NEML2 1.4.0
Loading...
Searching...
No Matches
WR2 Class Reference

A skew rank 2, represented as an axial vector. More...

Detailed Description

A skew rank 2, represented as an axial vector.

The logical storage space is (3).

#include <WR2.h>

Inheritance diagram for WR2:

Public Member Functions

 WR2 (const R2 &T)
 Skew-symmetrize a R2 then fill.
 
Scalar operator() (TorchSize i, TorchSize j) const
 Accessor.
 
Rot exp () const
 Exponential map to make this into a rotation (Rot)
 
R2 dexp () const
 Derivative of the exponential map.
 
- Public Member Functions inherited from VecBase< WR2 >
Scalar operator() (TorchSize i) const
 Accessor.
 
Scalar dot (const VecBase< Derived2 > &v) const
 dot product
 
WR2 cross (const VecBase< Derived2 > &v) const
 cross product
 
R2 outer (const VecBase< Derived2 > &v) const
 outer product
 
Scalar norm_sq () const
 Norm squared.
 
Scalar norm () const
 Norm.
 
WR2 rotate (const Rot &r) const
 Rotate using a Rodrigues vector.
 
WR2 rotate (const R2 &R) const
 Rotate using a rotation matrix.
 
R2 drotate (const Rot &r) const
 Derivative of the rotated vector w.r.t. the Rodrigues vector.
 
R3 drotate (const R2 &R) const
 Derivative of the rotated vector w.r.t. the rotation matrix.
 
- Public Member Functions inherited from FixedDimTensor< Derived, S >
 FixedDimTensor ()=default
 Default constructor.
 
 FixedDimTensor (const torch::Tensor &tensor, TorchSize batch_dim)
 Construct from another torch::Tensor given batch dimension.
 
 FixedDimTensor (const torch::Tensor &tensor)
 Construct from another torch::Tensor and infer batch dimension.
 
 operator BatchTensor () const
 Implicit conversion to a BatchTensor and loses information on the fixed base shape.
 
- Public Member Functions inherited from BatchTensorBase< Derived >
 BatchTensorBase ()=default
 Default constructor.
 
 BatchTensorBase (const torch::Tensor &tensor, TorchSize batch_dim)
 Construct from another torch::Tensor.
 
 BatchTensorBase (const Derived &tensor)
 Copy constructor.
 
 BatchTensorBase (Real)=delete
 
bool batched () const
 Whether the tensor is batched.
 
TorchSize batch_dim () const
 Return the number of batch dimensions.
 
TorchSizebatch_dim ()
 Return a writable reference to the batch dimension.
 
TorchSize base_dim () const
 Return the number of base dimensions.
 
TorchShapeRef batch_sizes () const
 Return the batch size.
 
TorchSize batch_size (TorchSize index) const
 Return the length of some batch axis.
 
TorchShapeRef base_sizes () const
 Return the base size.
 
TorchSize base_size (TorchSize index) const
 Return the length of some base axis.
 
TorchSize base_storage () const
 Return the flattened storage needed just for the base indices.
 
Derived batch_index (TorchSlice indices) const
 Get a batch.
 
BatchTensor base_index (const TorchSlice &indices) const
 Return an index sliced on the base dimensions.
 
void batch_index_put (TorchSlice indices, const torch::Tensor &other)
 Set a index sliced on the batch dimensions to a value.
 
void base_index_put (const TorchSlice &indices, const torch::Tensor &other)
 Set a index sliced on the base dimensions to a value.
 
Derived batch_expand (TorchShapeRef batch_size) const
 Return a new view of the tensor with values broadcast along the batch dimensions.
 
BatchTensor base_expand (TorchShapeRef base_size) const
 Return a new view of the tensor with values broadcast along the base dimensions.
 
template<class Derived2 >
Derived batch_expand_as (const Derived2 &other) const
 Expand the batch to have the same shape as another tensor.
 
template<class Derived2 >
Derived2 base_expand_as (const Derived2 &other) const
 Expand the base to have the same shape as another tensor.
 
Derived batch_expand_copy (TorchShapeRef batch_size) const
 Return a new tensor with values broadcast along the batch dimensions.
 
BatchTensor base_expand_copy (TorchShapeRef base_size) const
 Return a new tensor with values broadcast along the base dimensions.
 
Derived batch_reshape (TorchShapeRef batch_shape) const
 Reshape batch dimensions.
 
BatchTensor base_reshape (TorchShapeRef base_shape) const
 Reshape base dimensions.
 
