featomic-torch 0.6.0rc2__py3-none-win_amd64.whl → 0.6.2__py3-none-win_amd64.whl

featomic/torch/_build_versions.py

@@ -2,4 +2,4 @@
 
 
 # version of featomic used when compiling this package
-BUILD_FEATOMIC_VERSION = '0.6.0rc2'
+BUILD_FEATOMIC_VERSION = '0.6.1'

featomic/torch/system.py

@@ -85,22 +85,4 @@ def _is_torch_system(system):
     if not isinstance(system, torch.ScriptObject):
         return False
 
-    torch_version_tuple = tuple(map(int, torch.__version__.split(".")[:2]))
-    if torch_version_tuple >= (2, 1):
-        return system._type().name() == "System"
-
-    # For older torch version, we check that we have the right properties
-    properties = system._properties()
-    if len(properties) != 3:
-        return False
-
-    if properties[0].name != "species":
-        return False
-
-    if properties[1].name != "positions":
-        return False
-
-    if properties[2].name != "cell":
-        return False
-
-    return True
+    return system._type().name() == "System"

featomic/torch/torch-2.1/lib/cmake/featomic_torch/featomic_torch-config-version.cmake

@@ -10,13 +10,13 @@
 # The variable CVF_VERSION must be set before calling configure_file().
 
 
-set(PACKAGE_VERSION "0.6.0")
+set(PACKAGE_VERSION "0.6.2")
 
 if(PACKAGE_VERSION VERSION_LESS PACKAGE_FIND_VERSION)
   set(PACKAGE_VERSION_COMPATIBLE FALSE)
 else()
 
-  if("0.6.0" MATCHES "^([0-9]+)\\.([0-9]+)")
+  if("0.6.2" MATCHES "^([0-9]+)\\.([0-9]+)")
     set(CVF_VERSION_MAJOR "${CMAKE_MATCH_1}")
     set(CVF_VERSION_MINOR "${CMAKE_MATCH_2}")
 
@@ -27,7 +27,7 @@ else()
       string(REGEX REPLACE "^0+" "" CVF_VERSION_MINOR "${CVF_VERSION_MINOR}")
     endif()
   else()
-    set(CVF_VERSION_MAJOR "0.6.0")
+    set(CVF_VERSION_MAJOR "0.6.2")
     set(CVF_VERSION_MINOR "")
   endif()
 

featomic/torch/torch-2.1/lib/cmake/featomic_torch/featomic_torch-config.cmake

@@ -5,7 +5,7 @@ set(REQUIRED_FEATOMIC_VERSION 0.6)
 find_package(featomic ${REQUIRED_FEATOMIC_VERSION} CONFIG REQUIRED)
 
 # use the same version for metatensor_torch as the main CMakeLists.txt
-set(REQUIRED_METATENSOR_TORCH_VERSION 0.6)
+set(REQUIRED_METATENSOR_TORCH_VERSION 0.7)
 find_package(metatensor_torch ${REQUIRED_METATENSOR_TORCH_VERSION} CONFIG REQUIRED)
 
 # We can only load metatensorfeatomic_torch with the same minor version of Torch

featomic/torch/torch-2.1/lib/cmake/featomic_torch/featomic_torch-targets.cmake

@@ -7,7 +7,7 @@ if(CMAKE_VERSION VERSION_LESS "2.8.12")
    message(FATAL_ERROR "CMake >= 2.8.12 required")
 endif()
 cmake_policy(PUSH)
-cmake_policy(VERSION 2.8.12...3.29)
+cmake_policy(VERSION 2.8.12...3.30)
 #----------------------------------------------------------------
 # Generated CMake target import file.
 #----------------------------------------------------------------

featomic/torch/torch-2.2/lib/cmake/featomic_torch/featomic_torch-config-version.cmake

@@ -10,13 +10,13 @@
 # The variable CVF_VERSION must be set before calling configure_file().
 
