AdaptiveMonteCarloUtils Namespace Reference

Functions
Real	computeSTD (const std::vector< Real > &data, const unsigned int &start_index)
	compute the standard deviation of a data vector by only considering values from a specific index. More...
Real	computeMean (const std::vector< Real > &data, const unsigned int &start_index)
	compute the mean of a data vector by only considering values from a specific index. More...
std::vector< std::vector< Real > >	sortInput (const std::vector< std::vector< Real >> &inputs, const std::vector< Real > &outputs, const unsigned int samplessub, const Real subset_prob)
	return input values corresponding to the largest po percentile output values. More...
std::vector< Real >	sortOutput (const std::vector< Real > &outputs, const unsigned int samplessub, const Real subset_prob)
	return the largest po percentile output values. More...
Real	computeMin (const std::vector< Real > &data)
	return the minimum value in a vector. More...
std::vector< Real >	computeVectorABS (const std::vector< Real > &data)
	return the absolute values in a vector. More...
unsigned int	weightedResample (const std::vector< Real > &weights, Real rnd)
	return a resampled vector from a population given a weight vector. More...

Function Documentation

computeMean()

Real AdaptiveMonteCarloUtils::computeMean	(	const std::vector< Real > &	data,
		const unsigned int &	start_index
	)

compute the mean of a data vector by only considering values from a specific index.

Parameters

the	data vector
the	starting index

Definition at line 38 of file AdaptiveMonteCarloUtils.C.

Referenced by AdaptiveImportanceSampler::computeSample(), and computeSTD().

{
  if (data.size() < start_index)
    return 0.0;
  else
    return std::accumulate(data.begin() + start_index, data.end(), 0.0) /
           (data.size() - start_index);
}

computeMin()

Real AdaptiveMonteCarloUtils::computeMin

(

const std::vector< Real > &

data

)

return the minimum value in a vector.

Parameters

the	data vector

Definition at line 80 of file AdaptiveMonteCarloUtils.C.

Referenced by AdaptiveMonteCarloDecision::execute().

{
  return *std::min_element(data.begin(), data.end());
}

computeSTD()

Real AdaptiveMonteCarloUtils::computeSTD	(	const std::vector< Real > &	data,
		const unsigned int &	start_index
	)

compute the standard deviation of a data vector by only considering values from a specific index.

Parameters

the	data vector
the	starting index

Definition at line 21 of file AdaptiveMonteCarloUtils.C.

Referenced by AdaptiveImportanceSampler::computeSample().

{
  if (data.size() < start_index)
    return 0.0;
  else
  {
    const Real mean = computeMean(data, start_index);
    const Real sq_diff =
        std::accumulate(data.begin() + start_index,
                        data.end(),
                        0.0,
                        [&mean](Real x, Real y) { return x + (y - mean) * (y - mean); });
    return std::sqrt(sq_diff / (data.size() - start_index));
  }
}

computeVectorABS()

std::vector< Real > AdaptiveMonteCarloUtils::computeVectorABS

(

const std::vector< Real > &

data

)

return the absolute values in a vector.

Parameters

the	data vector

Definition at line 86 of file AdaptiveMonteCarloUtils.C.

Referenced by AdaptiveMonteCarloDecision::execute(), and ParallelSubsetSimulation::sampleSetUp().

{
  std::vector<Real> data_abs(data.size());
  for (unsigned int i = 0; i < data.size(); ++i)
    data_abs[i] = std::abs(data[i]);
  return data_abs;
}

sortInput()

std::vector< std::vector< Real > > AdaptiveMonteCarloUtils::sortInput	(	const std::vector< std::vector< Real >> &	inputs,
		const std::vector< Real > &	outputs,
		const unsigned int	samplessub,
		const Real	subset_prob
	)

return input values corresponding to the largest po percentile output values.

FOR PARALLEL SUBSET SIMULATION SAMPLER ********

Parameters

the	input vector
the	output vector
the	number of samples per subset
the	subset index
the	subset intermediate failure probability

Definition at line 48 of file AdaptiveMonteCarloUtils.C.

Referenced by AdaptiveMonteCarloDecision::execute(), and ParallelSubsetSimulation::sampleSetUp().

{
  std::vector<size_t> ind;
  Moose::indirectSort(outputs.begin(), outputs.end(), ind);

  std::vector<std::vector<Real>> out(inputs.size(), std::vector<Real>(samplessub * subset_prob));
  const size_t offset = std::ceil(samplessub * (1 - subset_prob));
  for (const auto & j : index_range(out))
    for (const auto & i : index_range(out[j]))
      out[j][i] = inputs[j][ind[i + offset]];

  return out;
}

sortOutput()

std::vector< Real > AdaptiveMonteCarloUtils::sortOutput	(	const std::vector< Real > &	outputs,
		const unsigned int	samplessub,
		const Real	subset_prob
	)

return the largest po percentile output values.

FOR PARALLEL SUBSET SIMULATION SAMPLER ********

Parameters

the	output vector
the	number of samples per subset
the	subset index
the	subset intermediate failure probability

Definition at line 66 of file AdaptiveMonteCarloUtils.C.

Referenced by AdaptiveMonteCarloDecision::execute().

{
  std::vector<size_t> ind;
  Moose::indirectSort(outputs.begin(), outputs.end(), ind);

  std::vector<Real> out(samplessub * subset_prob);
  const size_t offset = std::round(samplessub * (1 - subset_prob));
  for (const auto & i : index_range(out))
    out[i] = outputs[ind[i + offset]];

  return out;
}

weightedResample()

unsigned int AdaptiveMonteCarloUtils::weightedResample	(	const std::vector< Real > &	weights,
		Real	rnd
	)

return a resampled vector from a population given a weight vector.

Parameters

the	given inputs (population)
the	weight vector
the	number of dimensions
a	random number between 0 and 1

Definition at line 95 of file AdaptiveMonteCarloUtils.C.

Referenced by IndependentMHDecision::nextSamples().

{
  unsigned int req_index = 0;
  for (unsigned int i = 0; i < weights.size(); ++i)
  {
    if (rnd < weights[i])
    {
      req_index = i;
      break;
    }
    rnd -= weights[i];
  }
  return req_index;
}