Merge pull request #344 from NNPDF/expose-evolve-capi

Expose `grid.evolve` to the C-API

Author: Radonirinaunimi

.github/actions/cache-test-data/action.yml

@@ -0,0 +1,19 @@
+# .github/actions/cache-test-data/action.yml
+
+name: Cache Test Data
+description: Caches and downloads test data
+runs:
+  using: "composite"
+  steps:
+    - name: Cache test data
+      id: cache
+      uses: actions/cache@v4
+      with:
+        path: test-data
+        key: test-data-v21
+    - name: Download test data if cache miss
+      if: steps.cache.outputs.cache-hit != 'true'
+      run: |
+        cd maintainer
+        ./download-test-data.sh
+      shell: bash

.github/workflows/capi.yaml

@@ -23,6 +23,9 @@ jobs:
           cargo cinstall --verbose --prefix=/usr/local/ --libdir=/usr/local/lib
           ldconfig
 
+      - name: Get test data
+        uses: ./.github/actions/cache-test-data
+
       - name: Test C++ examples
         run: |
           cd examples/cpp

.github/workflows/python.yml

@@ -23,19 +23,8 @@ jobs:
       - name: Set up Rust
         run: |
           rustup component add llvm-tools
-
       - name: Get test data
-        id: cache-test-data
-        uses: actions/cache@v4
-        with:
-          path: test-data
-          key: test-data-v20
-      - name: Download test data
-        if: steps.cache-test-data.outputs.cache-hit != 'true'
-        run: |
-          cd maintainer
-          ./download-test-data.sh
-
+        uses: ./.github/actions/cache-test-data
       - name: Test
         env:
           # `-C link-dead-code` is needed to prevent 'warning: XX functions have mismatched data' warnings

.github/workflows/rust.yml

@@ -21,16 +21,7 @@ jobs:
       - uses: actions/checkout@v4
 
       - name: Get test data
-        id: cache-test-data
-        uses: actions/cache@v4
-        with:
-          path: test-data
-          key: test-data-v20
-      - name: Download test data
-        if: steps.cache-test-data.outputs.cache-hit != 'true'
-        run: |
-          cd maintainer
-          ./download-test-data.sh
+        uses: ./.github/actions/cache-test-data
 
       - name: Set RUSTDOCFLAGS
         run: |

Cargo.lock

@@ -14,6 +14,8 @@ PROGRAMS = \
 	convolve-grid-deprecated \
 	convolve-grid \
 	get-subgrids \
+	evolve-grid \
+	evolve-grid-identity \
 	deprecated \
 	display-channels-deprecated \
 	display-channels \
@@ -51,6 +53,12 @@ deprecated: deprecated.cpp
 get-subgrids: get-subgrids.cpp
 	$(CXX) $(CXXFLAGS) $< $(PINEAPPL_DEPS) -o $@
 
+evolve-grid: evolve-grid.cpp
+	$(CXX) $(CXXFLAGS) $< $(LHAPDF_DEPS) $(PINEAPPL_DEPS) -o $@
+
+evolve-grid-identity: evolve-grid-identity.cpp
+	$(CXX) $(CXXFLAGS) $< $(LHAPDF_DEPS) $(PINEAPPL_DEPS) -o $@
+
 display-channels-deprecated: display-channels-deprecated.cpp
 	$(CXX) $(CXXFLAGS) $< $(PINEAPPL_DEPS) -o $@
 

