Goto Chapter: Top 1 2 3 4 5 6 7 8 9 10 Ind
 [Top of Book]  [Contents]   [Previous Chapter]   [Next Chapter] 

9 CAP Operations for GradientBasedLearningForCAP
 9.1 Basic Operations
 9.2 Add-Methods

9 CAP Operations for GradientBasedLearningForCAP

9.1 Basic Operations

9.1-1 MultiplicationForMorphisms
‣ MultiplicationForMorphisms( alpha, beta )( operation )

Returns: a morphism

Compute the pointwise multiplication of two morphisms with the same source and target. For morphisms f, g: A \to B, returns a morphism whose output at each component is the product of the outputs of f and g.

9.1-2 ReverseDifferentialWithGivenObjects
‣ ReverseDifferentialWithGivenObjects( source, alpha, range )( operation )

Returns: a morphism

Compute the reverse differential of a morphism alpha with given source and target objects. For a morphism f: \mathbb{R}^m \to \mathbb{R}^n, the reverse differential is a morphism Df: \mathbb{R}^m \times \mathbb{R}^n \to \mathbb{R}^m that computes y \cdot J_f(x) where J_f is the Jacobian matrix of f.

9.1-3 ReverseDifferential
‣ ReverseDifferential( alpha )( attribute )

Returns: a morphism

Compute the reverse differential of a morphism alpha. This is equivalent to ReverseDifferentialWithGivenObjects with automatically computed source and target.

9.2 Add-Methods

9.2-1 AddMultiplicationForMorphisms
‣ AddMultiplicationForMorphisms( C, F )( operation )
‣ AddMultiplicationForMorphisms( C, F, weight )( operation )

Returns: nothing

The arguments are a category C and a function F. This operation adds the given function F to the category for the basic operation MultiplicationForMorphisms. Optionally, a weight (default: 100) can be specified which should roughly correspond to the computational complexity of the function (lower weight = less complex = faster execution). F: ( alpha, beta ) \mapsto \mathtt{MultiplicationForMorphisms}(alpha, beta).

9.2-2 AddReverseDifferential
‣ AddReverseDifferential( C, F )( operation )
‣ AddReverseDifferential( C, F, weight )( operation )

Returns: nothing

The arguments are a category C and a function F. This operation adds the given function F to the category for the basic operation ReverseDifferential. Optionally, a weight (default: 100) can be specified which should roughly correspond to the computational complexity of the function (lower weight = less complex = faster execution). F: ( alpha ) \mapsto \mathtt{ReverseDifferential}(alpha).

9.2-3 AddReverseDifferentialWithGivenObjects
‣ AddReverseDifferentialWithGivenObjects( C, F )( operation )
‣ AddReverseDifferentialWithGivenObjects( C, F, weight )( operation )

Returns: nothing

The arguments are a category C and a function F. This operation adds the given function F to the category for the basic operation ReverseDifferentialWithGivenObjects. Optionally, a weight (default: 100) can be specified which should roughly correspond to the computational complexity of the function (lower weight = less complex = faster execution). F: ( source, alpha, range ) \mapsto \mathtt{ReverseDifferentialWithGivenObjects}(source, alpha, range).

 [Top of Book]  [Contents]   [Previous Chapter]   [Next Chapter] 
Goto Chapter: Top 1 2 3 4 5 6 7 8 9 10 Ind

generated by GAPDoc2HTML