Files
Jungfraujoch/image_analysis/spot_finding/SpotUtils.cpp
T
leonarski_fandClaude Opus 5 54adcaafcc Estimate the resolution the merged data will reach, not the furthest spot found
The per-image resolution estimate was the 5th percentile of the spot d spacings -
an extreme order statistic, so it measured where detection stops rather than how
well the crystal diffracts. A large cell puts more reflections past the same
threshold and scored better than a small cell that diffracts further; intensity
was not used at all, so a weak crystal padded with spurious high-resolution
detections ran away; and nothing clamped the answer to what the detector can
deliver. Against the resolution the merged data actually reach it was 42% out in
log-RMS, with 1 of 38 rotation datasets inside 0.2 A.

Take instead the 1/d^2 beyond which 30% of the sum of sqrt(I) over the image's
non-ice spots lies, report 1/(2.25 sqrt of it), clamp at the detector corner, and
take the median over images. A quantile from the middle of the distribution
measures the shape of the falloff - the crystal's own exp(-B/2d^2) - where an
extreme one measures the threshold. sqrt(I) is the Poisson significance of a
summed photon count, so a marginal high-resolution detection cannot carry the
answer and neither can a handful of very strong low-resolution reflections. The
2.25 is the multiplicity gain: merging keeps measuring intensities a fixed factor
in 1/d past the point where a single frame detects them.

Spearman 0.881 -> 0.954, log-RMS 42% -> 8.9%, median error 0.79 -> 0.07 A, and 32
of 38 within 0.2 A. Both constants sit on a broad plateau, the scale is stable
across dataset halves and across resolution ranges, and no second predictor
survives leave-one-out. The residual is around 9%, set by multiplicity, symmetry
and radiation damage - none of which a spot list can see.

The estimate feeds only reporting: the image stream, HDF5, the plots, the scan
result and the preview ring. It sets no cutoff and no search limit, and the
scaling and merging output is byte-identical.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016NNnL26LAvruQ9eLUUWvrJ
2026-08-24 18:50:50 +02:00