 
-set(PACKAGE_VERSION "0.6.0")
+set(PACKAGE_VERSION "0.6.2")
 
 if(PACKAGE_VERSION VERSION_LESS PACKAGE_FIND_VERSION)
   set(PACKAGE_VERSION_COMPATIBLE FALSE)
 else()
 
-  if("0.6.0" MATCHES "^([0-9]+)\\.([0-9]+)")
+  if("0.6.2" MATCHES "^([0-9]+)\\.([0-9]+)")
     set(CVF_VERSION_MAJOR "${CMAKE_MATCH_1}")
     set(CVF_VERSION_MINOR "${CMAKE_MATCH_2}")
 
@@ -27,7 +27,7 @@ else()
       string(REGEX REPLACE "^0+" "" CVF_VERSION_MINOR "${CVF_VERSION_MINOR}")
     endif()
   else()
-    set(CVF_VERSION_MAJOR "0.6.0")
+    set(CVF_VERSION_MAJOR "0.6.2")
     set(CVF_VERSION_MINOR "")
   endif()
 

featomic/torch/torch-2.2/lib/cmake/featomic_torch/featomic_torch-config.cmake

@@ -5,7 +5,7 @@ set(REQUIRED_FEATOMIC_VERSION 0.6)
 find_package(featomic ${REQUIRED_FEATOMIC_VERSION} CONFIG REQUIRED)
 
 # use the same version for metatensor_torch as the main CMakeLists.txt
-set(REQUIRED_METATENSOR_TORCH_VERSION 0.6)
+set(REQUIRED_METATENSOR_TORCH_VERSION 0.7)
 find_package(metatensor_torch ${REQUIRED_METATENSOR_TORCH_VERSION} CONFIG REQUIRED)
 
 # We can only load metatensorfeatomic_torch with the same minor version of Torch

featomic/torch/torch-2.2/lib/cmake/featomic_torch/featomic_torch-targets.cmake

@@ -7,7 +7,7 @@ if(CMAKE_VERSION VERSION_LESS "2.8.12")
    message(FATAL_ERROR "CMake >= 2.8.12 required")
 endif()
 cmake_policy(PUSH)
-cmake_policy(VERSION 2.8.12...3.29)
+cmake_policy(VERSION 2.8.12...3.30)
 #----------------------------------------------------------------
 # Generated CMake target import file.
 #----------------------------------------------------------------

featomic/torch/torch-2.3/lib/cmake/featomic_torch/featomic_torch-config-version.cmake

@@ -10,13 +10,13 @@
 # The variable CVF_VERSION must be set before calling configure_file().
 
 
-set(PACKAGE_VERSION "0.6.0")
+set(PACKAGE_VERSION "0.6.2")
 
 if(PACKAGE_VERSION VERSION_LESS PACKAGE_FIND_VERSION)
   set(PACKAGE_VERSION_COMPATIBLE FALSE)
 else()
 
-  if("0.6.0" MATCHES "^([0-9]+)\\.([0-9]+)")
+  if("0.6.2" MATCHES "^([0-9]+)\\.([0-9]+)")
     set(CVF_VERSION_MAJOR "${CMAKE_MATCH_1}")
     set(CVF_VERSION_MINOR "${CMAKE_MATCH_2}")
 
@@ -27,7 +27,7 @@ else()
       string(REGEX REPLACE "^0+" "" CVF_VERSION_MINOR "${CVF_VERSION_MINOR}")
     endif()
   else()
-    set(CVF_VERSION_MAJOR "0.6.0")
+    set(CVF_VERSION_MAJOR "0.6.2")
     set(CVF_VERSION_MINOR "")
   endif()
 

featomic/torch/torch-2.3/lib/cmake/featomic_torch/featomic_torch-config.cmake

@@ -5,7 +5,7 @@ set(REQUIRED_FEATOMIC_VERSION 0.6)
 find_package(featomic ${REQUIRED_FEATOMIC_VERSION} CONFIG REQUIRED)
 
