TensorBase.h

// Copyright 2024, UChicago Argonne, LLC
// All Rights Reserved
// Software Name: NEML2 -- the New Engineering material Model Library, version 2
// By: Argonne National Laboratory
// OPEN SOURCE LICENSE (MIT)
//
// Permission is hereby granted, free of charge, to any person obtaining a copy
// of this software and associated documentation files (the "Software"), to deal
// in the Software without restriction, including without limitation the rights
// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
// copies of the Software, and to permit persons to whom the Software is
// furnished to do so, subject to the following conditions:
//
// The above copyright notice and this permission notice shall be included in
// all copies or substantial portions of the Software.
//
// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
// THE SOFTWARE.
 
#pragma once
 
#include "neml2/misc/utils.h"
 
namespace neml2
{
// Forward declarations
template <class Derived>
class TensorBase;
 
class Tensor;
 
template <class Derived>
class TensorBase : public torch::Tensor
{
public:
  TensorBase() = default;
 
  TensorBase(const torch::Tensor & tensor, Size batch_dim);
 
  TensorBase(const torch::Tensor & tensor, const TraceableTensorShape & batch_shape);
 
  template <class Derived2>
  TensorBase(const TensorBase<Derived2> & tensor);
 
  TensorBase(Real) = delete;
 
  [[nodiscard]] static Derived empty_like(const Derived & other);
  [[nodiscard]] static Derived zeros_like(const Derived & other);
  [[nodiscard]] static Derived ones_like(const Derived & other);
  [[nodiscard]] static Derived full_like(const Derived & other, Real init);
 
  [[nodiscard]] static Derived
  linspace(const Derived & start, const Derived & end, Size nstep, Size dim = 0);
  [[nodiscard]] static Derived
  logspace(const Derived & start, const Derived & end, Size nstep, Size dim = 0, Real base = 10);
 
  Derived clone(torch::MemoryFormat memory_format = torch::MemoryFormat::Contiguous) const;
  Derived detach() const;
  using torch::Tensor::detach_;
  Derived to(const torch::TensorOptions & options) const;
  using torch::Tensor::copy_;
  using torch::Tensor::zero_;
  using torch::Tensor::requires_grad;
  using torch::Tensor::requires_grad_;
  Derived operator-() const;
 
  using torch::Tensor::options;
  using torch::Tensor::scalar_type;
  using torch::Tensor::device;
  using torch::Tensor::dim;
  using torch::Tensor::sizes;
  using torch::Tensor::size;
  bool batched() const;
  Size batch_dim() const;
  Size base_dim() const;
  const TraceableTensorShape & batch_sizes() const;
  TraceableSize batch_size(Size index) const;
  TensorShapeRef base_sizes() const;
  Size base_size(Size index) const;
  Size base_storage() const;
 
  using torch::Tensor::index;
  using torch::Tensor::index_put_;
  Derived batch_index(indexing::TensorIndicesRef indices) const;
  neml2::Tensor base_index(indexing::TensorIndicesRef indices) const;
  void batch_index_put_(indexing::TensorIndicesRef indices, const torch::Tensor & other);
  void batch_index_put_(indexing::TensorIndicesRef indices, Real v);
  void base_index_put_(indexing::TensorIndicesRef indices, const torch::Tensor & other);
  void base_index_put_(indexing::TensorIndicesRef indices, Real v);
  Derived variable_data() const;
 
  Derived batch_expand(const TraceableTensorShape & batch_shape) const;
  Derived batch_expand(const TraceableSize & batch_size, Size dim) const;
  neml2::Tensor base_expand(TensorShapeRef base_shape) const;
  neml2::Tensor base_expand(Size base_size, Size dim) const;
  template <class Derived2>
  Derived batch_expand_as(const Derived2 & other) const;
  template <class Derived2>
  Derived2 base_expand_as(const Derived2 & other) const;
  Derived batch_expand_copy(const TraceableTensorShape & batch_shape) const;
  neml2::Tensor base_expand_copy(TensorShapeRef base_shape) const;
  Derived batch_reshape(const TraceableTensorShape & batch_shape) const;
  neml2::Tensor base_reshape(TensorShapeRef base_shape) const;
  Derived batch_unsqueeze(Size d) const;
  neml2::Tensor base_unsqueeze(Size d) const;
  Derived batch_transpose(Size d1, Size d2) const;
  neml2::Tensor base_transpose(Size d1, Size d2) const;
  neml2::Tensor base_flatten() const;
 
private:
  Size _batch_dim = {};
 
