Jungfraujoch/image_analysis/scale_merge/ScaleAndMerge.cpp

// SPDX-FileCopyrightText: 2025 Filip Leonarski, Paul Scherrer Institute <filip.leonarski@psi.ch>
// SPDX-License-Identifier: GPL-3.0-only

#include "ScaleAndMerge.h"

#include <ceres/ceres.h>

#include <algorithm>
#include <cmath>
#include <limits>
#include <stdexcept>
#include <tuple>
#include <unordered_map>
#include <utility>
#include <vector>

namespace {

struct HKLKey {
    int64_t h = 0;
    int64_t k = 0;
    int64_t l = 0;
    bool is_positive = true; // only relevant if opt.merge_friedel == false

    bool operator==(const HKLKey& o) const noexcept {
        return h == o.h && k == o.k && l == o.l && is_positive == o.is_positive;
    }
};

struct HKLKeyHash {
    size_t operator()(const HKLKey& key) const noexcept {
        auto mix = [](uint64_t x) {
            x ^= x >> 33;
            x *= 0xff51afd7ed558ccdULL;
            x ^= x >> 33;
            x *= 0xc4ceb9fe1a85ec53ULL;
            x ^= x >> 33;
            return x;
        };
        const uint64_t a = static_cast<uint64_t>(key.h);
        const uint64_t b = static_cast<uint64_t>(key.k);
        const uint64_t c = static_cast<uint64_t>(key.l);
        const uint64_t d = static_cast<uint64_t>(key.is_positive ? 1 : 0);
        return static_cast<size_t>(mix(a) ^ (mix(b) << 1) ^ (mix(c) << 2) ^ (mix(d) << 3));
    }
};

inline int RoundImageId(float image_number, double rounding_step) {
    if (!(rounding_step > 0.0))
        rounding_step = 1.0;
    const double x = static_cast<double>(image_number) / rounding_step;
    const double r = std::round(x) * rounding_step;
    return static_cast<int>(std::llround(r / rounding_step));
}

inline double SafeSigma(double s, double min_sigma) {
    if (!std::isfinite(s) || s <= 0.0)
        return min_sigma;
    return std::max(s, min_sigma);
}

inline double SafeD(double d) {
    if (!std::isfinite(d) || d <= 0.0)
        return std::numeric_limits<double>::quiet_NaN();
    return d;
}

inline int SafeToInt(int64_t x) {
    if (x < std::numeric_limits<int>::min() || x > std::numeric_limits<int>::max())
        throw std::out_of_range("HKL index out of int range for Gemmi");
    return static_cast<int>(x);
}

// Canonicalize HKL according to Gemmi Reciprocal ASU if space group is provided.
// If merge_friedel==true -> Friedel mates collapse (key.is_positive always true).
// If merge_friedel==false -> keep I+ vs I- separate by key.is_positive.
inline HKLKey CanonicalizeHKLKey(const Reflection& r, const ScaleMergeOptions& opt) {
    HKLKey key{};
    key.h = r.h;
    key.k = r.k;
    key.l = r.l;
    key.is_positive = true;

    // If no SG provided, we can still optionally separate Friedel mates deterministically.
    if (!opt.space_group.has_value()) {
        if (!opt.merge_friedel) {
            const HKLKey neg{-r.h, -r.k, -r.l, true};
            const bool pos = std::tie(key.h, key.k, key.l) >= std::tie(neg.h, neg.k, neg.l);
            if (!pos) {
                key.h = -key.h;
                key.k = -key.k;
                key.l = -key.l;
                key.is_positive = false;
            }
        }
        return key;
    }

    const gemmi::SpaceGroup& sg = *opt.space_group;
    const gemmi::GroupOps gops = sg.operations();
    const gemmi::ReciprocalAsu rasu(&sg);

    const gemmi::Op::Miller in{{SafeToInt(r.h), SafeToInt(r.k), SafeToInt(r.l)}};
    const auto [asu_hkl, sign_plus] = rasu.to_asu_sign(in, gops);

    key.h = asu_hkl[0];
    key.k = asu_hkl[1];
    key.l = asu_hkl[2];
    key.is_positive = opt.merge_friedel ? true : sign_plus;
    return key;
}

struct IntensityResidual {
    IntensityResidual(double Iobs, double sigma, double s2, bool refine_b)
        : Iobs_(Iobs), inv_sigma_(1.0 / sigma), s2_(s2), refine_b_(refine_b) {}

    template <typename T>
    bool operator()(const T* const log_k,
                    const T* const b,
                    const T* const log_Ihkl,
                    T* residual) const {
        const T k = ceres::exp(log_k[0]);
        const T Ihkl = ceres::exp(log_Ihkl[0]);

