The monomial expansion of a Schur polynomial #
The Schur polynomial s_μ is a sum of one monomial per semistandard tableau, so its coefficient at
an exponent vector d is the Kostka number counting the tableaux of shape μ and content d
(TauCeti.coeff_schurPoly). That content is a function on the alphabet, while the Kostka
numbers TauCeti.kostkaNumber are indexed by a partition: the two differ by sorting the
exponents into decreasing order. This file closes that gap and reads off the expansion of a Schur
polynomial in the monomial symmetric polynomials MvPolynomial.msymm,
s_μ = ∑_{ν ⊢ n} K_{μν} m_ν,
with the Kostka numbers as its coefficients.
The bridge is the symmetry of s_μ. Sorting the exponents of a monomial is a permutation of the
alphabet, and a permutation of the alphabet does not change the coefficients of a symmetric
polynomial, so the coefficient of s_μ at an arbitrary exponent vector of total degree n is its
coefficient at the sorted one, which TauCeti.coeff_schurPoly_partWeight already computes as a
Kostka number. Concretely, TauCeti.weightPartition records the multiset of nonzero exponents of
a monomial, TauCeti.exists_perm_mapDomain_eq_partWeight produces the permutation that sorts them,
and TauCeti.coeff_schurPoly_eq_kostkaNumber is the resulting coefficient formula.
The sum runs over all partitions of n, with no bound relating the number of parts of ν to the
size of the alphabet: a partition with more parts than the alphabet has letters contributes nothing
because its monomial symmetric polynomial vanishes there (TauCeti.msymm_eq_zero_of_card_lt),
exactly as the Schur polynomial itself vanishes for such a shape
(TauCeti.schurPoly_eq_zero_iff).
Main definitions #
TauCeti.weightSym: the multiset of letters of a monomial of total degreen, as an element ofSym σ n.TauCeti.weightPartition: the partition ofnrecording the multiset of nonzero exponents of a monomial of total degreen.
Main results #
TauCeti.exists_perm_mapDomain_eq_partWeight: a monomial of total degreenis a permutation of the alphabet away from the sorted monomialTauCeti.partWeightof itsTauCeti.weightPartition.TauCeti.coeff_msymm: a monomial symmetric polynomial has coefficient1at the monomials of its shape and0at every other monomial, andTauCeti.msymm_eq_zero_of_card_lt: it vanishes in an alphabet with fewer letters than its partition has parts.TauCeti.coeff_schurPoly_eq_kostkaNumber: the coefficient ofs_μat any monomial of degreenis a Kostka number, that ofμand the partition of the monomial's exponents.TauCeti.schurPoly_eq_sum_kostkaNumber_smul_msymm: the monomial expansions_μ = ∑_ν K_{μν} m_ν, writing a Schur polynomial as the combination of the monomial symmetric polynomials whose coefficients are the Kostka numbers. This is the expansion only: neither family is linearly independent as indexed here, both containing zero terms whenever the partition has more parts than the alphabet has letters. Restricted to the partitions the alphabet does record they are bases, and the Kostka numbers a genuine change-of-basis matrix, which isTauCeti/RingTheory/MvPolynomial/Symmetric/Schur/Basis.lean.TauCeti.degree_partWeight,TauCeti.weightPartition_partWeightandTauCeti.coeff_msymm_partWeight: the sorted monomial of a partition the alphabet records has degreenand shape that partition, som_νis the indicator of its own sorted monomial. This is what makes the monomial symmetric polynomials independent.TauCeti.isHomogeneous_msymm: a monomial symmetric polynomial is homogeneous.
Implementation notes #
One general fact is used and kept private here rather than stated for its own sake: that two
families on a finite type taking the same multiset of values differ by a permutation of the index
type. It is the shape of the sorting argument in this file and has no other consumer yet.
References #
- W. Fulton, Young Tableaux, Section 2.2, where
s_λ = ∑_μ K_{λμ} m_μis read off the tableau definition. - R. P. Stanley, Enumerative Combinatorics, Volume 2, §7.10.
- Schur--Weyl roadmap,
Layer 7, "the
msymm-to-schurPolychange of basis is the Kostka matrixKλμ".
Rearranging a family along a permutation #
The partition of the exponents of a monomial #
The number of letters of the monomial with exponent vector d is its total degree. This is
not a simp lemma: Finsupp.card_toMultiset already rewrites the left-hand side to
d.sum fun _ => id, so tagging it would leave it out of simp-normal form.
The multiset of letters of a monomial of total degree n, read as an element of
Sym σ n: the letter x occurs as often as the exponent of x prescribes.
Equations
- TauCeti.weightSym d h = ⟨Finsupp.toMultiset d, ⋯⟩
Instances For
The partition of the exponents of a monomial of total degree n: its parts are the nonzero
exponents, so it is the shape of the monomial once its exponents are sorted decreasingly.
Equations
Instances For
The parts of the partition of the exponents of a monomial are the exponents of the letters that actually occur in it.
A monomial has one nonzero exponent per letter occurring in it, so the partition of its exponents has no more parts than the alphabet has letters.
Sorting the exponents of a monomial #
A monomial of total degree n is a rearrangement of the sorted monomial of its shape: some
permutation of the alphabet carries it to the exponent vector TauCeti.partWeight recording the
parts of its TauCeti.weightPartition.
The coefficients of a monomial symmetric polynomial #
A monomial symmetric polynomial is homogeneous: it has no monomial whose total degree is not that of its partition.
The coefficients of a monomial symmetric polynomial are 0 and 1: m_ν is the sum of
the monomials of total degree n whose nonzero exponents are the parts of ν, each occurring
once.
A monomial symmetric polynomial vanishes when its partition has more parts than the alphabet has letters: no monomial in that alphabet uses that many distinct letters.
A monomial symmetric polynomial is homogeneous of degree the natural number its partition partitions: every monomial occurring in it has the parts of the partition as its exponents.
The sorted monomial of a partition #
The sorted monomial of a partition of n has total degree n, its exponents being the
parts of the partition padded with zeros. The hypothesis is what keeps every part inside the
alphabet: a longer partition would be truncated.
Sorting the exponents of an already sorted monomial changes nothing: the partition of the
exponents of TauCeti.partWeight σ ν is ν itself. Together with TauCeti.coeff_msymm this
says that the monomial symmetric polynomials take the value 1 at their own sorted monomial and
0 at every other one.
A monomial symmetric polynomial is the indicator of its own sorted monomial: m_ν has
coefficient 1 at the sorted monomial of ν and 0 at the sorted monomial of any other
partition. This is the statement that the monomial symmetric polynomials are dual to the sorted
monomials, and the reason they are linearly independent.
The monomial expansion #
The coefficient of a Schur polynomial at any monomial of degree n is a Kostka number:
that of the shape μ and the partition of the monomial's exponents. The exponents need not be
sorted, since sorting them is a permutation of the alphabet, which a symmetric polynomial does not
see.
The monomial expansion of a Schur polynomial: s_μ = ∑_ν K_{μν} m_ν, expanding s_μ in
the monomial symmetric polynomials with the Kostka numbers as its coefficients. The sum runs over
every partition of n: those with more parts than the alphabet has letters contribute nothing,
their monomial symmetric polynomial vanishing there. (Being an expansion, this does not by itself
say that the Kostka numbers are a change-of-basis matrix: no basis result is proved here.)