 # use the same version for metatensor_torch as the main CMakeLists.txt
-set(REQUIRED_METATENSOR_TORCH_VERSION 0.6)
+set(REQUIRED_METATENSOR_TORCH_VERSION 0.7)
 find_package(metatensor_torch ${REQUIRED_METATENSOR_TORCH_VERSION} CONFIG REQUIRED)
 
 # We can only load metatensorfeatomic_torch with the same minor version of Torch

featomic/torch/torch-2.3/lib/cmake/featomic_torch/featomic_torch-targets.cmake

@@ -7,7 +7,7 @@ if(CMAKE_VERSION VERSION_LESS "2.8.12")
    message(FATAL_ERROR "CMake >= 2.8.12 required")
 endif()
 cmake_policy(PUSH)
-cmake_policy(VERSION 2.8.12...3.29)
+cmake_policy(VERSION 2.8.12...3.30)
 #----------------------------------------------------------------
 # Generated CMake target import file.
 #----------------------------------------------------------------

featomic/torch/torch-2.4/lib/cmake/featomic_torch/featomic_torch-config-version.cmake

@@ -10,13 +10,13 @@
 # The variable CVF_VERSION must be set before calling configure_file().
 
 
-set(PACKAGE_VERSION "0.6.0")
+set(PACKAGE_VERSION "0.6.2")
 
 if(PACKAGE_VERSION VERSION_LESS PACKAGE_FIND_VERSION)
   set(PACKAGE_VERSION_COMPATIBLE FALSE)
 else()
 
-  if("0.6.0" MATCHES "^([0-9]+)\\.([0-9]+)")
+  if("0.6.2" MATCHES "^([0-9]+)\\.([0-9]+)")
     set(CVF_VERSION_MAJOR "${CMAKE_MATCH_1}")
     set(CVF_VERSION_MINOR "${CMAKE_MATCH_2}")
 
@@ -27,7 +27,7 @@ else()
       string(REGEX REPLACE "^0+" "" CVF_VERSION_MINOR "${CVF_VERSION_MINOR}")
     endif()
   else()
-    set(CVF_VERSION_MAJOR "0.6.0")
+    set(CVF_VERSION_MAJOR "0.6.2")
     set(CVF_VERSION_MINOR "")
   endif()
 

featomic/torch/torch-2.4/lib/cmake/featomic_torch/featomic_torch-config.cmake

@@ -5,7 +5,7 @@ set(REQUIRED_FEATOMIC_VERSION 0.6)
 find_package(featomic ${REQUIRED_FEATOMIC_VERSION} CONFIG REQUIRED)
 
 # use the same version for metatensor_torch as the main CMakeLists.txt
-set(REQUIRED_METATENSOR_TORCH_VERSION 0.6)
+set(REQUIRED_METATENSOR_TORCH_VERSION 0.7)
 find_package(metatensor_torch ${REQUIRED_METATENSOR_TORCH_VERSION} CONFIG REQUIRED)
 
 # We can only load metatensorfeatomic_torch with the same minor version of Torch

featomic/torch/torch-2.4/lib/cmake/featomic_torch/featomic_torch-targets.cmake

@@ -7,7 +7,7 @@ if(CMAKE_VERSION VERSION_LESS "2.8.12")
    message(FATAL_ERROR "CMake >= 2.8.12 required")
 endif()
 cmake_policy(PUSH)
-cmake_policy(VERSION 2.8.12...3.29)
+cmake_policy(VERSION 2.8.12...3.30)
 #----------------------------------------------------------------
 # Generated CMake target import file.
 #----------------------------------------------------------------

featomic/torch/torch-2.5/lib/cmake/featomic_torch/featomic_torch-config-version.cmake

@@ -10,13 +10,13 @@
 # The variable CVF_VERSION must be set before calling configure_file().
 
