
| if os.path.isfile(bh_file): | |||
| with open(bh_file) as file: | |||
| for line in file: | |||
| measure, value = line.strip().split(' ') | |||
| file_summary.write(value+'\t') | |||
| if line != 'None': | |||
| @@ -87,4 +102,4 @@ def build_headers(path): | |||
| if 'body' in file: | |||
| id = file.split('.')[0].split('_')[-1] | |||
| file_summary.write(gen+'\t'+id+'\n') | |||
| file_summary.close() | |||
| file_summary.close() | |||
|
|
||
| measures_snapshots_all = sqldf("select * from measures_snapshots_all where fitness IS NOT NULL") | ||
|
|
||
| fail_test = sqldf(paste("select method,run,generation,count(*) as c from measures_snapshots_all group by 1,2,3 having c<",gens," order by 4")) |
There was a problem hiding this comment.
Why is this rearrangement necessary?
| @@ -46,6 +41,97 @@ def handle_swap(genotype): | |||
| return genotype | |||
|
|
|||
|
|
|||
| def handle_addition(genotype, genotype_conf): | |||
There was a problem hiding this comment.
Review difference with development function.
| for subitem in item: | ||
| clause += str(subitem) + ' ' | ||
|
|
||
| clause_is_true = eval(clause) |
There was a problem hiding this comment.
Why eval? Please no!
We will never be able to debug any of this
| # generates clause and rule for each flavor of the letter | ||
| for flavor in range(0, conf.max_clauses): | ||
|
|
||
| grammar[symbol[0]].append([]) |
There was a problem hiding this comment.
why creating an empty list and then extend it? Is it not easier to just create the correct list in the fist place?
|
|
||
| def build_clause(environmental_conditions, logic_operators, max_terms_clause): | ||
|
|
||
| clause = [] |
There was a problem hiding this comment.
this creates a list of statements that later is concatenated in a string that is evaluated and assumes some variables present in the context of the function (namely: "hill", "hot" or any other string present in environmental_conditions).
This is so confusing I don't even know where to start.
| max_clauses=2, | ||
| max_terms_clause=2, | ||
| plastic=False, | ||
| environmental_conditions=['hill'], |
There was a problem hiding this comment.
If you want to add a new condition here, there is nowhere written that you also MUST create this variable inside the early_development function of plasticoding.py
And I don't like this way of doing things. You shouldn't touch multiple files all over the place when you need to configure one thing. And you shouldn't touch files outside of your experimental folder when configuring an experiment.
add connections of inpurs to oscillators
Formally used to review the changes between the multiobjective branch and the development branch.