def get_data(self):
colors = {100: "tomato", 500: "dodgerblue", 1000: "springgreen"}
markers = {100: "s", 500: "o", 1000: "D"}
for popsize in colors:
x = []
y = []
for data in self.stats_process.find(
{"parameters.popsize": popsize}):
x.append(get_reference_array()[10] / data['ten'])
y.append(get_reference_array()[len(GOOD_CANDS_LS)] /
data['all'])
plt.scatter(x,
y,
c=[colors[popsize]] * len(x),
marker=markers[popsize],
s=25,
label="popsize=" + str(popsize))
plt.legend(loc=4)
yxis = plt.ylim()
plt.ylim(0.75, 6.5)
xxis = plt.xlim()
plt.xlim(0.75, 11.5)
plt.xticks(np.arange(1, plt.xlim()[1], 1))
plt.yticks(np.arange(1, plt.ylim()[1], 1))
plt.annotate("best GA",
xy=(9.9, 5.82),
xytext=(10, 6),
arrowprops={
"arrowstyle": "->",
"connectionstyle": "arc"
})
def get_data(self):
colors = {100: "tomato", 500: "dodgerblue", 1000: "springgreen"}
markers = {100: "s", 500: "o", 1000: "D"}
for popsize in colors:
x = []
y = []
for data in self.stats_process.find(
{"parameters.popsize": popsize}):
x.append(data['breakouts_avg'])
y.append(get_reference_array()[len(GOOD_CANDS_LS)] /
data['all'])
plt.scatter(x,
y,
c=[colors[popsize]] * len(x),
marker=markers[popsize],
s=30,
label="popsize=" + str(popsize))
plt.legend(loc=0)
yxis = plt.ylim()
plt.ylim(0.75, yxis[1])
plt.yticks(np.arange(1, plt.ylim()[1], 1))
xxis = plt.xlim()
plt.xlim(0, xxis[1])
def get_data(self, parameter):
plt.title(get_pretty_name(parameter), fontname=self.fontname)
data = []
labels = self.stats_process.distinct("parameters." + parameter)
if "-" not in str(labels[0]):
labels.sort()
else:
labels.sort(
key=lambda label: ord(label.split("-")[0][1]) * 10 + float(
label.split("-")[1]) + float(label.split("-")[2]))
for label in labels:
# get the best
best = self.stats_process.find({"parameters." + parameter: label},
sort=[(self.criteria,
pymongo.ASCENDING)])[0]
if self.criteria == "all":
data.append(get_reference_array()[len(GOOD_CANDS_LS)] /
best["all"])
elif self.criteria == "ten":
data.append(get_reference_array()[10] / best["ten"])
'''
# get the average
m_sum = 0.0
npoints = 0.0
num_to_average = 1 # average the num_to_average BEST runs
for item in self.stats_process.find({"parameters." + parameter: label}, {"all": 1}, sort=[("all",pymongo.ASCENDING)]).limit(num_to_average):
m_sum = m_sum + get_reference_array()[len(GOOD_CANDS_LS)] / item["all"]
npoints = npoints + 1
data.append(m_sum / npoints)
'''
pretty_labels = [get_pretty_name(n) for n in labels]
xlocations = np.array(range(len(data))) + 0.5
width = 0.75
plt.bar(xlocations, data, width=width)
plt.xticks(xlocations + width / 2,
pretty_labels,
fontname=self.fontname)
plt.yticks(fontname=self.fontname)
plt.xlim(0, xlocations[-1] + width * 2)
def get_data(self):
x = []
y = []
idx = 1
for data in self.stats_process.find({},
sort=[("ten", pymongo.ASCENDING)]):
x.append(idx)
idx = idx + 1
y.append(get_reference_array()[10] / data["ten"])
plt.errorbar(x, y, linestyle="--", lw=self.lw, color="blue")
plt.annotate("10 materials",
xy=(x[50], y[50]),
xytext=(10, 0),
color="blue",
textcoords='offset points',
ha='left',
va='bottom',
fontname=self.fontname,
fontsize=self.fontsize,
arrowprops=None)
x = []
y = []
idx = 1
for data in self.stats_process.find({},
sort=[("all", pymongo.ASCENDING)]):
x.append(idx)
idx = idx + 1
y.append(get_reference_array()[len(GOOD_CANDS_LS)] / data["all"])
plt.errorbar(x, y, lw=self.lw, color="blue")
plt.annotate("all 20 materials",
xy=(x[100], y[100]),
xytext=(25, 5),
color="blue",
textcoords='offset points',
ha='left',
va='bottom',
fontname=self.fontname,
fontsize=self.fontsize,
arrowprops=None)
def get_score(self, constraint):
m_sum = 0.0
npoints = 0.0
num_to_average = 50000 # average the num_to_average BEST runs
for item in self.stats_process.find(constraint, {
"all": 1
},
sort=[("all", pymongo.ASCENDING)
]).limit(num_to_average):
m_sum = m_sum + get_reference_array()[len(
GOOD_CANDS_LS)] / item["all"]
npoints = npoints + 1
return (m_sum / npoints)
def get_reference_data(self):
x = [0, get_reference_array()[len(GOOD_CANDS_LS)]]
y = [0, len(GOOD_CANDS_LS)]
plt.errorbar(x, y, lw=self.lw, color="black")
plt.annotate("random",
xy=(x[1] * 0.15, y[1] * 0.15),
xytext=(-5, 5),
color="black",
textcoords='offset points',
ha='right',
va='bottom',
fontname=self.fontname,
fontsize=self.fontsize,
arrowprops=None)
'''
from ga_optimization_ternary.database import Stats_Database
