Weak Schur sampling weights of sum to one (RSK count)
OpenMixedCloning.Optimal.wss_normalizationcombinatoricsrepresentation-theory
Let and be integers. For a partition let be the number of standard Young tableaux of shape . Let be the number of semistandard Young tableaux of shape with entries in , that is, the Schur polynomial evaluated at . Then
Equivalently, for the weak Schur sampling weights of sum to . Bijectively, this is the RSK correspondence for words. A word corresponds to a pair of tableaux of the same shape , where is semistandard with entries in and is standard.
Differences from the source:
- encoding: partitions and tableaux with symbols starting at 1 → YoungDiagram cells and Fin alphabets starting at 0; reason: shifting indices preserves all shapes, orders and counts
- specialization: the symmetric-function identity → its evaluation at , with the sum restricted to ; reason: , counts SSYT with entries in , and when
- form: normalization of probabilities → identity multiplied by ; reason: avoids division, and it also holds at (both sides are for and for )
Preamble
import Definitions.Def_MixedCloning_Optimal_Schur noncomputable section open Matrix MeasureTheory open scoped Matrix ComplexConjugate ComplexOrder Matrix.Norms.Elementwise
Formal statement
namespace MixedCloning.Optimal
attribute [local instance] Classical.propDecidable Classical.decEq
theorem wss_normalization (n r : ℕ) :
∑ L ∈ partitions n r, (spechtDim L : ℝ) * schurCount L r = (r : ℝ)^n := by sorry
end MixedCloning.Optimal
Source
R. P. Stanley, Enumerative Combinatorics, Vol. 2, Cambridge Univ. Press, 1999 (doi:10.1017/CBO9780511609589), Ch. 7: Corollary 7.12.5, eq. (7.47) h_1^n = sum_{lambda |- n} f^lambda s_lambda, p. 324 (its bijective proof is RSK, Theorem 7.11.5, p. 319, applied to N-matrices with col(A) = (1^n), i.e. words); s_lambda as the SSYT generating function: Definition 7.10.1, p. 310. Used in Grinko–Fanizza–Scharnhorst–Spilecki, Optimal cloning of mixed states: PDF p. 15 (weak Schur sampling outputs lambda with probability dim(P_lambda) s_lambda(rho)) and PDF p. 18 (lambda ~ WSS_n(1^r/r) has probability dim(P_lambda) dim(Q^r_lambda)/r^n for l(lambda) <= r).