ParallelFor and ParallelChunks started fresh OS threads on every call, one per chunk, through std::async. There are 57 call sites and several sit inside iterative fits, so one run of the heaviest crystal created 13 497 threads and a 900-frame dataset 7 320. Both now run on a persistent pool. The contracts are unchanged: ParallelChunks keeps the same worker count and the same fixed split, so a reduction sums term for term as before, and ParallelFor keeps stealing per item. Two things in the pool are worth knowing. The caller is one of the hands - it claims its own region's tasks and then waits only on tasks already running - so a region entered from inside another region cannot deadlock at any depth, which a shared-queue pool would. And a task wakes one worker rather than the whole pool: on a large machine notify_all wakes every idle thread to find nothing, once per region, tens of thousands of times a run. Two hand-rolled copies of the same pattern now use it, in FrenchWilson and in the two histogram passes of ComputeAsuGroups. Be clear about what this buys today: nothing measurable. Thread creations drop 13 497 -> 770 and entering a parallel region goes from 1.2-2.2 ms to 112 us, an 11-20x cut, but wall clock on 48 threads is level with before, inside the +-5 % this machine's run-to-run placement is worth. What it removes is a cost that grows with the thread count - measured, entry is linear in it - and the machine this is heading for has four times the threads of the one it was measured on, where the same 335 regions a run would cost about 1.8 s of pure thread creation. ComputeAsuGroups' histogram also changes. It is an nthreads x n_groups table, 936 MB at -N 48 on the heaviest crystal and allocated five times a run, and the prefix over it walked DOWN a column - a 19.5 MB stride, so a cache and TLB miss per step, 234 M of them, serially. Both passes now walk rows and split over group ranges. The counts are integers, so the result is bit-identical. This one is reasoning, not measurement: at 48 threads it sits under this machine's noise and could not be shown either way. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
159 lines
7.4 KiB
C++
159 lines
7.4 KiB
C++
// SPDX-FileCopyrightText: 2026 Filip Leonarski, Paul Scherrer Institute <filip.leonarski@psi.ch>
|
|
// SPDX-License-Identifier: GPL-3.0-only
|
|
|
|
#include "FrenchWilson.h"
|
|
|
|
#include <algorithm>
|
|
#include <cmath>
|
|
#include <limits>
|
|
#include <vector>
|
|
|
|
#include "../../common/ParallelFor.h"
|
|
#include "../../common/ResolutionShells.h"
|
|
#include "gemmi/symmetry.hpp"
|
|
|
|
namespace {
|
|
|
|
struct Posterior {
|
|
double mean_I; // <J> (posterior mean true intensity)
|
|
double mean_F; // <|F|> (posterior mean amplitude)
|
|
};
|
|
|
|
// Posterior moments of the true intensity J >= 0 given a measurement I +/- sigma and the Wilson
|
|
// prior with mean sigma_wilson. Integrated numerically over J in [0, I + 8 sigma] with a log-shift
|
|
// so the exponentials never overflow/underflow. acentric: p(J) ~ exp(-J/S); centric:
|
|
// p(J) ~ exp(-J/2S)/sqrt(J).
|
|
// `logw` is caller-owned scratch of npts doubles (one per worker), so the integration allocates nothing.
|
|
Posterior integrate_posterior(double I, double sigma, double sigma_wilson, bool centric, int npts,
|
|
std::vector<double> &logw) {
|
|
const double inv_2s2 = 1.0 / (2.0 * sigma * sigma);
|
|
// The posterior is the Gaussian likelihood tilted by the exponential prior, so it peaks at
|
|
// I - sigma^2/S and decays over whichever of sigma and S is TIGHTER. Ranging to I + 8 sigma
|
|
// regardless is wrong once sigma greatly exceeds S: with npts fixed the whole prior then falls
|
|
// inside the first grid cell, the quadrature degenerates to that one point and returns
|
|
// F = sqrt(dj/2) with sigmaF -> 0 - i.e. a reflection we know nothing about comes back looking
|
|
// like the best measured one in the file.
|
|
const double prior_scale = centric ? 2.0 * sigma_wilson : sigma_wilson;
|
|
const double peak = std::max(I - sigma * sigma / prior_scale, 0.0);
|
|
const double width = peak > 0.0 ? sigma : std::min(sigma, prior_scale);
|
|
const double j_max = peak + 10.0 * width;
|
|
const double dj = j_max / npts;
|
|
|
|
double max_logw = -std::numeric_limits<double>::infinity();
|
|
for (int i = 0; i < npts; ++i) {
|
|
const double j = (i + 0.5) * dj;
|
|
const double diff = I - j;
|
|
const double log_prior = centric ? (-j / (2.0 * sigma_wilson) - 0.5 * std::log(j))
|
|
: (-j / sigma_wilson);
|
|
logw[i] = log_prior - diff * diff * inv_2s2;
|
|
max_logw = std::max(max_logw, logw[i]);
|
|
}
|
|
|
|
double sum_w = 0, sum_wI = 0, sum_wF = 0;
|
|
for (int i = 0; i < npts; ++i) {
|
|
const double j = (i + 0.5) * dj;
|
|
const double w = std::exp(logw[i] - max_logw);
|
|
if (!std::isfinite(w))
|
|
continue;
|
|
sum_w += w;
|
|
sum_wI += w * j;
|
|
sum_wF += w * std::sqrt(j);
|
|
}
|
|
if (sum_w <= 0.0) {
|
|
const double j = std::max(I, 0.0);
|
|
return {j, std::sqrt(j)};
|
|
}
|
|
return {sum_wI / sum_w, sum_wF / sum_w};
|
|
}
|
|
|
|
} // namespace
|
|
|
|
void ApplyFrenchWilson(std::vector<MergedReflection> &merged, int32_t space_group_number,
|
|
const FrenchWilsonOptions &opts) {
|
|
// Naive amplitude sqrt(max(I,0)) for a missing / strong / untrusted intensity; NaN in -> NaN out
|
|
// (a missing Bijvoet hand stays missing). Fills one (F, sigmaF) pair.
