Convert CorrelatedAngleEnergy

CMakeLists.txt

@@ -412,6 +412,7 @@ add_library(libopenmc SHARED
   src/position.cpp
   src/pugixml/pugixml_c.cpp
   src/random_lcg.cpp
+  src/secondary_correlated.cpp
   src/secondary_kalbach.cpp
   src/secondary_nbody.cpp
   src/secondary_uncorrelated.cpp

src/secondary_correlated.cpp

@@ -0,0 +1,240 @@
+#include "secondary_correlated.h"
+
+#include <algorithm> // for copy
+#include <cmath>
+#include <cstddef>   // for size_t
+#include <iterator>  // for back_inserter
+
+#include "hdf5_interface.h"
+#include "xtensor/xarray.hpp"
+#include "xtensor/xview.hpp"
+#include "endf.h"
+#include "random_lcg.h"
+#include "search.h"
+
+namespace openmc {
+
+CorrelatedAngleEnergy::CorrelatedAngleEnergy(hid_t group)
+{
+  // Open incoming energy dataset
+  hid_t dset = open_dataset(group, "energy");
+
+  // Get interpolation parameters
+  xt::xarray<int> temp;
+  read_attribute(dset, "interpolation", temp);
+
+  auto temp_b = xt::view(temp, 0); // view of breakpoints
+  auto temp_i = xt::view(temp, 1); // view of interpolation parameters
+
+  std::copy(temp_b.begin(), temp_b.end(), std::back_inserter(breakpoints_));
+  for (const auto i : temp_i)
+    interpolation_.push_back(int2interp(i));
+  n_region_ = breakpoints_.size();
+
+  // Get incoming energies
+  read_dataset(dset, energy_);
+  std::size_t n_energy = energy_.size();
+  close_dataset(dset);
+
+  // Get outgoing energy distribution data
+  dset = open_dataset(group, "energy_out");
+  std::vector<int> offsets;
+  std::vector<int> interp;
+  std::vector<int> n_discrete;
+  read_attribute(dset, "offsets", offsets);
+  read_attribute(dset, "interpolation", interp);
+  read_attribute(dset, "n_discrete_lines", n_discrete);
+
+  xt::xarray<double> eout;
+  read_dataset(dset, eout);
+  close_dataset(dset);
+
+  // Read angle distributions
+  xt::xarray<double> mu;
+  read_dataset(group, "mu", mu);
+
+  for (int i = 0; i < n_energy; ++i) {
+    // Determine number of outgoing energies
+    int j = offsets[i];
+    int n;
+    if (i < n_energy - 1) {
+      n = offsets[i+1] - j;
+    } else {
+      n = eout.shape()[1] - j;
+    }
+
+    // Assign interpolation scheme and number of discrete lines
+    CorrTable d;
+    d.interpolation = int2interp(interp[i]);
+    d.n_discrete = n_discrete[i];
+
+    // Copy data
+    d.e_out = xt::view(eout, 0, xt::range(j, j+n));
+    d.p = xt::view(eout, 1, xt::range(j, j+n));
+    d.c = xt::view(eout, 2, xt::range(j, j+n));
+
+    // To get answers that match ACE data, for now we still use the tabulated
+    // CDF values that were passed through to the HDF5 library. At a later
+    // time, we can remove the CDF values from the HDF5 library and
+    // reconstruct them using the PDF
+    if (false) {
+      // Calculate cumulative distribution function -- discrete portion
+      for (int k = 0; k < d.n_discrete; ++k) {
+        if (k == 0) {
+          d.c[k] = d.p[k];
+        } else {
+          d.c[k] = d.c[k-1] + d.p[k];
+        }
+      }
+
+      // Continuous portion
+      for (int k = d.n_discrete; k < n; ++k) {
+        if (k == d.n_discrete) {
+          d.c[k] = d.c[k-1] + d.p[k];
+        } else {
+          if (d.interpolation == Interpolation::histogram) {
+            d.c[k] = d.c[k-1] + d.p[k-1]*(d.e_out[k] - d.e_out[k-1]);
+          } else if (d.interpolation == Interpolation::lin_lin) {
+            d.c[k] = d.c[k-1] + 0.5*(d.p[k-1] + d.p[k]) *
+                  (d.e_out[k] - d.e_out[k-1]);
+          }
+        }
+      }
+
+      // Normalize density and distribution functions
+      d.p /= d.c[n - 1];
+      d.c /= d.c[n - 1];
+    }
+
+    for (j = 0; j < n; ++j) {
+      // Get interpolation scheme
+      int interp_mu = std::lround(eout(3, offsets[i] + j));
+
+      // Determine offset and size of distribution
+      int offset_mu = std::lround(eout(4, offsets[i] + j));
+      int m;
+      if (offsets[i] + j + 1 < eout.shape()[1]) {
+        m = std::lround(eout(4, offsets[i]+j+1)) - offset_mu;
+      } else {
+        m = mu.shape()[1] - offset_mu;
+      }
+
+      auto interp = int2interp(interp_mu);
+      auto xs = xt::view(mu, 0, xt::range(offset_mu, offset_mu + m));
+      auto ps = xt::view(mu, 1, xt::range(offset_mu, offset_mu + m));
+      auto cs = xt::view(mu, 2, xt::range(offset_mu, offset_mu + m));
+
+      std::vector<double> x {xs.begin(), xs.end()};
+      std::vector<double> p {ps.begin(), ps.end()};
+      std::vector<double> c {cs.begin(), cs.end()};
+
+      // To get answers that match ACE data, for now we still use the tabulated
