Tensor products of cones #
Given ordered modules M and N, there are in general several distinct possible
orderings of the tensor product module M ⊗ N. Since the ordering of an ordered module
can be represented by its cone of nonnegative elements, there are likewise multiple
ways to construct a cone in M ⊗ N from cones in M and N. Such constructions
are referred to as tensor products of cones.
"Sufficiently nice" candidates for tensor products of cones are bounded by the minimal and maximal tensor products. These products are generally distinct but coincide in special cases.
We define the minimal and maximal tensor products of pointed cones:
minTensorProduct C₁ C₂: all conical combinations of elementary tensor productsx ⊗ₜ ywithx ∈ C₁andy ∈ C₂.maxTensorProduct C₁ C₂: the dual cone of the minimal tensor product of the dual cones.
Main results #
minTensorProduct_le_maxTensorProduct: the minimal tensor product is less than or equal to the maximal tensor product
Notation #
- no special notation defined
- x, y, z are elements of the (original) cones
- φ, ψ are elements of the dual cones
References #
The minimal tensor product of two cones is given by all conical combinations of elementary
tensor products x ⊗ₜ y with x ∈ C₁ and y ∈ C₂.
Equations
- C₁.minTensorProduct C₂ = PointedCone.span R (Set.image2 (fun (x1 : G) (x2 : H) => x1 ⊗ₜ[R] x2) ↑C₁ ↑C₂)
Instances For
The maximal tensor product of two cones is the dual (pointed cone) of the minimal tensor product of the dual cones.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Characterization of the maximal tensor product: z lies in maxTensorProduct C₁ C₂ iff
all pairings with elementary dual tensors are nonnegative.
Elementary tensors are members of the maximal tensor product.
Elementary tensors are members of the minimal tensor product.
The maximal tensor product contains the set of all elementary tensors.
The minimal tensor product contains the set of all elementary tensors.
The minimal tensor product is less than or equal to the maximal tensor product.