        T atten = T(1);
        if (refine_b_) {
            // I_pred = k * exp(-B*s^2) * I_hkl
            atten = ceres::exp(-b[0] * T(s2_));
        }

        const T Ipred = k * atten * Ihkl;
        residual[0] = (Ipred - T(Iobs_)) * T(inv_sigma_);
        return true;
    }

    double Iobs_;
    double inv_sigma_;
    double s2_;
    bool refine_b_;
};

} // namespace

ScaleMergeResult ScaleAndMergeReflectionsCeres(const std::vector<Reflection>& observations,
                                              const ScaleMergeOptions& opt) {
    ScaleMergeResult out;

    struct ObsRef {
        const Reflection* r = nullptr;
        int img_id = 0;     // rounded/quantized image id
        int img_slot = -1;  // compact [0..nimg)
        int hkl_slot = -1;  // compact [0..nhkl)
        double s2 = 0.0;
        double sigma = 0.0;
    };

    std::vector<ObsRef> obs;
    obs.reserve(observations.size());

    std::unordered_map<int, int> imgIdToSlot;
    imgIdToSlot.reserve(256);

    std::unordered_map<HKLKey, int, HKLKeyHash> hklToSlot;
    hklToSlot.reserve(observations.size());

    for (const auto& r : observations) {
        const double d = SafeD(r.d);
        if (!std::isfinite(d))
            continue;
        if (!std::isfinite(r.I))
            continue;

        const double sigma = SafeSigma(static_cast<double>(r.sigma), opt.min_sigma);
        const double s2 = 1.0 / (d * d);

        const int img_id = RoundImageId(r.image_number, opt.image_number_rounding);

        int img_slot;
        {
            auto it = imgIdToSlot.find(img_id);
            if (it == imgIdToSlot.end()) {
                img_slot = static_cast<int>(imgIdToSlot.size());
                imgIdToSlot.emplace(img_id, img_slot);
            } else {
                img_slot = it->second;
            }
        }

        int hkl_slot;
        try {
            const HKLKey key = CanonicalizeHKLKey(r, opt);
            auto it = hklToSlot.find(key);
            if (it == hklToSlot.end()) {
                hkl_slot = static_cast<int>(hklToSlot.size());
                hklToSlot.emplace(key, hkl_slot);
            } else {
                hkl_slot = it->second;
            }
        } catch (...) {
            continue; // skip problematic HKL
        }

        ObsRef o;
        o.r = &r;
        o.img_id = img_id;
        o.img_slot = img_slot;
        o.hkl_slot = hkl_slot;
        o.s2 = s2;
        o.sigma = sigma;
        obs.push_back(o);
    }

    const int nimg = static_cast<int>(imgIdToSlot.size());
    const int nhkl = static_cast<int>(hklToSlot.size());

    out.image_scale_k.assign(nimg, 1.0);
    out.image_b_factor.assign(nimg, 0.0);
    out.image_ids.assign(nimg, 0);

    for (const auto& kv : imgIdToSlot) {
        out.image_ids[kv.second] = kv.first;
    }

    std::vector<double> log_k(nimg, 0.0);
    std::vector<double> b(nimg, 0.0);
    std::vector<double> log_Ihkl(nhkl, 0.0);

