fnn_layer Module Reference

Module dedicated to the class layer. More...

Data Types
type	layer
	Base class for all layers. Do not instanciate. More...

Functions/Subroutines
integer(ik) function	layer_get_num_parameters (self)
	Implements layer::get_num_parameters. More...
subroutine	layer_set_parameters (self, new_parameters)
	Implements layer::set_parameters. More...
subroutine	layer_get_parameters (self, parameters)
	Implements layer::get_parameters. More...
subroutine	layer_tofile (self, unit_num)
	Implements layer::tofile. More...
subroutine	layer_apply_forward (self, train, member, x, y)
	Implements layer::apply_forward. More...
subroutine	layer_apply_tangent_linear (self, member, dp, dx, dy)
	Implements layer::apply_tangent_linear. More...
subroutine	layer_apply_adjoint (self, member, dy, dp, dx)
	Implements layer::apply_adjoint. More...

Detailed Description

Module dedicated to the class layer.

Function/Subroutine Documentation

layer_apply_adjoint()

subroutine fnn_layer::layer_apply_adjoint ( class(layer), intent(in)  self, integer(ik), intent(in)  member, real(rk), dimension(:), intent(inout)  dy, real(rk), dimension(:), intent(out)  dp, real(rk), dimension(:), intent(out)  dx  )

private

Implements layer::apply_adjoint.

Applies the adjoint of the layer.

The adjoint operator reads

where is the TL of with respect to the component and is the TL of with respect to the component.

Note

This should be overridden by each subclass.

The intent of dy is declared inout instead of in because, for certain subclasses (e.g. fnn_layer_dense::denselayer) the value of dy gets overwritten during the process (bad side-effect).

Parameters

[in]	self	The layer.
[in]	member	The index inside the batch.
[in,out]	dy	The output perturbation.
[out]	dp	The parameter perturbation.
[out]	dx	The state perturbation.

layer_apply_forward()

subroutine fnn_layer::layer_apply_forward ( class(layer), intent(inout)  self, logical, intent(in)  train, integer(ik), intent(in)  member, real(rk), dimension(:), intent(in)  x, real(rk), dimension(:), intent(out)  y  )

private

Implements layer::apply_forward.

Applies and linearises the layer.

The forward function reads

where is the vector of parameters.

Note

This should be overridden by each subclass.

The intent for self is declared inout instead of in because, for certain subclasses (e.g. fnn_layer_dense::denselayer) the linearisation is stored inside the layer.

Parameters

[in,out]	self	The layer.
[in]	train	Whether the model is used in training mode.
[in]	member	The index inside the batch.
[in]	x	The input of the layer.
[out]	y	The output of the layer.

layer_apply_tangent_linear()

subroutine fnn_layer::layer_apply_tangent_linear ( class(layer), intent(in)  self, integer(ik), intent(in)  member, real(rk), dimension(:), intent(in)  dp, real(rk), dimension(:), intent(in)  dx, real(rk), dimension(:), intent(out)  dy  )

private

Implements layer::apply_tangent_linear.

Applies the TL of the layer.

The TL operator reads

where is the TL of with respect to the component and is the TL of with respect to the component.

Note

This should be overridden by each subclass.

Parameters

[in]	self	The layer.
[in]	member	The index inside the batch.
[in]	dp	The parameter perturbation.
[in]	dx	The state perturbation.
[out]	dy	The output perturbation.

layer_get_num_parameters()

integer(ik) function fnn_layer::layer_get_num_parameters

(

class(layer), intent(in)

self

)

Implements layer::get_num_parameters.

Returns the number of parameters.

Parameters

[in]

self

The layer.

Returns

The number of parameters.

layer_get_parameters()

subroutine fnn_layer::layer_get_parameters ( class(layer), intent(in)  self, real(rk), dimension(:), intent(out)  parameters  )

private

Implements layer::get_parameters.

Getter for layer::parameters.

Parameters

[in]	self	The layer.
[out]	parameters	The vector of parameters.

layer_set_parameters()

subroutine fnn_layer::layer_set_parameters ( class(layer), intent(inout)  self, real(rk), dimension(:), intent(in)  new_parameters  )

private

Implements layer::set_parameters.

Setter for layer::parameters.

Parameters

[in,out]	self	The layer.
[in]	new_parameters	The new values for the parameters.

layer_tofile()

subroutine fnn_layer::layer_tofile ( class(layer), intent(in)  self, integer(ik), intent(in)  unit_num  )

private

Implements layer::tofile.

Saves the layer.

Note

This should be overridden by each subclass.

Parameters

[in]	self	The layer.
[in]	unit_num	The unit number for the write statement.