Physlib

Physlib.Relativity.Tensors.Contraction.CrossToSlot

Result-to-slot contraction against a rank-2 tensor

i. Overview

`crossToEnd` deposits the surviving index of a slot contraction at the end of the survivor list. Raising and lowering a named index instead wants the replacement to stay *in place*: contracting slot `i` of `t` against a rank-2 tensor `M` (a metric, the unit tensor) should return `t` with only slot `i`'s color changed, `T^{μ}{}_{νρ} ↦ T_{μνρ}`. `crossToSlot i j hc M` is that operation, and `crossToSlot_eq_crossToEnd` is the one place the two conventions meet.

This module stays at the `CommRing` altitude of `crossToEnd`; the `RCLike` round trips built on the operation live with the unit-tensor collapse theory in `Physlib.Relativity.Tensors.Contraction.UnitTensorContraction`.

ii. Key results

- `TensorSpecies.Tensor.crossToSlot` : contract slot `i` against slot `j` of a rank-2 tensor and rotate the survivor back into position `i`; raising and lowering a named index. - `TensorSpecies.Tensor.crossToSlot_eq_crossToEnd` : the bridge to the result-to-end convention. - `TensorSpecies.Tensor.crossToSlotInv` : the returning half of a round trip, the contraction against the second factor with the round trip's color cast absorbed. - `TensorSpecies.Tensor.crossToSlot_permT_right_id` : an identity reindexing of the rank-2 tensor passes through the contraction. - `TensorSpecies.Tensor.crossToSlot_equivariant` : the contraction commutes with the `G`-action.

iii. Table of contents

  • A. The result-to-slot contraction

iv. References

A. The result-to-slot contraction

6 declarations

theorem

(cycle(i,nA))−1(\text{cycle}(i, n_A))^{-1} is a reindexing for result-to-slot contraction

Let nAn_A be a natural number, c:{0,…,nA}→Cc: \{0, \dots, n_A\} \to C be a sequence of tensor index types (colors), and cM:{0,1}→Cc_M: \{0, 1\} \to C be the colors of a rank-2 tensor. For any index i∈{0,…,nA}i \in \{0, \dots, n_A\} and j∈{0,1}j \in \{0, 1\}, the inverse of the cyclic permutation on the interval from ii to nAn_A, denoted by σ=(cycle(i,nA))−1\sigma = (\text{cycle}(i, n_A))^{-1}, is a reindexing from a source color sequence to a target color sequence. The source sequence is the concatenation of cc (omitting index ii) and cMc_M (omitting index jj). The target sequence is the original sequence cc where the ii-th entry is updated to be the surviving index type of cMc_M (the entry not at index jj).

definition

Result-to-slot contraction of slot ii with slot jj of a rank-2 tensor MM

Let SS be a tensor species over a ring kk. Let c:{0,…,n}→Cc: \{0, \dots, n\} \to C be a sequence of colors for a tensor tt, and let cM:{0,1}→Cc_M: \{0, 1\} \to C be the color sequence for a rank-2 tensor MM. Given indices i∈{0,…,n}i \in \{0, \dots, n\} and j∈{0,1}j \in \{0, 1\}, such that the color c(i)c(i) is dual to the color cM(j)c_M(j) (i.e., S.τ(c(i))=cM(j)S.\tau(c(i)) = c_M(j)), the kk-linear map `crossToSlot` performs a contraction between the ii-th slot of tt and the jj-th slot of MM. In this operation, the surviving index of MM (the index j′≠jj' \neq j) is placed back into the original position ii of the resulting tensor. The color sequence of the result is cc with the ii-th entry updated to the color cM(j′)c_M(j'). This operation corresponds to raising or lowering the ii-th index of a tensor using a rank-2 tensor such as a metric.

