def _produceSamples(self):
""" Append batch size new samples and evaluate them. """
reuseindices = []
if self.numLearningSteps == 0 or not self.importanceMixing:
[
self._oneEvaluation(self._sample2base(self._produceSample()))
for _ in range(self.batchSize)
]
self._pointers = list(
range(
len(self._allEvaluated) - self.batchSize,
len(self._allEvaluated)))
else:
reuseindices, newpoints = importanceMixing(
list(map(self._base2sample,
self._currentEvaluations)), self._oldpdf,
self._newpdf, self._produceSample, self.forcedRefresh)
[self._oneEvaluation(self._sample2base(s)) for s in newpoints]
self._pointers = ([self._pointers[i] for i in reuseindices] + list(
range(
len(self._allEvaluated) - self.batchSize +
len(reuseindices), len(self._allEvaluated))))
self._allGenSteps.append(self._allGenSteps[-1] + self.batchSize -
len(reuseindices))
self._allPointers.append(self._pointers)
def _produceSamples(self):
""" Append batch size new samples and evaluate them. """
reuseindices = []
if self.numLearningSteps == 0 or not self.importanceMixing:
[self._oneEvaluation(self._sample2base(self._produceSample())) for _ in range(self.batchSize)]
self._pointers = list(range(len(self._allEvaluated)-self.batchSize, len(self._allEvaluated)))
else:
reuseindices, newpoints = importanceMixing(map(self._base2sample, self._currentEvaluations),
self._oldpdf, self._newpdf, self._produceSample, self.forcedRefresh)
[self._oneEvaluation(self._sample2base(s)) for s in newpoints]
self._pointers = ([self._pointers[i] for i in reuseindices]+
range(len(self._allEvaluated)-self.batchSize+len(reuseindices), len(self._allEvaluated)))
self._allGenSteps.append(self._allGenSteps[-1]+self.batchSize-len(reuseindices))
self._allPointers.append(self._pointers)