244 lines
11 KiB
C++

// SPDX-FileCopyrightText: 2025 Filip Leonarski, Paul Scherrer Institute <filip.leonarski@psi.ch>
// SPDX-License-Identifier: GPL-3.0-only
#include "../../common/JFJochMath.h"
#include "SpotUtils.h"
#include "../../common/ResolutionShells.h"
void CountSpots(DataMessage &msg,
const std::vector<SpotToSave> &spots,
float d_min_A) {
int64_t low_res = 0;
int64_t ice_ring = 0;
for (auto &s: spots) {
if (s.ice_ring)
ice_ring++;
if (s.d_A > d_min_A)
low_res++;
}
msg.spot_count = spots.size();
msg.spot_count_low_res = low_res;
msg.spot_count_ice_rings = ice_ring;
}
// Spots in the ice-free control flanks either side of the hexagonal rings, rescaled to the ring bands'
// own q width. The control for one ring is the two intervals [w, 2w) beside it - same total width as
// the ring band, and symmetric, so the fall-off of spot density with resolution cancels to first
// order. A flank that lands on another ring is not a control and is dropped, its width with it; the
// three rings at 1.947/1.916/1.882 A are 0.05-0.06 apart in q and usually lose both.
float CountIceRingControlSpots(const std::vector<SpotToSave> &spots, float w) {
if (!(w > 0.0f))
return 0.0f;
float control = 0.0f;
for (const float d : ICE_RING_RES_A) {
const float q_ring = 2 * PI / d;
bool lo_free = true, hi_free = true;
for (const float other : ICE_RING_RES_A) {
const float q_other = 2 * PI / other;
if (q_other > q_ring && q_other < q_ring + 3 * w) hi_free = false;
if (q_other < q_ring && q_other > q_ring - 3 * w) lo_free = false;
}
const int free_flanks = (lo_free ? 1 : 0) + (hi_free ? 1 : 0);
if (free_flanks == 0)
continue;
int64_t n = 0;
for (const auto &s: spots) {
if (!(s.d_A > 0.0f)) continue;
const float dq = 2 * PI / s.d_A - q_ring;
if (hi_free && dq >= w && dq < 2 * w) n++;
if (lo_free && dq <= -w && dq > -2 * w) n++;
}
// One free flank covers half the ring band's width, so it counts double.
control += static_cast<float>(n) * 2.0f / static_cast<float>(free_flanks);
}
return control;
}
void MarkIceRings(std::vector<SpotToSave> &spots, float tolerance_q_recipA) {
std::vector<float> ice_rings_q;
for (const auto &i: ICE_RING_RES_A)
ice_rings_q.push_back(2 * PI / i);
for (auto &s: spots) {
auto spot_q = 2 * PI / s.d_A;
bool tmp = false;
for (const auto &q: ice_rings_q)
tmp |= (fabs(spot_q - q) < tolerance_q_recipA);
s.ice_ring = tmp;
}
}
void FilterSpotsByCount(std::vector<SpotToSave> &input, int64_t count, bool deprioritise_ice) {
size_t output_size = std::min<size_t>(input.size(), count);
std::ranges::partial_sort(input, input.begin() + output_size,
std::ranges::less{}, // comparator on the projected key
[deprioritise_ice](const SpotToSave &s) {
// projection: non-ice first (false < true), then strongest intensity
// first. Where the run has no measurable ice the flag marks ordinary
// reflections that happen to lie in the fixed bands, so ordering on it
// would discard a fifth of the strongest spots for nothing.
return std::tuple{deprioritise_ice && s.ice_ring, -s.intensity};
});
input.resize(output_size);
}
void FilterSpuriousHighResolutionSpots(std::vector<SpotToSave> &spots, float threshold) {
std::ranges::sort(spots, [](SpotToSave &a, SpotToSave &b) {
return a.d_A > b.d_A;
});
// Apply 1/d gap threshold: find first gap in q = 1/d exceeding dist_threshold and ignore spots after it
if (spots.size() >= 2 && threshold > 0.0f) {
size_t cut_index = spots.size(); // default: keep all
// d_A sorted descending → q = 1/d_A sorted ascending
// We check consecutive q gaps: Δq_i = (1/d_i) - (1/d_{i+1})
for (size_t i = 0; i + 1 < spots.size(); ++i) {
float d1 = spots[i].d_A;
float d2 = spots[i + 1].d_A;
// Avoid division by zero; d_A should be > 0 in valid data
if (d1 <= 0.0f || d2 <= 0.0f)
continue;
float q1 = 2 * PI / d1;
float q2 = 2 * PI / d2;
float dq = q2 - q1; // should be >= 0 due to sorting
if (dq > threshold) {
cut_index = i + 1; // keep up to i inclusive
break;
}
}
if (cut_index < spots.size())
spots.resize(cut_index);
}
}
namespace {
// Fraction of the image's weighted spot signal that is allowed to lie beyond the quantile read
// off below. A quantile near the middle of the distribution measures the shape of the fall-off,
// which is the crystal's own; the extreme end of it measures the detection threshold and how many
// reflections the unit cell puts on the frame, which are not.
constexpr float SPOT_RESOLUTION_TAIL_FRACTION = 0.30f;
// How much further in 1/d the merged data reach than that quantile. Merging averages many
// observations of each reflection, so intensities go on being measurable well past the point where
// one image's spot finder still detects them. Calibrated on rotation data against the resolution at
// which per-shell CC1/2 falls through 0.30.