 
-set(PACKAGE_VERSION "0.6.0")
+set(PACKAGE_VERSION "0.6.2")
 
 if(PACKAGE_VERSION VERSION_LESS PACKAGE_FIND_VERSION)
   set(PACKAGE_VERSION_COMPATIBLE FALSE)
 else()
 
-  if("0.6.0" MATCHES "^([0-9]+)\\.([0-9]+)")
+  if("0.6.2" MATCHES "^([0-9]+)\\.([0-9]+)")
     set(CVF_VERSION_MAJOR "${CMAKE_MATCH_1}")
     set(CVF_VERSION_MINOR "${CMAKE_MATCH_2}")
 
@@ -27,7 +27,7 @@ else()
       string(REGEX REPLACE "^0+" "" CVF_VERSION_MINOR "${CVF_VERSION_MINOR}")
     endif()
   else()
-    set(CVF_VERSION_MAJOR "0.6.0")
+    set(CVF_VERSION_MAJOR "0.6.2")
     set(CVF_VERSION_MINOR "")
   endif()
 

featomic/torch/torch-2.5/lib/cmake/featomic_torch/featomic_torch-config.cmake

@@ -5,7 +5,7 @@ set(REQUIRED_FEATOMIC_VERSION 0.6)
 find_package(featomic ${REQUIRED_FEATOMIC_VERSION} CONFIG REQUIRED)
 
 # use the same version for metatensor_torch as the main CMakeLists.txt
-set(REQUIRED_METATENSOR_TORCH_VERSION 0.6)
+set(REQUIRED_METATENSOR_TORCH_VERSION 0.7)
 find_package(metatensor_torch ${REQUIRED_METATENSOR_TORCH_VERSION} CONFIG REQUIRED)
 
 # We can only load metatensorfeatomic_torch with the same minor version of Torch

featomic/torch/torch-2.5/lib/cmake/featomic_torch/featomic_torch-targets.cmake

@@ -7,7 +7,7 @@ if(CMAKE_VERSION VERSION_LESS "2.8.12")
    message(FATAL_ERROR "CMake >= 2.8.12 required")
 endif()
 cmake_policy(PUSH)
-cmake_policy(VERSION 2.8.12...3.29)
+cmake_policy(VERSION 2.8.12...3.30)
 #----------------------------------------------------------------
 # Generated CMake target import file.
 #----------------------------------------------------------------

featomic/torch/torch-2.6/include/featomic/torch/autograd.hpp

@@ -0,0 +1,58 @@
+// IWYU pragma: private; include "featomic/torch.hpp"
+
+#ifndef FEATOMIC_TORCH_AUTOGRAD_HPP
+#define FEATOMIC_TORCH_AUTOGRAD_HPP
+
+#include <ATen/core/ivalue.h>
+#include <torch/autograd.h>
+
+#include <metatensor/torch.hpp>
+
+#include "featomic/torch/exports.h"
+
+namespace featomic_torch {
+
+/// Custom torch::autograd::Function integrating featomic with torch autograd.
+///
+/// This is a bit more complex than your typical autograd because there is some
+/// impedance mismatch between featomic and torch. Most of it should be taken
+/// care of by the `compute` function below.
+class FEATOMIC_TORCH_EXPORT FeatomicAutograd: public torch::autograd::Function<FeatomicAutograd> {
+public:
+    /// Register a pseudo node in Torch's computational graph going from
+    /// `all_positions` and `all_cell` to the values in `block`; using the
+    /// pre-computed gradients in `block`.
+    ///
+    /// If `all_positions.requires_grad` is True, `block` must have a
+    /// `"positions"` gradient; and `systems_start` should contain the index of
+    /// the first atom of each system in `all_positions`.
+    ///
+    /// If `all_cells.requires_grad` is True, `block` must have a `"cell"`
+    /// gradient, and the block samples must contain a `"stucture"` dimension.
+    ///
+    /// This function returns a vector with one element corresponding to
+    /// `block.values`, which should be left unused. It is only there to make
+    /// sure torch registers a `grad_fn` for the tensors stored inside the
+    /// TensorBlock (the values in the TensorBlock are references to the ones
+    /// returned by this function, so when a `grad_fn` is added to one, it is
+    /// also added to the other).
+    static std::vector<torch::Tensor> forward(
+        torch::autograd::AutogradContext *ctx,
+        torch::Tensor all_positions,
+        torch::Tensor all_cells,
+        torch::IValue systems_start,
+        metatensor_torch::TorchTensorBlock block
+    );
+
+    /// Backward step: get the gradients of some quantity `A` w.r.t. the outputs
+    /// of `forward`; and compute the gradients of the same quantity `A` w.r.t.
+    /// the inputs of `forward` (i.e. cell and positions).
+    static std::vector<torch::Tensor> backward(
+        torch::autograd::AutogradContext *ctx,
+        std::vector<torch::Tensor> grad_outputs
+    );
+};
+
+}
+
+#endif