+      // CDF values that were passed through to the HDF5 library. At a later
+      // time, we can remove the CDF values from the HDF5 library and
+      // reconstruct them using the PDF
+      Tabular* mudist = new Tabular{x.data(), p.data(), m, interp, c.data()};
+
+      d.angle.emplace_back(mudist);
+    } // outgoing energies
+
+    distribution_.push_back(std::move(d));
+  } // incoming energies
+}
+
+void CorrelatedAngleEnergy::sample(double E_in, double& E_out, double& mu) const
+{
+  // <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< REMOVE THIS <<<<<<<<<<<<<<<<<<<<<<<<<<<<<
+  // Before the secondary distribution refactor, an isotropic polar cosine was
+  // always sampled but then overwritten with the polar cosine sampled from the
+  // correlated distribution. To preserve the random number stream, we keep
+  // this dummy sampling here but can remove it later (will change answers)
+  mu = 2.0*prn() - 1.0;
+  // <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< REMOVE THIS <<<<<<<<<<<<<<<<<<<<<<<<<<<<<
+
+  // Find energy bin and calculate interpolation factor -- if the energy is
+  // outside the range of the tabulated energies, choose the first or last bins
+  auto n_energy_in = energy_.size();
+  int i;
+  double r;
+  if (E_in < energy_[0]) {
+    i = 0;
+    r = 0.0;
+  } else if (E_in > energy_[n_energy_in - 1]) {
+    i = n_energy_in - 2;
+    r = 1.0;
+  } else {
+    i = lower_bound_index(energy_.begin(), energy_.end(), E_in);
+    r = (E_in - energy_[i]) / (energy_[i+1] - energy_[i]);
+  }
+
+  // Sample between the ith and [i+1]th bin
+  int l = r > prn() ? i + 1 : i;
+
+  // Interpolation for energy E1 and EK
+  int n_energy_out = distribution_[i].e_out.size();
+  double E_i_1 = distribution_[i].e_out[0];
+  double E_i_K = distribution_[i].e_out[n_energy_out - 1];
+
+  n_energy_out = distribution_[i+1].e_out.size();
+  double E_i1_1 = distribution_[i+1].e_out[0];
+  double E_i1_K = distribution_[i+1].e_out[n_energy_out - 1];
+
+  double E_1 = E_i_1 + r*(E_i1_1 - E_i_1);
+  double E_K = E_i_K + r*(E_i1_K - E_i_K);
+
+  // Determine outgoing energy bin
+  n_energy_out = distribution_[l].e_out.size();
+  double r1 = prn();
+  double c_k = distribution_[l].c[0];
+  double c_k1;
+  int k;
+  for (k = 0; k < n_energy_out - 2; ++k) {
+    c_k1 = distribution_[l].c[k+1];
+    if (r1 < c_k1) break;
+    c_k = c_k1;
+  }
+
+  // Check to make sure 1 <= k <= NP - 1
+  k = std::max(0, std::min(k, n_energy_out - 2));
+
+  double E_l_k = distribution_[l].e_out[k];
+  double p_l_k = distribution_[l].p[k];
+  if (distribution_[l].interpolation == Interpolation::histogram) {
+    // Histogram interpolation
+    if (p_l_k > 0.0) {
+      E_out = E_l_k + (r1 - c_k)/p_l_k;
+    } else {
+      E_out = E_l_k;
+    }
+
+  } else if (distribution_[l].interpolation == Interpolation::lin_lin) {
+    // Linear-linear interpolation
+    double E_l_k1 = distribution_[l].e_out[k+1];
+    double p_l_k1 = distribution_[l].p[k+1];
+
+    double frac = (p_l_k1 - p_l_k)/(E_l_k1 - E_l_k);
+    if (frac == 0.0) {
+      E_out = E_l_k + (r1 - c_k)/p_l_k;
+    } else {
+      E_out = E_l_k + (std::sqrt(std::max(0.0, p_l_k*p_l_k +
+                        2.0*frac*(r1 - c_k))) - p_l_k)/frac;
+    }
+
+  }
+
+  // Now interpolate between incident energy bins i and i + 1
+  if (l == i) {
+    E_out = E_1 + (E_out - E_i_1)*(E_K - E_1)/(E_i_K - E_i_1);
+  } else {
+    E_out = E_1 + (E_out - E_i1_1)*(E_K - E_1)/(E_i1_K - E_i1_1);
+  }
+
+  // Find correlated angular distribution for closest outgoing energy bin
+  if (r1 - c_k < c_k1 - r1) {
+    mu = distribution_[l].angle[k]->sample();
+  } else {
+    mu = distribution_[l].angle[k + 1]->sample();
+  }
+}
+
+}

src/secondary_correlated.h

@@ -0,0 +1,37 @@
+#ifndef OPENMC_SECONDARY_CORRELATED_H
+#define OPENMC_SECONDARY_CORRELATED_H
+
+#include <vector>
+
+#include "hdf5.h"
+#include "xtensor/xtensor.hpp"
+#include "angle_energy.h"
+#include "endf.h"
+#include "distribution.h"
+
+namespace openmc {
+
+class CorrelatedAngleEnergy : public AngleEnergy {
+public:
+  explicit CorrelatedAngleEnergy(hid_t group);
+  void sample(double E_in, double& E_out, double& mu) const;
+private:
+  struct CorrTable {
+    int n_discrete;
+    Interpolation interpolation;
+    xt::xtensor<double, 1> e_out;
+    xt::xtensor<double, 1> p;
+    xt::xtensor<double, 1> c;
+    std::vector<UPtrDist> angle;
+  };
+
+  int n_region_;
+  std::vector<int> breakpoints_;
+  std::vector<Interpolation> interpolation_;
+  std::vector<double> energy_;
+  std::vector<CorrTable> distribution_;
+};
+
+}
+
+#endif // OPENMC_SECONDARY_CORRELATED_H