Symmetry, Integrability and Geometry: Methods and Applications SIGMA 12 (2016), 033, 27 pages
Orthogonality Measure on the Torus
for Vector-Valued Jack Polynomials?
Charles F. DUNKL
Department of Mathematics, University of Virginia,
PO Box 400137, Charlottesville VA 22904-4137, USA
E-mail: cfd5z@virginia.edu
URL: http://people.virginia.edu/~cfd5z/
Received November 26, 2015, in final form March 23, 2016; Published online March 27, 2016
http://dx.doi.org/10.3842/SIGMA.2016.033
Abstract. For each irreducible module of the symmetric group on N objects there is
a set of parametrized nonsymmetric Jack polynomials in N variables taking values in the
module. These polynomials are simultaneous eigenfunctions of a commutative set of opera-
tors, self-adjoint with respect to certain Hermitian forms. These polynomials were studied
by the author and J.-G. Luque using a Yang–Baxter graph technique. This paper con-
structs a matrix-valued measure on the N -torus for which the polynomials are mutually
orthogonal. The construction uses Fourier analysis techniques. Recursion relations for the
Fourier–Stieltjes coefficients of the measure are established, and used to identify parameter
values for which the construction fails. It is shown that the absolutely continuous part of
the measure satisfies a first-order system of differential equations.
Key words: nonsymmetric Jack polynomials; Fourier–Stieltjes coefficients; matrix-valued
measure; symmetric group modules
2010 Mathematics Subject Classification: 33C52; 42B10; 20C30; 46G10; 35F35
1 Introduction
The Jack polynomials form a parametrized basis of symmetric polynomials. A special case of
these consists of the Schur polynomials, important in the character theory of the symmetric
groups. By means of a commutative algebra of differential-difference operators the theory was
extended to nonsymmetric Jack polynomials, again a parametrized basis but now for all poly-
nomials in N variables. These polynomials are orthogonal for several different inner products,
and in each case they are simultaneous eigenfunctions of a commutative set of self-adjoint ope-
rators. These inner products are invariant under permutations of the coordinates, that is, the
symmetric group. One of these inner products is that of L2
(
TN ,Kκ(x)dm(x)
)
, where
TN :=
{
x ∈ CN : |xj | = 1, 1 ≤ j ≤ N
}
,
dm(x) = (2pi)−Ndθ1 · · · dθN , xj = exp(iθj), −pi < θj ≤ pi, 1 ≤ j ≤ N,
Kκ(x) =
∏
1≤i<j≤N
|xi − xj |
2κ, κ > −
1
N
;
defining the N -torus, the Haar measure on the torus, and the weight function respectively.
Beerends and Opdam [1] discovered this orthogonality property of symmetric Jack polyno-
mials. Opdam [12] established orthogonality structures on the torus for trigonometric polyno-
mials associated with Weyl groups; the nonsymmetric Jack polynomials form a special case.
?This paper is a contribution to the Special Issue on Orthogonal Polynomials, Special Functions and Applica-
tions. The full collection is available at http://www.emis.de/journals/SIGMA/OPSFA2015.html
2 C.F. Dunkl
Details on the derivation of the norm formulae can be found in the treatise by Xu and the
author [6, Section 10.6.3]. The weight function Kκ turned out to be the square of the base
state for the Calogero–Sutherland quantum mechanical model of N identical particles located
at x1, x2, . . . , xN on the circle with a 1/r2 potential. This means that the particles repel each
other with a force corresponding to a potential C|xi − xj |−2. See Lapointe and Vinet [10] for
the construction of wavefunctions in terms of Jack polynomials for this model. More recently
Griffeth [8] constructed vector-valued Jack polynomials for the family G (n, p,N) of complex
reflection groups. These are the groups of permutation matrices (exactly one nonzero entry in
each row and each column) whose nonzero entries are nth roots of unity and the product of these
entries is a (n/p)th root of unity. The symmetric groups and the hyperoctahedral groups are the
special cases G(1, 1, N) and G(2, 1, N) respectively. The term “vector-valued” means that the
polynomials take values in irreducible modules of the underlying group, and the action of the
group is on the range as well as the domain of the polynomials. The author [3] together with
Luque [5] investigated the symmetric group case more intensively. The results from these two
papers are the foundation for the present work.
Since the torus structure is such an important aspect of the theory of Jack polynomials it
seemed like an obvious research topic to find the role of the torus in the vector-valued Jack case.
Is there a matrix-valued weight function on the torus for which the vector-valued Jack polyno-
mials are mutually orthogonal? Some explorations in the N = 3 and N = 4 situation showed
that the theory is much more complicated than the ordinary (scalar) case. For two-dimensional
representations the weight function has hypergeometric function entries (see [4]); this is quite
different from the rather natural product
∏
1≤i<j≤N
|xi − xj |2κ, a power of the discriminant.
In this paper we will produce a matrix-valued measure on the torus for which the vector-
valued nonsymmetric polynomials are mutually orthogonal. The result applies to arbitrary
irreducible representations of the symmetric groups. In each case there is a permitted range
of the parameter. We start with a concise outline of the definitions and construction of the
polynomials using the Yang–Baxter graph technique in Section 2, based on [3] and [5]. Section 3
