Base class for nonlinear activation functions. Do not instanciate. More...
Inheritance diagram for fnn_activation_nonlinear::nonlinearactivation:
Collaboration diagram for fnn_activation_nonlinear::nonlinearactivation:
Public Member Functions | |
| procedure, pass, public | apply_tangent_linear => nonlinear_apply_tangent_linear |
| Applies the TL of the activation function. More... | |
| procedure, pass, public | apply_adjoint => nonlinear_apply_adjoint |
| Applies the adjoint of the activation function. More... | |
 Public Member Functions inherited from fnn_activation_linear::linearactivation | |
| procedure, pass, public | tofile => linear_tofile |
| Saves the activation function. More... | |
| procedure, pass, public | apply_forward => linear_apply_forward |
| Applies and linearises the activation function. More... | |
| procedure, pass, public | apply_tangent_linear => linear_apply_tangent_linear |
| Applies the TL of the activation function. More... | |
| procedure, pass, public | apply_adjoint => linear_apply_adjoint |
| Applies the adjoint of the activation function. More... | |
Public Attributes | |
| real(rk), dimension(:, :), allocatable, public | z_prime |
| The storage for the linearisation. More... | |
| Public Attributes inherited from fnn_activation_linear::linearactivation | |
| integer(ik) | batch_size |
| The batch size. More... | |
Detailed Description
Base class for nonlinear activation functions. Do not instanciate.
This class only exists to implement the features which are common to all nonlinear activation functions:
- the storage for the linearisation;
- the TL of the activation function;
- the adjoint of the activation function.
Member Function/Subroutine Documentation
◆ apply_adjoint()
| procedure, pass, public fnn_activation_nonlinear::nonlinearactivation::apply_adjoint |
Applies the adjoint of the activation function.
Implemented by nonlinear_apply_adjoint.
◆ apply_tangent_linear()
| procedure, pass, public fnn_activation_nonlinear::nonlinearactivation::apply_tangent_linear |
Applies the TL of the activation function.
Implemented by nonlinear_apply_tangent_linear.
Member Data Documentation
◆ z_prime
| real(rk), dimension(:, :), allocatable, public fnn_activation_nonlinear::nonlinearactivation::z_prime |
The storage for the linearisation.
Array of reals with shape (linearactivation::self_size, linearactivation::batch_size).
The documentation for this type was generated from the following file:
