The GPU predictors claim their output slot with atomicAdd(counter, 1), so a reflection's position in the array is whatever order the blocks happened to finish in. That position is not private to the predictor. BraggOwnerKey packs it into the owner map as the tie-break between two centres equidistant from a shared pixel - the map's atomicMin is order-independent, but the number it compares is not - and the ingest and post-refine bucket sorts, whose comparators are deliberately not total, resolve their ties by the order they are handed. So two runs of the same binary on the same images integrated a different set of reflections. Measured on a large-cell rotation dataset: 63301112 observations against 63301139, and 89% of the merged intensities differing by more than 1% of themselves, median 1.8%. Single-threaded as well as at -N 48, which is what ruled out thread ordering and pointed here. Order the downloaded list by (h, k, l, delta_phi) before TruncateToOutput, whose own pick is then reproducible as well. hkl is a property of the reflection rather than of the schedule, and delta_phi separates the two rocking solutions one hkl can have. The CPU predictors already emit in hkl order, so the two paths now agree on it. Sorting a 20-byte key and gathering once, rather than sorting the 88-byte reflections in place, keeps this off the clock: on a crystal predicting some 35000 reflections a frame the run measures 52.2 s against 52.3 s before. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011n8riB6X59oRjkrSHzNPAU
212 lines
9.1 KiB
C++
212 lines
9.1 KiB
C++
// SPDX-FileCopyrightText: 2025 Filip Leonarski, Paul Scherrer Institute <filip.leonarski@psi.ch>
|
||
// SPDX-License-Identifier: GPL-3.0-only
|
||
|
||
#include <algorithm>
|
||
|
||
#include "../../common/JFJochMath.h"
|
||
#include "../../common/Logger.h"
|
||
#include "BraggPrediction.h"
|
||
#include "../bragg_integration/SystematicAbsence.h"
|
||
|
||
void BraggPrediction::GrowCapacity(int count) {
|
||
reflections.resize(count);
|
||
max_reflections = count;
|
||
}
|
||
|
||
void BraggPrediction::OrderOutput(int count) {
|
||
order_keys.resize(count);
|
||
for (int i = 0; i < count; i++) {
|
||
const Reflection &r = reflections[i];
|
||
order_keys[i] = {r.h, r.k, r.l, r.delta_phi_deg, i};
|
||
}
|
||
std::sort(order_keys.begin(), order_keys.end(),
|
||
[](const OrderKey &a, const OrderKey &b) {
|
||
if (a.h != b.h) return a.h < b.h;
|
||
if (a.k != b.k) return a.k < b.k;
|
||
if (a.l != b.l) return a.l < b.l;
|
||
return a.delta_phi_deg < b.delta_phi_deg;
|
||
});
|
||
order_scratch.resize(count);
|
||
for (int i = 0; i < count; i++) order_scratch[i] = reflections[order_keys[i].index];
|
||
std::copy(order_scratch.begin(), order_scratch.end(), reflections.begin());
|
||
}
|
||
|
||
int BraggPrediction::TruncateToOutput(int count) {
|
||
if (count <= output_limit)
|
||
return count;
|
||
|
||
// Partiality first, then excitation error: exactly one of the two says anything on each path. On the
|
||
// rotation path dist_ewald is identically zero - BraggPredictionRot picks the rocking coordinate so
|
||
// that 2*S0.p + p.p = 0, i.e. |S| = 1/lambda exactly - so ranking by it there left an hkl-lexicographic
|
||
// prefix rather than the best-recorded reflections. On the still path every partiality is 1, so the
|
||
// ranking falls through to dist_ewald, which is what it has always been.
|
||
std::partial_sort(reflections.begin(), reflections.begin() + output_limit,
|
||
reflections.begin() + count,
|
||
[](const Reflection &a, const Reflection &b) {
|
||
if (a.partiality != b.partiality) return a.partiality > b.partiality;
|
||
if (a.dist_ewald != b.dist_ewald) return a.dist_ewald < b.dist_ewald;
|
||
if (a.h != b.h) return a.h < b.h;
|
||
if (a.k != b.k) return a.k < b.k;
|
||
return a.l < b.l;
|
||
});
|
||
return output_limit;
|
||
}
|
||
|
||
namespace {
|
||
// Number of bandwidth sigmas included in the (radially thickened) Ewald-shell
|
||
// acceptance window. 3σ captures essentially the whole pink-beam smear; matches
|
||
// the conservative end of the mosaicity cutoff used by callers.