contains the construction of the abstract Hermitian form which is designed to act like an integral
over the torus; that is, multiplication by a coordinate function xj is an isometry. The method
is algebraic and based on the Yang–Baxter graph. In Section 4 we use techniques from Fourier
analysis to produce the desired measure. The Section begins by using the formulae from the
previous sections to define the hypothetical Fourier–Stieltjes coefficients, defined on ZN which
is the dual group of the torus, a multiplicative group, and then applying a matrix version
of a theorem of Bochner about positive-definite functions to get the measure. There is an
application of approximate-identity theory using a Cesa`ro kernel to construct a sequence of
positive matrix-valued Laurent polynomials which converges to the orthogonality measure.
Section 5 develops a recurrence relation satisfied by the Fourier–Stieltjes coefficients of the
orthogonality measure. The relation allows an inductive calculation for the coefficients (but
actual work, even with symbolic computation software, may not be feasible unless the dimensions
are reasonably small), and it describes the list of parameter values (certain rational numbers)
for which the construction fails.
The scalar weight function on the torus
Kκ(x) :=
∏
1≤i<j≤N
{
(xi − xj)
(
x−1i − x
−1
j
)}κ
satisfies a first-order differential system,
xi
∂
∂xi
Kκ(x) = κK(x)
∑
j 6=i
xi + xj
xi − xj
, 1 ≤ i ≤ N.
Orthogonality Measure on the Torus for Vector-Valued Jack Polynomials 3
In Section 6 we show that there is an analogous matrix differential system which is solved in a
distribution sense by the orthogonality measure. We outline a result asserting that the orthog-
onality measure restricted to the complement of
⋃
1≤i<j≤N
{x : xi = xj} is equal to an analytic
solution of the differential system times the Haar measure dm. Finally there is Appendix A
containing some technical background results.
2 Vector-valued Jack polynomials and the Yang–Baxter graph
This is a summary of the definitions and results from [3] and [5]. For x = (x1, . . . , xN ) ∈ CN
the monomial xα :=
N∏
i=1
xαii , α = (α1, . . . , αN ) ∈ N
N
0 , |α| :=
N∑
i=1
αi, N0 := {0, 1, 2, 3, . . .}
and α is called a multi-index or a composition of |α|. We denote two distinguished elements
by 0 = (0, 0, . . . , 0), and 1 = (1, 1, . . . , 1). The degree of xα is |α|, and a polynomial is a finite
linear combination of monomials. The linear space of all polynomials is denoted by P, and
Pn = span{xα : |α| = n} is the subspace of polynomials homogeneous of degree n. The specific
polynomials considered here have coefficients in Q(κ) where κ is transcendental (indeterminate)
but which will also take on certain real values. The multi-indices α have an important partial
order: let α+ denote the nonincreasing rearrangement of α, for example if α = (1, 2, 1, 4) then
α+ = (4, 2, 1, 1). Let NN,+0 denote the set of partition multi-indices, that is,
{
λ ∈ NN0 : λ1 ≥
λ2 ≥ · · · ≥ λN
}
.
Definition 2.1.
α ≺ β ⇐⇒
i∑
j=1
αj ≤
i∑
j=1
βj , 1 ≤ i ≤ N, α 6= β,
α C β ⇐⇒ (|α| = |β|) ∧
[
(α+ ≺ β+) ∨ (α+ = β+ ∧ α ≺ β)
]
.
For example (3, 2, 1) C (0, 2, 4) C (4, 0, 2), while (4, 1, 1), (3, 3, 0) are not C-comparable. The
symmetric group SN , the set of permutations of {1, 2, . . . , N}, acts on CN by permutation of
coordinates. The action is extended to polynomials by wp(x) = p(xw) where (xw)i = xw(i)
(consider x as a row vector and w as a permutation matrix, [w]ij = δi,w(j), then xw = x[w]).
This is a representation of SN , that is, w1(w2p)(x) = (w2p)(xw1) = p(xw1w2) = (w1w2)p(x) for
all w1, w2 ∈ SN .
Furthermore SN is generated by reflections in the mirrors {x : xi−xj = 0} for 1 ≤ i < j ≤ N .
These are transpositions, denoted by (i, j), interchanging xi and xj . Define the SN -action on
α ∈ NN0 so that (xw)
α = xwα
(xw)α =
N∏
i=1
xαiw(i) =
N∏
j=1
x
αw−1(j)
j ,
that is (wα)i = αw−1(i) (take α as column vector, then wα = [w]α).
The simple reflections si := (i, i+ 1), 1 ≤ i ≤ N − 1, generate SN . They are the key devices
for applying inductive methods, and satisfy the braid relations:
sisj = sjsi, |i− j| ≥ 2;
sisi+1si = si+1sisi+1.
We consider the situation where the group SN acts on the range as well as on the domain
of the polynomials. We use vector spaces (called SN -modules) on which SN has an irreducible
unitary (orthogonal) representation: τ : SN → Om(R) (τ(w)−1 = τ(w−1) = τ(w)T ). See James
4 C.F. Dunkl
and Kerber [9] for representation theory, including a modern discussion of Young’s methods. We
will specify an orthogonal basis and the images τ(si) for each i, which suffices for our purposes.
Identify τ with a partition of N : (τ1, τ2, . . .) ∈ N
N,+
0 such that |τ | = N . The length of τ is
`(τ) = max{i : τi > 0}. There is a Ferrers diagram of shape τ (this diagram is given the same
name), with boxes at points (i, j) with 1 ≤ i ≤ `(τ) and 1 ≤ j ≤ τi. A tableau of shape τ is
a filling of the boxes with numbers, and a reverse standard Young tableau (RSYT) is a filling with
the numbers {1, 2, . . . , N} so that the entries decrease in each row and each column. We exclude
the one-dimensional representations corresponding to one-row (N) or one-column (1, 1, . . . , 1)
partitions, that is, we require dimVτ ≥ 2. The hook-length of the node (i, j) ∈ τ is defined to be
hook(τ ; i, j) := τi − j + #
{
k : i < k ≤ `(τ) ∧ j ≤ τk
}
+ 1.
We will need the key quantity hτ := hook(τ ; 1, 1) = τ1 + `(τ)− 1, the maximum hook-length of
the diagram.
Example 2.2. Here are the Ferrers diagram, a (column-strict) tableau, and an RSYT, all of
shape (5, 3, 2)
    
