def test(self):
numberOfBins = 4
multiplicity = 2
h = Histogram(numberOfBins, multiplicity)
h.setCurrentToMinimum()
h.accumulate(0, 1)
h.accumulate(1, 2)
h.accumulate(2, 2)
h.accumulate(3, 1)
frameTimes = [0, 1]
recordedSpecies = [0, 1, 2]
x = HistogramFrames(numberOfBins, multiplicity, recordedSpecies)
x.setFrameTimes(frameTimes)
for i in range(2):
x.setCurrentToMinimum()
for i in range(len(frameTimes)):
for j in range(len(recordedSpecies)):
x.histograms[i][j].merge(h)
assert not x.hasErrors()
stream = StringIO()
writer = XmlWriter(stream)
writer.beginDocument()
x.writeXml(writer, 'model', 'method')
writer.endDocument()
def testMoreBasic(self):
# More basic tests.
x = Histogram(8, 2)
x.setCurrentToMinimum()
x.accumulate(4, 1.)
assert x.getMean() == 4
assert x.getUnbiasedVariance() == float('inf')
assert x.size() == 8
assert x.min() == 4.
assert x.max() == 5.
assert x._upperBound() == 8.
def testConstantValueBins(self):
# Basic tests with constant-value bins.
size = 8
values = range(size)
weights = [1.] * size
mean, variance = _computeMeanAndVariance(values, weights)
x = Histogram(size, 2)
x.setCurrentToMinimum()
for i in range(8):
x.accumulate(i, 1.)
self.assertAlmostEqual(x.getMean(), mean)
self.assertAlmostEqual(x.getUnbiasedVariance(), variance)
self.assertEqual(x.size(), 8)
self.assertEqual(x.min(), 0.)
self.assertEqual(x.max(), 8.)
self.assertEqual(x._upperBound(), 8.)
p = x.getProbabilities()
self.assertTrue((p == 1./8.).all())
def testPoisson2(self):
# Poisson with mean 4. Use accumulate.
lam = 4.
size = 20
x = Histogram(size, 2)
x.setCurrentToMinimum()
weight = math.exp(-lam)
for i in range(50):
x.accumulate(i, weight)
weight *= lam / (i + 1)
assert abs(x.getMean() - 4) < 1e-4
assert abs(x.getUnbiasedVariance() - 4) < 1e-1
# Uniform, 20 bins.
size = 20
weights = [1.] * size
cardinality = size
sumOfWeights = sum(weights)
mean = 0.
for i in range(size):
mean += weights[i] * i
mean /= sumOfWeights
summedSecondCenteredMoment = 0.
for i in range(size):
summedSecondCenteredMoment += weights[i] * (i - mean)**2
stream = StringIO(repr(cardinality) + '\n' +
repr(sumOfWeights) + '\n' +
repr(mean) + '\n' +
repr(summedSecondCenteredMoment) + '\n' +
'0\n1\n' +
''.join([str(_x) + ' ' for _x in [1.] * size]) +
'\n' +
'0 ' * size + '\n')
y = Histogram()
y.read(stream, 2)
# Distance between Poisson and uniform.
assert histogramDistance(x, y) > 0
# Distance between Poisson and Poisson.
assert histogramDistance(x, x) == 0
# Uniform, 30 bins.
x.merge(y)
def testAccumulate1(self):
x = Histogram(8, 2)
x.setCurrentToMinimum()
x.accumulate(0, 1)
assert x.size() == 8
assert x.min() == 0.
assert x.max() == 1.
assert x._upperBound() == 8.
x.accumulate(7, 1)
assert x.size() == 8
assert x.min() == 0.
assert x.max() == 8.
assert x._upperBound() == 8.
x.accumulate(8, 1)
assert x.size() == 8
assert x.min() == 0.
assert x.max() == 10.
assert x._upperBound() == 16.
x.accumulate(15, 1)
assert x.size() == 8
assert x.min() == 0.
assert x.max() == 16.
assert x._upperBound() == 16.
def testAccumulate2(self):
x = Histogram(8, 2)
x.setCurrentToMinimum()
x.accumulate(10, 1)
assert x.size() == 8
assert x.min() == 10.
assert x.max() == 11.
assert x._upperBound() == 18.
x.accumulate(5, 1)
assert x.size() == 8
assert x.min() == 5.
assert x.max() == 11.
assert x._upperBound() == 13.
x.accumulate(0, 1)
assert x.size() == 8
assert x.min() == 0.
assert x.max() == 12.
assert x._upperBound() == 16.
# For each sample size.
histogram = Histogram(converged[0].size())
for sample in samples:
# The average run time.
time = 0.
for start in range(0, size, sample):
if start + sample > size:
break
for i in range(start, start + sample):
time += float(lines[i * (numberOfHistograms + 1)])
time /= size // sample
# Compute the average error.
averageError = 0.
# For each recorded species.
for h in range(numberOfHistograms):
error = 0.
for start in range(0, size, sample):
histogram.clear()
if start + sample > size:
break
for i in range(start, start + sample):
n = i * (numberOfHistograms + 1) + h + 1
data = [float(_x) for _x in lines[n].split()]
for j in range(0, len(data), 2):
histogram.accumulate(data[j], data[j + 1], 0)
error += histogramDistance(histogram, converged[h])
error /= size // sample
averageError += error
averageError /= numberOfHistograms
report.write('%s %s\n' % (time, averageError))
converged.read(open('converged.txt', 'r'))
histogram = Histogram(converged.size())
lines = open(directory + '/output.txt', 'r').readlines()
size = len(lines)
report = open(directory + '/error.txt', 'w')
report.write('#time %s\n' % directory)
# Determine the samples.
samples = []
sample = size
while sample > 0:
samples.append(sample)
sample //= 2
samples.sort()
# For each sample size.
for sample in samples:
error = 0.
time = 0.
for start in range(0, size, sample):
histogram.clear()
if start + sample > size:
break
for i in range(start, start + sample):
data = [float(_x) for _x in lines[i].split()]
time += data[0]
for i in range(1, len(data), 2):
histogram.accumulate(data[i], data[i + 1], 0)
error += histogramDistance(histogram, converged)
time /= size // sample
error /= size // sample
report.write('%s %s\n' % (time, error))