featomic/torch/torch-2.6/include/featomic/torch/calculator.hpp

@@ -0,0 +1,134 @@
+#ifndef FEATOMIC_TORCH_CALCULATOR_HPP
+#define FEATOMIC_TORCH_CALCULATOR_HPP
+
+#include <torch/script.h>
+
+#include <featomic.hpp>
+#include <metatensor/torch.hpp>
+#include <metatensor/torch/atomistic.hpp>
+
+#include "featomic/torch/exports.h"
+
+namespace featomic_torch {
+class FeatomicAutograd;
+
+class CalculatorHolder;
+using TorchCalculator = torch::intrusive_ptr<CalculatorHolder>;
+
+class CalculatorOptionsHolder;
+using TorchCalculatorOptions = torch::intrusive_ptr<CalculatorOptionsHolder>;
+
+/// Options for a single calculation
+class FEATOMIC_TORCH_EXPORT CalculatorOptionsHolder: public torch::CustomClassHolder {
+public:
+    /// get the current selected samples
+    torch::IValue selected_samples() const {
+        return selected_samples_;
+    }
+    /// Set the selected samples to `selection`.
+    ///
+    /// The `IValue` can be `None` (no selection), an instance of
+    /// `metatensor_torch::TorchLabels`, or an instance of
+    /// `metatensor_torch::TorchTensorMap`.
+    void set_selected_samples(torch::IValue selection);
+
+    /// Get the selected samples in the format used by featomic
+    featomic::LabelsSelection selected_samples_featomic() const;
+
+    /// get the current selected properties
+    torch::IValue selected_properties() const {
+        return selected_properties_;
+    }
+    /// Set the selected properties to `selection`.
+    ///
+    /// The `IValue` can be `None` (no selection), an instance of
+    /// `metatensor_torch::TorchLabels`, or an instance of
+    /// `metatensor_torch::TorchTensorMap`.
+    void set_selected_properties(torch::IValue selection);
+
+    /// Get the selected properties in the format used by featomic
+    featomic::LabelsSelection selected_properties_featomic() const;
+
+    /// get the current selected keys
+    torch::IValue selected_keys() const {
+        return selected_keys_;
+    }
+
+    /// Set the selected properties to `selection`.
+    ///
+    /// The `IValue` can be `None` (no selection), or an instance of
+    /// `metatensor_torch::TorchLabels`.
+    void set_selected_keys(torch::IValue selection);
+
+    /// which gradients to keep in the output of a calculation
+    std::vector<std::string> gradients = {};
+
+private:
+    torch::IValue selected_samples_ = torch::IValue();
+    torch::IValue selected_properties_ = torch::IValue();
+    torch::IValue selected_keys_ = torch::IValue();
+};
+
+/// Custom class holder to store, serialize and load featomic calculators
+/// inside Torch(Script) modules.
+class FEATOMIC_TORCH_EXPORT CalculatorHolder: public torch::CustomClassHolder {
+public:
+    /// Create a new calculator with the given `name` and JSON `parameters`
+    CalculatorHolder(std::string name, std::string parameters):
+        c_name_(std::move(name)),
+        calculator_(c_name_, std::move(parameters))
+    {}
+
+    /// Get the name of this calculator
+    std::string name() const {
+        return calculator_.name();
+    }
+
+    /// Get the name used to register this calculator
+    std::string c_name() const {
+        return c_name_;
+    }
+
+    /// Get the parameters of this calculator
+    std::string parameters() const {
+        return calculator_.parameters();
+    }
+
+    /// Get all radial cutoffs used by this `Calculator`'s neighbors lists
+    std::vector<double> cutoffs() const {
+        return calculator_.cutoffs();
+    }
+
+    /// Run a calculation for the given `systems` using the given options
+    metatensor_torch::TorchTensorMap compute(
+        std::vector<metatensor_torch::System> systems,
+        TorchCalculatorOptions options = {}
+    );
+
+private:
+    std::string c_name_;
+    featomic::Calculator calculator_;
+};
+
+
+/// Register autograd nodes between `system.positions` and `system.cell` for
+/// each of the systems and the values in the `precomputed` TensorMap.
+///
+/// This is an advanced function must users should not need to use.
+///
+/// The autograd nodes `backward()` function will use the gradients already
+/// stored in `precomputed`, meaning that if any of the system's positions
+/// `requires_grad`, `precomputed` must contain `"positions"` gradients.
+/// Similarly, if any of the system's cell `requires_grad`, `precomputed` must
+/// contain `"cell"` gradients.
+///
+/// `forward_gradients` controls which gradients are left inside the TensorMap.
+metatensor_torch::TorchTensorMap FEATOMIC_TORCH_EXPORT register_autograd(
+    std::vector<metatensor_torch::System> systems,
+    metatensor_torch::TorchTensorMap precomputed,
+    std::vector<std::string> forward_gradients
+);
+
+}
+
+#endif