Derived batch_unsqueeze (TorchSize d) const
 Unsqueeze a batch dimension.
 
Derived list_unsqueeze () const
 Unsqueeze on the special list batch dimension.
 
BatchTensor base_unsqueeze (TorchSize d) const
 Unsqueeze a base dimension.
 
Derived batch_transpose (TorchSize d1, TorchSize d2) const
 Transpose two batch dimensions.
 
BatchTensor base_transpose (TorchSize d1, TorchSize d2) const
 Transpose two base dimensions.
 
BatchTensor base_movedim (TorchSize d1, TorchSize d2) const
 Move two base dimensions.
 
Derived clone (torch::MemoryFormat memory_format=torch::MemoryFormat::Contiguous) const
 Clone (take ownership)
 
Derived detach () const
 Discard function graph.
 
Derived to (const torch::TensorOptions &options) const
 Send to options.
 
Derived operator- () const
 Negation.
 
Derived batch_sum (TorchSize d) const
 Sum on a batch index.
 
Derived list_sum () const
 Sum on the list index (TODO: replace with class)
 

Additional Inherited Members

- Static Public Member Functions inherited from VecBase< WR2 >
static WR2 fill (const Real &v1, const Real &v2, const Real &v3, const torch::TensorOptions &options=default_tensor_options())
 
static WR2 fill (const Scalar &v1, const Scalar &v2, const Scalar &v3)
 
static R2 identity_map (const torch::TensorOptions &options=default_tensor_options())
 The derivative of a vector with respect to itself.
 
- Static Public Member Functions inherited from FixedDimTensor< Derived, S >
static Derived empty (const torch::TensorOptions &options=default_tensor_options())
 Unbatched empty tensor.
 
static Derived empty (TorchShapeRef batch_shape, const torch::TensorOptions &options=default_tensor_options())
 Empty tensor given batch shape.
 
static Derived zeros (const torch::TensorOptions &options=default_tensor_options())
 Unbatched zero tensor.
 
static Derived zeros (TorchShapeRef batch_shape, const torch::TensorOptions &options=default_tensor_options())
 Zero tensor given batch shape.
 
static Derived ones (const torch::TensorOptions &options=default_tensor_options())
 Unbatched unit tensor.
 
static Derived ones (TorchShapeRef batch_shape, const torch::TensorOptions &options=default_tensor_options())
 Unit tensor given batch shape.
 
static Derived full (Real init, const torch::TensorOptions &options=default_tensor_options())
 Unbatched tensor filled with a given value given base shape.
 
static Derived full (TorchShapeRef batch_shape, Real init, const torch::TensorOptions &options=default_tensor_options())
 Full tensor given batch shape.
 
static BatchTensor identity_map (const torch::TensorOptions &)
 Derived tensor classes should define identity_map where appropriate.
 
- Static Public Member Functions inherited from BatchTensorBase< Derived >
static Derived empty_like (const Derived &other)
 Empty tensor like another, i.e. same batch and base shapes, same tensor options, etc.
 
static Derived zeros_like (const Derived &other)
 Zero tensor like another, i.e. same batch and base shapes, same tensor options, etc.
 
static Derived ones_like (const Derived &other)
 Unit tensor like another, i.e. same batch and base shapes, same tensor options, etc.
 
static Derived full_like (const Derived &other, Real init)
 
static Derived linspace (const Derived &start, const Derived &end, TorchSize nstep, TorchSize dim=0, TorchSize batch_dim=-1)
 Create a new tensor by adding a new batch dimension with linear spacing between start and end.
 
static Derived logspace (const Derived &start, const Derived &end, TorchSize nstep, TorchSize dim=0, TorchSize batch_dim=-1, Real base=10)
 log-space equivalent of the linspace named constructor
 
- Static Public Attributes inherited from FixedDimTensor< Derived, S >
static const TorchShape const_base_sizes = {S...}
 The base shape.
 
static constexpr TorchSize const_base_dim = sizeof...(S)
 The base dim.
 
static const TorchSize const_base_storage = utils::storage_size({S...})
 The base storage.
 

Constructor & Destructor Documentation

◆ WR2()

WR2 ( const R2 & T)

Skew-symmetrize a R2 then fill.

Member Function Documentation

◆ dexp()

R2 dexp ( ) const

Derivative of the exponential map.

◆ exp()

Rot exp ( ) const

Exponential map to make this into a rotation (Rot)

◆ operator()()

Scalar operator() ( TorchSize i,
TorchSize j ) const

Accessor.