    // Initialize Ihkl from per-HKL median of observed intensities.
    {
        std::vector<std::vector<double>> per_hkl_I(nhkl);
        for (const auto& o : obs) {
            per_hkl_I[o.hkl_slot].push_back(static_cast<double>(o.r->I));
        }
        for (int h = 0; h < nhkl; ++h) {
            auto& v = per_hkl_I[h];
            if (v.empty()) {
                log_Ihkl[h] = std::log(std::max(opt.min_sigma, 1e-6));
                continue;
            }
            std::nth_element(v.begin(), v.begin() + static_cast<long>(v.size() / 2), v.end());
            double med = v[v.size() / 2];
            if (!std::isfinite(med) || med <= opt.min_sigma)
                med = opt.min_sigma;
            log_Ihkl[h] = std::log(med);
        }
    }

    ceres::Problem problem;

    std::unique_ptr<ceres::LossFunction> loss;
    if (opt.use_huber_loss) {
        loss = std::make_unique<ceres::HuberLoss>(opt.huber_delta);
    }

    for (const auto& o : obs) {
        const double Iobs = static_cast<double>(o.r->I);

        auto* cost = new ceres::AutoDiffCostFunction<IntensityResidual, 1, 1, 1, 1>(
            new IntensityResidual(Iobs, o.sigma, o.s2, opt.refine_b_factor));

        problem.AddResidualBlock(cost,
                                 loss.get(),
                                 &log_k[o.img_slot],
                                 &b[o.img_slot],
                                 &log_Ihkl[o.hkl_slot]);
    }

    // Fix gauge freedom: anchor first image scale to 1.0
    if (opt.fix_first_image_scale && nimg > 0) {
        log_k[0] = 0.0;
        problem.SetParameterBlockConstant(&log_k[0]);
    }

    if (!opt.refine_b_factor) {
        for (int i = 0; i < nimg; ++i) {
            b[i] = 0.0;
            problem.SetParameterBlockConstant(&b[i]);
        }
    } else {
        for (int i = 0; i < nimg; ++i) {
            if (opt.b_min)
                problem.SetParameterLowerBound(&b[i], 0, *opt.b_min);
            if (opt.b_max)
                problem.SetParameterUpperBound(&b[i], 0, *opt.b_max);
        }
    }

    ceres::Solver::Options options;
    options.linear_solver_type = ceres::SPARSE_NORMAL_CHOLESKY;
    options.minimizer_progress_to_stdout = false;
    options.logging_type = ceres::LoggingType::SILENT;
    options.max_num_iterations = opt.max_num_iterations;
    options.max_solver_time_in_seconds = opt.max_solver_time_s;

    ceres::Solver::Summary summary;
    ceres::Solve(options, &problem, &summary);

    for (int i = 0; i < nimg; ++i) {
        out.image_scale_k[i] = std::exp(log_k[i]);
        out.image_b_factor[i] = opt.refine_b_factor ? b[i] : 0.0;
    }

    // Reverse maps for merged output
    std::vector<HKLKey> slotToHKL(nhkl);
    for (const auto& kv : hklToSlot) {
        slotToHKL[kv.second] = kv.first;
    }

    // crude sigma estimate from model residual scatter
    std::vector<int> n_per_hkl(nhkl, 0);
    std::vector<double> ss_per_hkl(nhkl, 0.0);

    for (const auto& o : obs) {
        const int i = o.img_slot;
        const int h = o.hkl_slot;

        const double k = std::exp(log_k[i]);
        const double atten = opt.refine_b_factor ? std::exp(-b[i] * o.s2) : 1.0;
        const double Ihkl = std::exp(log_Ihkl[h]);

        const double Ipred = k * atten * Ihkl;
        const double r = (Ipred - static_cast<double>(o.r->I));

        n_per_hkl[h] += 1;
        ss_per_hkl[h] += r * r;
    }

    out.merged.reserve(nhkl);
    for (int h = 0; h < nhkl; ++h) {
        MergedReflection m{};
        m.h = slotToHKL[h].h;
        m.k = slotToHKL[h].k;
        m.l = slotToHKL[h].l;
        m.I = static_cast<float>(std::exp(log_Ihkl[h]));

        if (n_per_hkl[h] >= 2) {
            const double rms = std::sqrt(ss_per_hkl[h] / static_cast<double>(n_per_hkl[h] - 1));
            m.sigma = static_cast<float>(rms / std::sqrt(static_cast<double>(n_per_hkl[h])));
        } else {
            m.sigma = std::numeric_limits<float>::quiet_NaN();
        }

        out.merged.push_back(m);
    }

    return out;
}