|
||
constexpr float kBandwidthCutoffSigmas = 3.0f;
|
||
}
|
||
|
||
BraggPrediction::BraggPrediction(int max_reflections)
|
||
: max_reflections(max_reflections), reflections(max_reflections) {}
|
||
|
||
const std::vector<Reflection> &BraggPrediction::GetReflections() const {
|
||
return reflections;
|
||
}
|
||
|
||
int BraggPrediction::Calc(const DiffractionExperiment &experiment, const CrystalLattice &lattice,
|
||
const BraggPredictionSettings &settings) {
|
||
|
||
const auto geom = experiment.GetDiffractionGeometry();
|
||
const auto det_width_pxl = static_cast<float>(experiment.GetXPixelsNum());
|
||
const auto det_height_pxl = static_cast<float>(experiment.GetYPixelsNum());
|
||
|
||
const float one_over_dmax = 1.0f / settings.high_res_A;
|
||
const float one_over_dmax_sq = one_over_dmax * one_over_dmax;
|
||
|
||
float one_over_wavelength = 1.0f / geom.GetWavelength_A();
|
||
|
||
const Coord Astar = lattice.Astar();
|
||
const Coord Bstar = lattice.Bstar();
|
||
const Coord Cstar = lattice.Cstar();
|
||
const Coord S0 = geom.GetScatteringVector();
|
||
|
||
std::vector<float> rot = geom.GetPoniRotMatrix().transpose().arr();
|
||
|
||
// Precompute detector geometry constants
|
||
float beam_x = geom.GetBeamX_pxl();
|
||
float beam_y = geom.GetBeamY_pxl();
|
||
float det_distance = geom.GetDetectorDistance_mm();
|
||
float pixel_size = geom.GetPixelSize_mm();
|
||
float F = det_distance / pixel_size;
|
||
|
||
const float epsilon = 1e-5f;
|
||
const float s0_sq = S0 * S0;
|
||
const float rad_to_deg = 180.0f / static_cast<float>(PI);
|
||
|
||
int i = 0;
|
||
|
||
for (int h = -settings.max_h; h <= settings.max_h; h++) {
|
||
// Precompute A* h contribution
|
||
const float Ah_x = Astar.x * h;
|
||
const float Ah_y = Astar.y * h;
|
||
const float Ah_z = Astar.z * h;
|
||
|
||
for (int k = -settings.max_k; k <= settings.max_k; k++) {
|
||
// Accumulate B* k contribution
|
||
const float AhBk_x = Ah_x + Bstar.x * k;
|
||
const float AhBk_y = Ah_y + Bstar.y * k;
|
||
const float AhBk_z = Ah_z + Bstar.z * k;
|
||
|
||
for (int l = -settings.max_l; l <= settings.max_l; l++) {
|
||
if (systematic_absence(h, k, l, settings.centering))
|
||
continue;
|
||
|
||
if (i >= max_reflections)
|
||
continue;
|
||
|
||
float recip_x = AhBk_x + Cstar.x * l;
|
||
float recip_y = AhBk_y + Cstar.y * l;
|
||
float recip_z = AhBk_z + Cstar.z * l;
|
||
|
||
float recip_sq = recip_x * recip_x + recip_y * recip_y + recip_z * recip_z;
|
||
if (recip_sq > one_over_dmax_sq)
|
||
continue;
|
||
|
||
float S_x = recip_x + S0.x;
|
||
float S_y = recip_y + S0.y;
|
||
float S_z = recip_z + S0.z;
|
||
float S_len = sqrtf(S_x * S_x + S_y * S_y + S_z * S_z);
|
||
float dist_ewald_sphere = std::fabs(S_len - one_over_wavelength);
|
||
|
||
// Energy bandwidth thickens the Ewald shell radially: at the
|
||
// diffraction condition |S|-1/λ shifts by recip_z·(Δλ/λ), i.e.
