Singleton upper quantum functional and weighted flattening-rank product
DefinitionAsymSpecConn_Supplement_singletonUpperQuantumalgebraic-complexityp2o-gran-per-chapterp2o-plan-paperp2o-v1quantum-functionaltensor-rank
For a probability vector , the singleton upper quantum functional assigns zero to the zero tensor and, to a nonzero complex tensor , assigns
The associated weighted flattening-rank product is , with .
This functional is the equivalent moment-polytope form of the isotypic-support supremum used in §5.2 and is the object in Theorem 5.4.
Formalization Note The imported entropy uses natural logarithms, so exponentiating its maximum gives the same value as raised to the source's base-two entropy. Chosen bases convert abstract finite-dimensional legs to coordinate arrays; the value is invariant under invertible changes of basis.
Definition code
import Mathlib
import Definitions.Def_MomentPolytopes_QuantumFunctionals_QuantumFunctional
import Definitions.Def_AsymSpecConn_Fw
import Definitions.Def_AsymSpecConn_irreducibleSet
import Definitions.Def_AsymSpecConn_flatRank
set_option autoImplicit false
open scoped NNReal ENNReal
namespace AsymSpecConn.Supplement
/-- Coordinate dimensions obtained from Mathlib's chosen bases of the legs of a concrete
complex tensor. -/
noncomputable def coordDim {k : ℕ} (X : MME.TensorObj ℂ k) (i : Fin k) : ℕ :=
Fintype.card (Module.Free.ChooseBasisIndex ℂ (X.V i))
/-- A concrete tensor written as an ordinary coordinate array using arbitrary chosen bases.
Changing those bases does not change the entanglement polytope, hence does not change the
quantum functional. -/
noncomputable def tensorCoordinates {k : ℕ} (X : MME.TensorObj ℂ k) :
MomentPolytopes.QuantumFunctionals.Ten k 1 (coordDim X) :=
fun x =>
let basis := Basis.piTensorProduct (fun i => Module.Free.chooseBasis ℂ (X.V i))
basis.repr X.t (fun i => (Fintype.equivFin (Module.Free.ChooseBasisIndex ℂ (X.V i))).symm (x.2 i))
/-- The singleton upper quantum functional `F^theta` of §5.2 (p.39) on a concrete tensor:
zero at the zero tensor, and `exp(max ∑_i theta_i H(p_i))` otherwise.
**Formalization Note** The imported moment-polytope definition uses natural logarithms, so
`exp(E_nat)` equals the source's `2^(E_base2)`. The entanglement-polytope maximum is equivalent
to the source's Schur--Weyl isotypic-support supremum. Boundary weights `theta_i = 0` are retained. -/
noncomputable def singletonUpperQuantumObj {k : ℕ} (theta : Fin k → ℝ)
(X : MME.TensorObj ℂ k) : ℝ≥0 := by
classical
exact if X.t = 0 then 0 else
Real.toNNReal (Real.exp
(MomentPolytopes.QuantumFunctionals.logQuantumFunctional theta (tensorCoordinates X)))
/-- The singleton upper quantum functional on tensor-equivalence classes. The selected quotient
representative is immaterial because the moment-polytope value is invariant under invertible
changes of basis and deletion of unused ambient directions. -/
noncomputable def singletonUpperQuantum {k : ℕ} (theta : Fin k → ℝ) :
MME.TensorQ ℂ k → ℝ≥0 :=
fun s => singletonUpperQuantumObj theta (Quotient.out s)
/-- `w_theta = ∏_i R_i^{theta_i}` from Corollary 5.2 and Theorem 5.4, with `0^0=1`. -/
noncomputable def singletonRankWeight {k : ℕ} (hk : 2 ≤ k) (theta : Fin k → ℝ) :
MME.TensorQ ℂ k → ℝ≥0 :=
fun s => ∏ i, (AsymSpecConn.flatRankQ hk i s : ℝ≥0) ^ theta i
end AsymSpecConn.Supplement
Source
Vrana and Zuiddam, The asymptotic spectrum of tensors is connected, preprint (STOC 2027 submission, version of November 2, 2026, sha256 85a0b2ac698c902a02f2df11dd0671a5d9ca373aed822ae44b7e2640fc152c7d), Definition 5.3 and Theorem 5.4 p.39