theorem

`crossToSlot` equals `crossToEnd` followed by an inverse cyclic permutation

Let SS be a tensor species over a ring kk. Let tt be a tensor with color sequence c:Fin(nA+1)→Cc: \text{Fin}(n_A + 1) \to C and MM be a rank-2 tensor with color sequence cM:Fin(2)→Cc_M: \text{Fin}(2) \to C. For indices i∈Fin(nA+1)i \in \text{Fin}(n_A + 1) and j∈Fin(2)j \in \text{Fin}(2) such that the color c(i)c(i) is dual to cM(j)c_M(j) (i.e., S.τ(c(i))=cM(j)S.\tau(c(i)) = c_M(j)), the result-to-slot contraction crossToSlot\text{crossToSlot} is related to the result-to-end contraction crossToEnd\text{crossToEnd} by a cyclic permutation. Specifically: crossToSlot(i,j,M,t)=permT(σ,crossToEnd(i,j,t,M))\text{crossToSlot}(i, j, M, t) = \text{permT}\left(\sigma, \text{crossToEnd}(i, j, t, M)\right) where σ=(cycle(i,nA))−1\sigma = (\text{cycle}(i, n_A))^{-1} is the inverse of the cyclic permutation on the interval [i,nA][i, n_A], which rotates the surviving index from the last slot back into the ii-th slot.

definition

Inverse result-to-slot contraction of slot ii with rank-2 tensor M′M'

Let SS be a tensor species over a ring kk. Let c:{0,…,n}→Cc: \{0, \dots, n\} \to C be a color sequence and i∈{0,…,n}i \in \{0, \dots, n\} be a fixed index such that the color at that position is c(i)=ec(i) = e. Let d,b∈Cd, b \in C be colors such that the dual of dd is bb (i.e., S.τ(d)=bS.\tau(d) = b). Given a rank-2 tensor M′M' with color sequence (b,e)(b, e), the kk-linear map `crossToSlotInv` transforms a tensor tt from the space Tensor(S,c[i↦d])\text{Tensor}(S, c[i \mapsto d]) (where the ii-th color of cc has been replaced by dd) to the original space Tensor(S,c)\text{Tensor}(S, c). This is performed by contracting the ii-th slot of tt (color dd) with the first slot of M′M' (color bb), and placing the surviving second slot of M′M' (color ee) back into the ii-th position. This operation serves as the "returning" half of an index-shifting round trip (such as raising then lowering an index), ensuring the result aligns with the original color sequence cc.

theorem

Identity reindexing of the rank-2 tensor in `crossToSlot` contraction

Let SS be a tensor species over a ring kk. Let tt be a tensor of rank n+1n+1 with color sequence c:{0,…,n}→Cc: \{0, \dots, n\} \to C, and let MM be a rank-2 tensor with color sequence cM:{0,1}→Cc_M: \{0, 1\} \to C. Suppose cM′c_M' is another color sequence such that the identity map is a reindexing from cMc_M to cM′c_M' (meaning cM(k)=cM′(k)c_M(k) = c_M'(k) for k∈{0,1}k \in \{0, 1\}). Let i∈{0,…,n}i \in \{0, \dots, n\} and j∈{0,1}j \in \{0, 1\} be indices such that the ii-th color of tt is dual to the jj-th color of MM under the sequence cM′c_M', i.e., S.τ(c(i))=cM′(j)S.\tau(c(i)) = c_M'(j). Then, the `crossToSlot` contraction of tt with MM (after MM has been reindexed to cM′c_M' via the identity map) is equal to the identity reindexing of the `crossToSlot` contraction of tt with MM using its original colors cMc_M.

theorem

The Result-to-Slot Contraction is GG-Equivariant

Let SS be a tensor species over a ring kk and GG be a group acting on the tensors of SS. Let tt be a tensor of rank n+1n+1 with color sequence c:{0,…,n}→Cc: \{0, \dots, n\} \to C, and MM be a rank-2 tensor with color sequence cM:{0,1}→Cc_M: \{0, 1\} \to C. Given indices i∈{0,…,n}i \in \{0, \dots, n\} and j∈{0,1}j \in \{0, 1\} such that the color c(i)c(i) is dual to cM(j)c_M(j) (i.e., S.τ(c(i))=cM(j)S.\tau(c(i)) = c_M(j)), the operation `crossToSlot` contracts the ii-th slot of tt with the jj-th slot of MM and places the surviving index of MM into the original position ii. This operation is GG-equivariant, meaning that for any g∈Gg \in G: crossToSlot(i,j,g⋅M,g⋅t)=g⋅(crossToSlot(i,j,M,t)) \text{crossToSlot}(i, j, g \cdot M, g \cdot t) = g \cdot (\text{crossToSlot}(i, j, M, t))