examples/cpp/Makefile

@@ -0,0 +1,257 @@
+#include <LHAPDF/PDF.h>
+#include <pineappl_capi.h>
+
+#include <algorithm>
+#include <cstdint>
+#include <cstdlib>
+#include <cassert>
+#include <cstddef>
+#include <string>
+#include <vector>
+#include <numeric>
+
+// NOTE: Uses the scale of the Grid as the starting scale such that we can use an IDENTITY EKO.
+double FAC0 = 6456.44;
+
+/** @brief This struct can contain arbitrary parameters that need to be passed to Evolution
+ * Operator Callback (`generated_fake_ekos`).
+ */
+struct OperatorParams {
+    std::vector<pineappl_conv_type> conv_types;
+};
+
+std::vector<std::size_t> unravel_index(std::size_t flat_index, const std::vector<std::size_t>& shape) {
+    std::size_t ndim = shape.size();
+    std::vector<std::size_t> coords(ndim);
+
+    for (int i = ndim - 1; i >= 0; --i) {
+        coords[i] = flat_index % shape[i];
+        flat_index /= shape[i];
+    }
+
+    return coords;
+}
+
+extern "C" void generate_fake_ekos(
+    std::size_t op_index,
+    double /*fac1*/,
+    const int* /*pids_in*/,
+    const double* /*x_in*/,
+    const int* /*pids_out*/,
+    const double* /*x_out*/,
+    const std::size_t* eko_shape,
+    double* eko_buffer,
+    void* params_state
+) {
+    // select the type of convolution based on the Operator index
+    OperatorParams* op_params = static_cast<OperatorParams*>(params_state);
+    pineappl_conv_type _ = op_params->conv_types[op_index];
+
+    // NOTE: This has to work because the Evolution Operator is always 4D
+    std::vector<std::size_t> shape(eko_shape, eko_shape + 4);
+    // Compute the length of the flattened shape by multiplying the entries
+    std::size_t flat_len = std::accumulate(shape.begin(),
+        shape.end(), 1, std::multiplies<std::size_t>());
+    for (std::size_t i = 0; i != flat_len; i++) {
+        std::vector<std::size_t> coords = unravel_index(i, shape);
+
+        double delta_ik = (coords[0] == coords[2]) ? 1.0 : 0.0;
+        double delta_jl = (coords[1] == coords[3]) ? 1.0 : 0.0;
+
+        eko_buffer[i] = delta_ik * delta_jl;
+    }
+}
+
+void print_results(std::vector<double> dxsec_grid, std::vector<double> dxsec_fktable) {
+    const int idx_width = 6;
+    const int num_width = 15;
+    const int dif_width = 15;
+
+    // Print headers
+    std::cout << std::setw(idx_width) << "Bin"
+              << std::setw(num_width) << "Grid"
+              << std::setw(num_width) << "FkTable"
+              << std::setw(dif_width) << "reldiff" << std::endl;
+
+    // Print dashed lines
+    std::cout << std::setw(idx_width) << std::string(idx_width - 2, '-')
+              << std::setw(num_width) << std::string(num_width - 2, '-')
+              << std::setw(num_width) << std::string(num_width - 2, '-')
+              << std::setw(dif_width) << std::string(dif_width - 2, '-') << std::endl;
+
+    // Print the data
+    std::cout << std::scientific << std::setprecision(6);
+    for (size_t i = 0; i < dxsec_grid.size(); ++i) {
+        double reldiff = (dxsec_fktable[i] - dxsec_grid[i]) / dxsec_grid[i];
+        std::cout << std::setw(idx_width) << i
+                  << std::setw(num_width) << dxsec_grid[i]
+                  << std::setw(num_width) << dxsec_fktable[i]
+                  << std::setw(dif_width) << reldiff
+                  << std::endl;
+    }
+}
+
+int main() {
+    // TODO: How to get a Grid that can be evolved??
+    std::string filename = "../../test-data/LHCB_WP_7TEV_opt.pineappl.lz4";
+
+    // disable LHAPDF banners to guarantee deterministic output
+    LHAPDF::setVerbosity(0);
+    std::string pdfset = "NNPDF31_nlo_as_0118_luxqed";
