Each layer is composed of data, mappings, and transformations. Datasets can be replaced, mappings can be merged, and transformations can be concatenated. These operations, taken together, define an associative operation on layers, which we call multiplication
Multiplication is primarily useful to combine partially defined layers.
+ is used to superimpose separate layers.
a + b has as many layers as
la + lb, where
lb are the number of layers in
Multiplication naturally extends to lists of layers. Given two
lb layers respectively, the product
a * b contains
la * lb layers—all possible pair-wise products.