|
||
// σ_bw = |recip_z|·bandwidth_sigma (= bλ/2d²). Broaden the acceptance
|
||
// window in quadrature so high-resolution shells (smeared most, ∝1/d²)
|
||
// are not clipped.
|
||
float radial_cutoff = settings.ewald_dist_cutoff;
|
||
if (settings.bandwidth_sigma > 0.0f) {
|
||
const float bw_tol = kBandwidthCutoffSigmas * settings.bandwidth_sigma * std::fabs(recip_z);
|
||
radial_cutoff = std::sqrt(radial_cutoff * radial_cutoff + bw_tol * bw_tol);
|
||
}
|
||
|
||
if (dist_ewald_sphere <= radial_cutoff ) {
|
||
const float s0_p0 = S0.x * recip_x + S0.y * recip_y + S0.z * recip_z;
|
||
const float val = s0_sq * recip_sq - s0_p0 * s0_p0;
|
||
|
||
float delta_phi_deg = NAN;
|
||
if (std::fabs(val) >= epsilon && s0_sq > epsilon) {
|
||
const float a_num = (s0_sq - 0.25f * recip_sq) * recip_sq;
|
||
if (a_num >= 0.0f) {
|
||
const float A = std::sqrt(a_num / val);
|
||
const float B = (A * s0_p0 + 0.5f * recip_sq) / s0_sq;
|
||
|
||
const float p_star_x = A * recip_x - B * S0.x;
|
||
const float p_star_y = A * recip_y - B * S0.y;
|
||
const float p_star_z = A * recip_z - B * S0.z;
|
||
|
||
const float p_star_sq = p_star_x * p_star_x + p_star_y * p_star_y + p_star_z * p_star_z;
|
||
const float denom = std::sqrt(p_star_sq * recip_sq);
|
||
|
||
if (denom >= epsilon) {
|
||
float c = (p_star_x * recip_x + p_star_y * recip_y + p_star_z * recip_z) / denom;
|
||
c = std::fmax(-1.0f, std::fmin(1.0f, c));
|
||
delta_phi_deg = std::acos(c) * rad_to_deg;
|
||
}
|
||
}
|
||
}
|
||
|
||
// Inlined RecipToDector with rot1 and rot2 (rot3 = 0)
|
||
// Apply rotation matrix transpose
|
||
float S_rot_x = rot[0] * S_x + rot[1] * S_y + rot[2] * S_z;
|
||
float S_rot_y = rot[3] * S_x + rot[4] * S_y + rot[5] * S_z;
|
||
float S_rot_z = rot[6] * S_x + rot[7] * S_y + rot[8] * S_z;
|
||
|
||
if (S_rot_z <= 0)
|
||
continue;
|
||
|
||
// Project to detector coordinates
|
||
// Assume detector is along x,y,z coordinates after rotation
|
||
float x = beam_x + F * S_rot_x / S_rot_z;
|
||
float y = beam_y + F * S_rot_y / S_rot_z;
|
||
|
||
if ((x < 0) || (x >= det_width_pxl) || (y < 0) || (y >= det_height_pxl))
|
||
continue;
|
||
|
||
float d = 1.0f / sqrtf(recip_sq);
|
||
reflections[i] = Reflection{
|
||
.h = h,
|
||
.k = k,
|
||
.l = l,
|
||
.delta_phi_deg = delta_phi_deg,
|
||
.predicted_x = x,
|
||
.predicted_y = y,
|
||
.observed_x = NAN,
|
||
.observed_y = NAN,
|
||
.d = d,
|
||
.dist_ewald = dist_ewald_sphere,
|
||
.rlp = 1.0,
|
||
.partiality = 1.0f,
|
||
.zeta = 1.0,
|
||
.image_scale_corr = 1.0
|
||
};
|
||
++i;
|
||
}
|
||
}
|
||
}
|
||
}
|
||
return TruncateToOutput(i);
|
||
}
|