from ga_optimization_ternary.utils import get_reference_array
__author__ = "Anubhav Jain"
__copyright__ = "Copyright 2012, The Materials Project"
__version__ = "0.1"
__maintainer__ = "Anubhav Jain"
__email__ = "*****@*****.**"
__date__ = "Oct 4, 2012"
if __name__ == "__main__":
sdb = Stats_Database()
sdb_os = Stats_Database(extension="_OS")
print get_reference_array()[20]
print get_reference_array()[10]
# print headers
params = ["crossover_fnc", "popsize", "selection_overall", "mutation_rate", "elitism_num", "fitness_fnc"]
for param in params:
print param,
print 'ten', 'all', 'half (OS)', 'all (OS)'
for item in sdb._stats_process.find():
unique_key = item['unique_key']
if "eval_fitness_complex_product" in unique_key:
unique_key = unique_key.replace("eval_fitness_complex_product", "eval_fitness_complex_product_oxide_shield")
else:
unique_key = unique_key.replace("eval_fitness_complex", "eval_fitness_complex_oxide_shield")
unique_key = unique_key.replace("eval_fitness_simple", "eval_fitness_simple_oxide_shield")
def get_data(self):
colors = {
"eval_fitness_simple": "dodgerblue",
"eval_fitness_complex": "dodgerblue",
"eval_fitness_complex_product": "dodgerblue"
}
markers = {
"eval_fitness_simple": "o",
"eval_fitness_complex": "o",
"eval_fitness_complex_product": "o"
}
#colors = {"eval_fitness_simple": "tomato", "eval_fitness_complex": "dodgerblue", "eval_fitness_complex_product": "springgreen"}
#markers = {"eval_fitness_simple": "s", "eval_fitness_complex": "o", "eval_fitness_complex_product": "D"}
for ff in colors:
x = []
y = []
for data in self.stats_process.find({"parameters.fitness_fnc":
ff}):
unique_key = data['unique_key']
if "eval_fitness_complex_product" in unique_key:
unique_key = unique_key.replace(
"eval_fitness_complex_product",
"eval_fitness_complex_product_oxide_shield")
else:
unique_key = unique_key.replace(
"eval_fitness_complex",
"eval_fitness_complex_oxide_shield")
unique_key = unique_key.replace(
"eval_fitness_simple",
"eval_fitness_simple_oxide_shield")
data2 = self.stats_process_os.find_one(
{"unique_key": unique_key})
m_idx = len(GOOD_CANDS_LS)
m_idx_OS = len(GOOD_CANDS_OS)
if self.crit == "half":
m_idx = int(len(GOOD_CANDS_LS) / 2)
m_idx_OS = int(len(GOOD_CANDS_OS) / 2)
x.append(get_reference_array()[m_idx] / data[self.crit])
y.append(get_reference_array_OS()[m_idx_OS] / data2[self.crit])
plt.scatter(x,
y,
c=[colors[ff]] * len(x),
marker=markers[ff],
s=25,
label=get_pretty_name(ff))
#plt.legend(loc=4)
yxis = plt.ylim()
topy = 10 if self.crit == "half" else 9
topx = 11.5 if self.crit == "half" else 7
plt.ylim(0.75, topy)
xxis = plt.xlim()
plt.xlim(0.75, topx)
plt.xticks(np.arange(1, plt.xlim()[1], 1))
plt.yticks(np.arange(1, plt.ylim()[1], 1))
self.stats_process = self.db._stats_process
for it in self.stats_process.find():
p = it['parameters']
print("{}\t{}\t{}\t{}\t{}\t{}\t{}").format(
p['popsize'], get_pretty_name(p['selection_fnc']),
get_pretty_name(p['fitness_fnc']),
get_pretty_name(p['crossover_fnc']), p['elitism_num'],
it['ten'], it['all'])
if __name__ == "__main__":
format = "png"
print get_reference_array()[20]
print get_reference_array()[20] / 1547.35
print get_reference_array()[10] / 532.05
if format:
mpl.rcParams['savefig.dpi'] = 160
#LSOSPlot(format=format, crit="all")
#HeatMapPlot(format=format)
#BreakoutPlot(format=format)
#PerformancePlot(format=format)
#ComparisonPlot(format=format)
#ParametersPlot(format=format)
PerformancePlotExclusion(format=format)
#TenAllPlot(format=format)
else:
'''
from ga_optimization_ternary.database import Stats_Database
from ga_optimization_ternary.utils import get_reference_array
__author__ = "Anubhav Jain"
__copyright__ = "Copyright 2012, The Materials Project"
__version__ = "0.1"
__maintainer__ = "Anubhav Jain"
__email__ = "*****@*****.**"
__date__ = "Oct 4, 2012"
if __name__ == "__main__":
sdb = Stats_Database()
sdb_os = Stats_Database(extension="_OS")
print get_reference_array()[20]
print get_reference_array()[10]
# print headers
params = [
"crossover_fnc", "popsize", "selection_overall", "mutation_rate",
"elitism_num", "fitness_fnc"
]
for param in params:
print param,
print 'ten', 'all', 'half (OS)', 'all (OS)'
for item in sdb._stats_process.find():
unique_key = item['unique_key']
if "eval_fitness_complex_product" in unique_key:
unique_key = unique_key.replace(