|
|
auto naive_one = [](float I, float sigma, float &F, float &sigF) {
|
|
if (!std::isfinite(I)) { F = NAN; sigF = NAN; return; }
|
|
const double ip = std::max(I, 0.0f);
|
|
F = static_cast<float>(std::sqrt(ip));
|
|
sigF = (ip > 0.0 && std::isfinite(sigma)) ? static_cast<float>(sigma / (2.0 * std::sqrt(ip))) : NAN;
|
|
};
|
|
// The mean intensity and each measured hand share the reflection's Wilson prior, so fill all three.
|
|
auto naive_all = [&](MergedReflection &r) {
|
|
naive_one(r.I, r.sigma, r.F, r.sigmaF);
|
|
naive_one(r.I_plus, r.sigma_plus, r.F_plus, r.sigmaF_plus);
|
|
naive_one(r.I_minus, r.sigma_minus, r.F_minus, r.sigmaF_minus);
|
|
};
|
|
|
|
const gemmi::SpaceGroup *sg = gemmi::find_spacegroup_by_number(space_group_number);
|
|
if (sg == nullptr || merged.empty()) {
|
|
for (auto &r : merged) naive_all(r);
|
|
return;
|
|
}
|
|
const gemmi::GroupOps gops = sg->operations();
|
|
|
|
float d_min = std::numeric_limits<float>::max(), d_max = 0.0f;
|
|
for (const auto &r : merged)
|
|
if (std::isfinite(r.d) && r.d > 0.0f) {
|
|
d_min = std::min(d_min, r.d);
|
|
d_max = std::max(d_max, r.d);
|
|
}
|
|
if (!(d_min < d_max && d_min > 0.0f)) {
|
|
for (auto &r : merged) naive_all(r);
|
|
return;
|
|
}
|
|
|
|
// Wilson mean intensity <I/epsilon> per resolution shell.
|
|
ResolutionShells shells(d_min * 0.999f, d_max * 1.001f, opts.num_shells);
|
|
std::vector<double> shell_sum(opts.num_shells, 0.0);
|
|
std::vector<int> shell_count(opts.num_shells, 0);
|
|
double global_sum = 0.0;
|
|
int global_count = 0;
|
|
auto epsilon = [&](const MergedReflection &r) {
|
|
return std::max(1, gops.epsilon_factor_without_centering({{r.h, r.k, r.l}}));
|
|
};
|
|
for (const auto &r : merged) {
|
|
if (!std::isfinite(r.I) || !std::isfinite(r.sigma) || r.sigma <= 0.0f)
|
|
continue;
|
|
const double i_over_eps = r.I / epsilon(r);
|
|
global_sum += i_over_eps;
|
|
++global_count;
|
|
if (const auto s = shells.GetShell(r.d)) {
|
|
shell_sum[*s] += i_over_eps;
|
|
++shell_count[*s];
|
|
}
|
|
}
|
|
const double global_mean = global_count > 0 ? std::max(global_sum / global_count, 1e-10) : 1.0;
|
|
std::vector<double> shell_mean(opts.num_shells, global_mean);
|
|
for (int s = 0; s < opts.num_shells; ++s)
|
|
if (shell_count[s] >= opts.min_reflections_per_shell)
|
|
shell_mean[s] = std::max(shell_sum[s] / shell_count[s], 1e-10);
|
|
|
|
// French-Wilson |F| for one intensity of reflection r (its mean, or one Bijvoet hand); the shell
|
|
// Wilson prior, epsilon and centric flag are the reflection's, shared by all three.
|
|
auto fw_one = [&](const MergedReflection &r, float I, float sigma, float &F, float &sigF,
|
|
std::vector<double> &logw) {
|
|
if (!std::isfinite(I) || !std::isfinite(sigma) || sigma <= 0.0f) { naive_one(I, sigma, F, sigF); return; }
|
|
// Strong reflections: the FW correction is negligible, <|F|> = sqrt(I).
|
|
if (I > opts.strong_cutoff * sigma) { naive_one(I, sigma, F, sigF); return; }
|
|
const auto s = shells.GetShell(r.d);
|
|
const double sigma_wilson = epsilon(r) * (s ? shell_mean[*s] : global_mean);
|
|
const bool centric = gops.is_reflection_centric({{r.h, r.k, r.l}});
|
|
const Posterior post = integrate_posterior(I, sigma, sigma_wilson, centric,
|
|
opts.integration_points, logw);
|
|
F = static_cast<float>(post.mean_F);
|
|
sigF = static_cast<float>(std::sqrt(std::max(0.0, post.mean_I - post.mean_F * post.mean_F)));
|
|
};
|
|
// Each reflection's amplitudes depend only on itself and the shell priors above, so the loop is
|
|
// data-parallel over contiguous chunks and gives the same result whatever the worker count.
|
|
ParallelChunks(static_cast<int>(merged.size()), std::max(1, opts.num_threads), [&](int lo, int hi) {
|
|
std::vector<double> logw(opts.integration_points);
|
|
for (int i = lo; i < hi; ++i) {
|
|
MergedReflection &r = merged[i];
|
|
fw_one(r, r.I, r.sigma, r.F, r.sigmaF, logw);
|
|
fw_one(r, r.I_plus, r.sigma_plus, r.F_plus, r.sigmaF_plus, logw);
|
|
fw_one(r, r.I_minus, r.sigma_minus, r.F_minus, r.sigmaF_minus, logw);
|
|
}
|
|
});
|
|
}
|