  TraceableTensorShape _batch_sizes;
};
 
template <class Derived>
template <class Derived2>
TensorBase<Derived>::TensorBase(const TensorBase<Derived2> & tensor)
  : _batch_dim(tensor.batch_dim()),
    _batch_sizes(tensor.batch_sizes())
{
  torch::Tensor::operator=(tensor);
}
 
template <class Derived>
template <class Derived2>
Derived
TensorBase<Derived>::batch_expand_as(const Derived2 & other) const
{
  return batch_expand(other.batch_sizes());
}
 
template <class Derived>
template <class Derived2>
Derived2
TensorBase<Derived>::base_expand_as(const Derived2 & other) const
{
  return base_expand(other.base_sizes());
}
 
template <class Derived,
          typename = typename std::enable_if_t<std::is_base_of_v<TensorBase<Derived>, Derived>>>
Derived
operator+(const Derived & a, const Real & b)
{
  return Derived(torch::operator+(a, b), a.batch_sizes());
}
 
template <class Derived,
          typename = typename std::enable_if_t<std::is_base_of_v<TensorBase<Derived>, Derived>>>
Derived
operator+(const Real & a, const Derived & b)
{
  return b + a;
}
 
template <class Derived,
          typename = typename std::enable_if_t<std::is_base_of_v<TensorBase<Derived>, Derived>>>
Derived
operator+(const Derived & a, const Derived & b)
{
  neml_assert_broadcastable_dbg(a, b);
  return Derived(torch::operator+(a, b), broadcast_batch_dim(a, b));
}
 
template <class Derived,
          typename = typename std::enable_if_t<std::is_base_of_v<TensorBase<Derived>, Derived>>>
Derived
operator-(const Derived & a, const Real & b)
{
  return Derived(torch::operator-(a, b), a.batch_sizes());
}
 
template <class Derived,
          typename = typename std::enable_if_t<std::is_base_of_v<TensorBase<Derived>, Derived>>>
Derived
operator-(const Real & a, const Derived & b)
{
  return -b + a;
}
 
template <class Derived,
          typename = typename std::enable_if_t<std::is_base_of_v<TensorBase<Derived>, Derived>>>
Derived
operator-(const Derived & a, const Derived & b)
{
  neml_assert_broadcastable_dbg(a, b);
  return Derived(torch::operator-(a, b), broadcast_batch_dim(a, b));
}
 
template <class Derived,
          typename = typename std::enable_if_t<std::is_base_of_v<TensorBase<Derived>, Derived>>>
Derived
operator*(const Derived & a, const Real & b)
{
  return Derived(torch::operator*(a, b), a.batch_sizes());
}
 
template <class Derived,
          typename = typename std::enable_if_t<std::is_base_of_v<TensorBase<Derived>, Derived>>>
Derived
operator*(const Real & a, const Derived & b)
{
  return b * a;
}
 
template <class Derived,
          typename = typename std::enable_if_t<std::is_base_of_v<TensorBase<Derived>, Derived>>>
Derived
operator/(const Derived & a, const Real & b)
{
  return Derived(torch::operator/(a, b), a.batch_sizes());
}
 
template <class Derived,
          typename = typename std::enable_if_t<std::is_base_of_v<TensorBase<Derived>, Derived>>>
Derived
operator/(const Real & a, const Derived & b)
{
  return Derived(torch::operator/(a, b), b.batch_sizes());
}
 
template <class Derived,
          typename = typename std::enable_if_t<std::is_base_of_v<TensorBase<Derived>, Derived>>>
Derived
operator/(const Derived & a, const Derived & b)
{
  neml_assert_broadcastable_dbg(a, b);
  return Derived(torch::operator/(a, b), broadcast_batch_dim(a, b));
}
} // namespace neml2