featomic/torch/torch-2.6/include/featomic/torch/exports.h

@@ -0,0 +1,43 @@
+
+#ifndef FEATOMIC_TORCH_EXPORT_H
+#define FEATOMIC_TORCH_EXPORT_H
+
+#ifdef FEATOMIC_TORCH_STATIC_DEFINE
+#  define FEATOMIC_TORCH_EXPORT
+#  define FEATOMIC_TORCH_NO_EXPORT
+#else
+#  ifndef FEATOMIC_TORCH_EXPORT
+#    ifdef featomic_torch_EXPORTS
+        /* We are building this library */
+#      define FEATOMIC_TORCH_EXPORT __declspec(dllexport)
+#    else
+        /* We are using this library */
+#      define FEATOMIC_TORCH_EXPORT __declspec(dllimport)
+#    endif
+#  endif
+
+#  ifndef FEATOMIC_TORCH_NO_EXPORT
+#    define FEATOMIC_TORCH_NO_EXPORT 
+#  endif
+#endif
+
+#ifndef FEATOMIC_TORCH_DEPRECATED
+#  define FEATOMIC_TORCH_DEPRECATED __declspec(deprecated)
+#endif
+
+#ifndef FEATOMIC_TORCH_DEPRECATED_EXPORT
+#  define FEATOMIC_TORCH_DEPRECATED_EXPORT FEATOMIC_TORCH_EXPORT FEATOMIC_TORCH_DEPRECATED
+#endif
+
+#ifndef FEATOMIC_TORCH_DEPRECATED_NO_EXPORT
+#  define FEATOMIC_TORCH_DEPRECATED_NO_EXPORT FEATOMIC_TORCH_NO_EXPORT FEATOMIC_TORCH_DEPRECATED
+#endif
+
+/* NOLINTNEXTLINE(readability-avoid-unconditional-preprocessor-if) */
+#if 0 /* DEFINE_NO_DEPRECATED */
+#  ifndef FEATOMIC_TORCH_NO_DEPRECATED
+#    define FEATOMIC_TORCH_NO_DEPRECATED
+#  endif
+#endif
+
+#endif /* FEATOMIC_TORCH_EXPORT_H */