+    auto pdf = std::unique_ptr<LHAPDF::PDF>(LHAPDF::mkPDF(pdfset, 0));
+
+    auto xfx = [](int32_t id, double x, double q2, void* pdf) {
+        return static_cast <LHAPDF::PDF*> (pdf)->xfxQ2(id, x, q2);
+    };
+    auto alphas = [](double q2, void* pdf) {
+        return static_cast <LHAPDF::PDF*> (pdf)->alphasQ2(q2);
+    };
+
+    std::vector<LHAPDF::PDF*> pdfs = {pdf.get(), pdf.get()};
+    void** pdf_states = reinterpret_cast<void**>(pdfs.data());
+
+    // read the grid from a file
+    auto* grid = pineappl_grid_read(filename.c_str());
+
+    // Get the PID basis representation
+    pineappl_pid_basis pid_basis = pineappl_grid_pid_basis(grid);
+    assert(pid_basis == PINEAPPL_PID_BASIS_PDG);
+
+    // Get the number of convolutions and their types
+    std::size_t n_convs = pineappl_grid_convolutions_len(grid);
+    std::vector<pineappl_conv_type> conv_types(n_convs);
+    pineappl_grid_conv_types(grid, conv_types.data());
+
+    // Fill the vector of unique convolution types. If the Operators required for the Grid
+    // are the same, then it suffices to only pass ONE single Operator.
+    std::vector<pineappl_conv_type> unique_convs;
+    for (std::size_t i = 0; i != n_convs; i++) {
+        pineappl_conv_type conv = conv_types[i];
+        if (std::find(unique_convs.begin(), unique_convs.end(), conv) == unique_convs.end()) {
+            unique_convs.push_back(conv);
+        }
+    }
+
+    // Get the shape of the evolve info objects
+    std::vector<std::size_t> evinfo_shape(5);
+    // NOTE: The argument of `pineappl_grid_evolve_info_shape` must follow the following orders:
+    //     - `grid`: PineAPPL Grid
+    //     - `order_mask`: array of booleans to mask the order(s) to apply the Evolution to,
+    //                     `nullptr` selects all the orders
+    //     - `evinfo_shape`: placeholder to store the shape of the Evolution Operator
+    pineappl_grid_evolve_info_shape(grid, nullptr, evinfo_shape.data());
+
+    // Get the values of the evolve info parameters. These contain, for example, the
+    // information on the `x`-grid and `PID` used to interpolate the Grid.
+    // NOTE: These are used to construct the Evolution Operator
+    std::vector<double> fac1(evinfo_shape[0]);
+    std::vector<double> frg1(evinfo_shape[1]);
+    std::vector<int> pids_in(evinfo_shape[2]);
+    std::vector<double> x_in(evinfo_shape[3]);
+    std::vector<double> ren1(evinfo_shape[4]);
+    // NOTE: The argument of `pineappl_grid_evolve_info` must follow the following orders:
+    //     - `grid`: PineAPPL Grid
+    //     - `order_mask`: array of booleans to mask the order(s) to apply the Evolution to,
+    //                     `nullptr` selects all the orders
+    // The rest of the arguments are placeholders to store data
+    pineappl_grid_evolve_info(grid, nullptr, fac1.data(),
+        frg1.data(), pids_in.data(), x_in.data(), ren1.data());
+
+    // ------------------ Construct the Operator Info ------------------
+    // The Operator Info is a vector with length `N_conv * N_Q2_slices` whose
+    // elements are `OperatorInfo` objects.
+    std::vector<pineappl_conv_type> convtypes(unique_convs.size());
+    std::vector<pineappl_operator_info> opinfo_slices(unique_convs.size() * fac1.size());
+    for (std::size_t i = 0; i != unique_convs.size(); i++) {
+        for (std::size_t j = 0; j != fac1.size(); j++) {
+            pineappl_operator_info opinfo = {
+                FAC0, // fac0
+                fac1[j], // fac1
+                pid_basis,
+                unique_convs[i],
+            };
+            opinfo_slices[i * fac1.size() + j] = opinfo;
+        }