constexpr float SPOT_RESOLUTION_MERGE_REACH = 2.25f;
// Fewer spots than this and the quantile is not a fall-off, it is a handful of points.
constexpr size_t SPOT_RESOLUTION_MIN_SPOTS = 4;
}
std::optional<float> GetResolution(const std::vector<SpotToSave> &spots, float detector_d_min_A) {
// Each spot enters weighted by its own signal-to-noise. The intensity is a summed photon count, so
// it is Poisson and its significance is sqrt(I): that keeps a marginal high-resolution detection
// from counting for as much as a real reflection, without letting the handful of very strong
// low-resolution reflections - which say nothing about how far the crystal diffracts - decide the
// answer, as weighting by intensity itself would.
std::vector<std::pair<float, float>> spot_1_over_d2_weight; // (1/d^2, sqrt(intensity))
spot_1_over_d2_weight.reserve(spots.size());
float total_weight = 0.0f;
for (const auto &spot: spots) {
if (spot.ice_ring || !(spot.d_A > 0.0f) || !(spot.intensity > 0.0f))
continue;
const float weight = std::sqrt(spot.intensity);
spot_1_over_d2_weight.emplace_back(1.0f / (spot.d_A * spot.d_A), weight);
total_weight += weight;
}
if (spot_1_over_d2_weight.size() < SPOT_RESOLUTION_MIN_SPOTS || !(total_weight > 0.0f))
return std::nullopt;
// Walk in from the highest-resolution spot until the tail fraction of the weight is behind us.
std::ranges::sort(spot_1_over_d2_weight, std::ranges::greater{},
[](const std::pair<float, float> &s) { return s.first; });
float walked = 0.0f;
float one_over_d2 = spot_1_over_d2_weight.front().first;
for (const auto &[s, weight]: spot_1_over_d2_weight) {
walked += weight;
one_over_d2 = s;
if (walked >= SPOT_RESOLUTION_TAIL_FRACTION * total_weight)
break;
}
const float d_A = 1.0f / (SPOT_RESOLUTION_MERGE_REACH * std::sqrt(one_over_d2));
// However far the crystal diffracts, no merge reaches past the corner of the detector.
return detector_d_min_A > 0.0f ? std::max(d_A, detector_d_min_A) : d_A;
}
void GenerateSpotPlot(DataMessage &msg, const std::vector<SpotToSave> &spots, float d_min_A) {
const int nshells = 20;
// The geometry gives no usable high-resolution corner (no distance or no wavelength), so there is
// no resolution axis to plot the spots against. ResolutionShells would throw on it, once per image.
if (d_min_A <= 0.0f || d_min_A >= 50.0f)
return;
ResolutionShells shells(d_min_A, 50.0, nshells);
std::vector<float> intensity(nshells);
std::vector<float> count(nshells);
for (const auto &s: spots) {
if (s.ice_ring)
continue;
if (auto shell = shells.GetShell(s.d_A)) {
intensity[*shell] += s.intensity;
count[*shell] += 1.0f;
}
}
std::vector<float> result(nshells);
for (int i = 0; i < nshells; ++i) {
if (count[i] > 0)
result[i] = intensity[i] / count[i];
else
result[i] = 0.0f;
}
msg.spot_plot_one_over_d_square = shells.GetShellMeanOneOverResSq();
msg.spot_plot_intensity = result;
msg.spot_plot_count = count;
}
void SpotAnalyze(const DiffractionExperiment &experiment,
const SpotFindingSettings &spot_finding_settings,
const std::vector<DiffractionSpot> &spots,
DataMessage &output) {
auto geom = experiment.GetDiffractionGeometry();
std::vector<SpotToSave> spots_out;
for (const auto &spot: spots) {
if (auto s = spot.Export(geom, output.number); s.has_value())
spots_out.push_back(s.value());
}
if (spot_finding_settings.high_res_gap_Q_recipA.has_value())
FilterSpuriousHighResolutionSpots(spots_out, spot_finding_settings.high_res_gap_Q_recipA.value());
if (experiment.GetDatasetSettings().IsDetectIceRings() && spot_finding_settings.ice_ring_width_Q_recipA > 0.0f) {
MarkIceRings(spots_out, spot_finding_settings.ice_ring_width_Q_recipA);
// Before FilterSpotsByCount below, which orders ice spots LAST and would throw them away first.
output.spot_count_ice_control =
CountIceRingControlSpots(spots_out, spot_finding_settings.ice_ring_width_Q_recipA);
}
CountSpots(output, spots_out, spot_finding_settings.cutoff_spot_count_low_res);
// 0 spells "no limit" everywhere else the limit is read (value_or(0) then compares against it), so it
// has to mean the same here - passing it on as a resolution makes ResolutionShells throw per image.
const auto &spot_d_min = spot_finding_settings.high_resolution_limit;
GenerateSpotPlot(output, spots_out,
spot_d_min.value_or(0.0f) > 0 ? *spot_d_min : experiment.GetDetectorMaxResolution_A());
output.resolution_estimate = GetResolution(spots_out, experiment.GetDetectorMaxResolution_A());
// One decision drives both: if indexing is to use the ice-band spots, the spot budget must not
// throw them away before it gets the chance.
FilterSpotsByCount(spots_out, experiment.GetMaxSpotCount(),
!experiment.GetIndexingSettings().GetIndexIceRings());
output.spots = spots_out;
}