featomic/torch/torch-2.6/include/featomic/torch/system.hpp

@@ -0,0 +1,112 @@
+#ifndef FEATOMIC_TORCH_SYSTEM_HPP
+#define FEATOMIC_TORCH_SYSTEM_HPP
+
+#include <vector>
+#include <map>
+
+#include <torch/script.h>
+
+#include <featomic.hpp>
+#include <metatensor/torch/atomistic.hpp>
+
+#include "featomic/torch/exports.h"
+
+namespace featomic_torch {
+
+/// Implementation of `featomic::System` using `metatensor_torch::System` as
+/// backing memory for all the data.
+///
+/// This can either be used with the Rust neighbor list implementation; or a set
+/// of pre-computed neighbor lists can be added to the system.
+class FEATOMIC_TORCH_EXPORT SystemAdapter final: public featomic::System {
+public:
+    /// Create a `SystemAdapter` wrapping an existing `metatensor_torch::System`
+    SystemAdapter(metatensor_torch::System system);
+
+    ~SystemAdapter() override = default;
+
+    /// `SystemAdapter` is copy-constructible
+    SystemAdapter(const SystemAdapter&) = default;
+    /// `SystemAdapter` is move-constructible
+    SystemAdapter(SystemAdapter&&) = default;
+
+    /// `SystemAdapter` can be copy-assigned
+    SystemAdapter& operator=(const SystemAdapter&) = default;
+    /// `SystemAdapter` can be move-assigned
+    SystemAdapter& operator=(SystemAdapter&&) = default;
+
+    /*========================================================================*/
+    /*            Functions to implement featomic::System                    */
+    /*========================================================================*/
+
+    /// @private
+    uintptr_t size() const override {
+        return static_cast<uintptr_t>(types_.size(0));
+    }
+
+    /// @private
+    const int32_t* types() const override {
+        return types_.data_ptr<int32_t>();
+    }
+
+    /// @private
+    const double* positions() const override {
+        return positions_.data_ptr<double>();
+    }
+
+    /// @private
+    CellMatrix cell() const override {
+        auto* data = cell_.data_ptr<double>();
+        return CellMatrix{{
+            {{data[0], data[1], data[2]}},
+            {{data[3], data[4], data[5]}},
+            {{data[6], data[7], data[8]}},
+        }};
+    }
+
+    /// @private
+    void compute_neighbors(double cutoff) override;
+
+    /// @private
+    const std::vector<featomic_pair_t>& pairs() const override;
+
+    /// @private
+    const std::vector<featomic_pair_t>& pairs_containing(uintptr_t atom) const override;
+
+    /*========================================================================*/
+    /*                 Functions to re-use pre-computed pairs                 */
+    /*========================================================================*/
+
+    /// Should we copy data to featomic internal data structure and compute the
+    /// neighbor list there? This is set to `true` by default, or `false` if
+    /// a neighbor list has been added with `set_precomputed_pairs`.
+    bool use_native_system() const;
+
+private:
+    // the origin of all the data
+    metatensor_torch::System system_;
+
+    /// atomic types tensor, contiguous and on CPU
+    torch::Tensor types_;
+    /// positions tensor, contiguous, on CPU and with dtype=float64
+    torch::Tensor positions_;
+    /// cell tensor, contiguous, on CPU and with dtype=float64
+    torch::Tensor cell_;
+
+
+    struct PrecomputedPairs {
+        std::vector<featomic_pair_t> pairs_;
+        std::vector<std::vector<featomic_pair_t>> pairs_by_atom_;
+    };
+
+    void set_precomputed_pairs(double cutoff, std::vector<featomic_pair_t> pairs);
+
+    // all precomputed pairs we know about
+    std::map<double, PrecomputedPairs> precomputed_pairs_;
+    // last custom requested by `compute_neighbors`
+    double last_cutoff_ = -1.0;
+};
+
+}
+
+#endif

featomic/torch/torch-2.6/include/featomic/torch.hpp

@@ -0,0 +1,10 @@
+#ifndef FEATOMIC_TORCH_HPP
+#define FEATOMIC_TORCH_HPP
+
+#include "featomic/torch/exports.h" // IWYU pragma: export
+
+#include "featomic/torch/system.hpp" // IWYU pragma: export
+#include "featomic/torch/calculator.hpp" // IWYU pragma: export
+
+
+#endif