+        conv_types[i] = unique_convs[i];
+    }
+
+    // ------------------ Construct the Evolution Operator ------------------
+    // Choose a different PID basis for the FK table
+    // std::vector<int> pids_out = {-6, -5, -4, -3, -2, -1, 1, 2, 3, 4, 5, 6, 21, 22};
+    std::vector<int> pids_out = pids_in;
+
+    // Construct the values of alphas table
+    std::vector<double> alphas_table;
+    for (double q2 : ren1) {
+        double alpha = alphas(q2, pdf.get());
+        alphas_table.push_back(alpha);
+    }
+
+    // Construct the Parameters that will get passed to the Callback
+    OperatorParams* op_params = new OperatorParams;
+    op_params->conv_types = convtypes;
+    void* params = static_cast<void*>(op_params);
+
+    std::vector<double> xi = {1.0, 1.0, 1.0};
+    // NOTE: The EKO has to have as shape: (pids_in, x_in, pids_out, x_out)
+    std::vector<std::size_t> tensor_shape = {pids_in.size(), x_in.size(), pids_out.size(), x_in.size()};
+
+    // NOTE: The arguments of `pineappl_grid_evolve` must follow the following orders:
+    //     - `grid`: PineAPPL Grid
+    //     - `nb_slices`: the number of convolution(s)/Evolution Operator(s) required
+    //     - `slices`: callback that returns the evolution operator(s) in slices
+    //     - `operator_info`: operator info
+    //     - `pids_in`: PIDs basis representation of the Grid
+    //     - `x_in`: x-grid of the Grid
+    //     - `pids_out`: PIDs basis representation of the FK table
+    //     - `x_out`: x-grid of the FK table
+    //     - `state`: parameters that get passed to `operator`
+    //     - `order_mask`: array of booleans to mask the order(s) to apply the Evolution to,
+    //                     `nullptr` selects all the orders
+    //     - `xi`: scale variation
+    //     - `ren1`: values of the renormalization scales
+    //     - `alphas_table`: values of alphas for each renormalization scales
+    //     - `eko_shape`: shape of the evolution operators
+    pineappl_grid* fktable = pineappl_grid_evolve(
+        grid,                 // `grid`
+        unique_convs.size(),  // `nb_slices`
+        generate_fake_ekos,   // `slices`
+        opinfo_slices.data(), // `operator_info`
+        pids_in.data(),       // `pids_in`
+        x_in.data(),          // `x_in`
+        pids_out.data(),      // `pids_out`
+        x_in.data(),          // `x_out`
+        tensor_shape.data(),  // `eko_shape`
+        params,               // `state`
+        nullptr,              // `order_mask`
+        xi.data(),            // `xi`
+        ren1.data(),          // `ren1`
+        alphas_table.data()   // `alphas_table`
+    );
+
+    // ------------------ Compare Grid & FK after convolution ------------------
+    // how many bins does this grid have?
+    std::size_t bins = pineappl_grid_bin_count(grid);
+
+    // [ convolve the Grid ]
+    std::vector<double> mu_scales = { 1.0, 1.0, 1.0 };
+    std::vector<double> dxsec_grid(bins);
+    pineappl_grid_convolve(grid, xfx, alphas, pdf_states, pdf.get(),
+        nullptr, nullptr, nullptr, 1,
+        mu_scales.data(), dxsec_grid.data());
+
+    // [ convolve the FK Table ]
+    std::vector<double> dxsec_fktable(bins);
+    auto as_one = [](double /*q2*/, void* /*pdf*/) { return 1.0; };
+    pineappl_grid_convolve(fktable, xfx, as_one, pdf_states, nullptr,
+        nullptr, nullptr, nullptr, 1,
+        mu_scales.data(), dxsec_fktable.data());
+
+    // Print the results
+    print_results(dxsec_grid, dxsec_fktable);
+
+    pineappl_grid_write(fktable, "evolved-grid-identity.pineappl.lz4");
+
+    pineappl_grid_delete(grid);
+    pineappl_grid_delete(fktable);
+}