  
 
,


0 0 1 2 3
1 2 2
2 4

 ,


10 7 4 2 1
9 6 3
8 5

 .
Denote the set of RSYT’s of shape τ by Y(τ) and let Vτ := span{T : T ∈ Y(τ)} (the field
is C(κ)) with orthogonal basis Y(τ). Furthermore dimVτ = #Y(τ) = N !/
∏
(i,j)∈τ
hook(τ ; i, j).
For 1 ≤ i ≤ N and T ∈ Y(τ) the entry i is at coordinates (rw(i, T ), cm(i, T )) and the content
is c(i, T ) := cm(i, T ) − rw(i, T ). Each T ∈ Y(τ) is uniquely determined by its content vector
[c(i, T )]Ni=1. For the example τ = (3, 1)
[
4 2 1
3
]
,
[
4 3 1
2
]
,
[
4 3 2
1
]
the list of content vectors is [2, 1,−1, 0], [2,−1, 1, 0], [−1, 2, 1, 0]. To recover T from its content
vector fill in the entries starting with N , then N − 1 (c(N − 1, T ) = ±1) has two possibilities
and so on.
Example 2.3. The list of Y(τ) for τ = (3, 1, 1), N = 5


5 2 1
4
3

 ,


5 3 1
4
2

 ,


5 3 2
4
1

 ,


5 4 1
3
2

 ,


5 4 2
3
1

 ,


5 4 3
2
1

 .
The corresponding list of content vectors is [2, 1,−2,−1, 0], [2,−2, 1,−1, 0], [−2, 2, 1,−1, 0],
[2,−2,−1, 1, 0], [−2, 2,−1, 1, 0], [−2,−1, 2, 1, 0].
The representation theory can be developed using the content vectors in place of tableaux;
this is due to Okounkov and Vershik [14].
2.1 Description of the representation τ
The formulae for the action of τ(si) on the basis Y(τ) are from Murphy [11, Theorem 3.12].
Define bi(T ) := 1/(c(i, T ) − c(i + 1, T )). Note that c(i, T ) − c(i + 1, T ) = 0 is impossible for
RSYT’s. If |c(i, T ) − c(i + 1, T )| ≥ 2 let T (i) ∈ Y(τ) denote T with i, i + 1 interchanged. The
following describes the action of τ(si) (in each case there is an informal subrectangle description
of the relative positions of i and i+ 1 in T ; in cases (3) and (4) i and i+ 1 are not necessarily
in adjacent rows or columns)
Orthogonality Measure on the Torus for Vector-Valued Jack Polynomials 5
1. If rw(i, T ) = rw(i+ 1, T ) then τ(si)T = T ; position is [i+ 1, i], bi(T ) = 1.
2. If cm(i, T ) = cm(i+ 1, T ) then τ(si)T = −T ; position is
[
i+ 1
i
]
, bi(T ) = −1.
3. if rw(i, T ) < rw(i + 1, T ) (then cm(i, T ) > cm(i + 1, T )), position
[
∗ i
i+ 1 ∗
]
, c(i, T ) ≥
(cm(i+ 1, T ) + 1)− (rw(i+ 1, T )− 1) ≥ c(i+ 1, T ) + 2, 0 < bi(T ) ≤ 12 then
τ(si)T = T
(i) + bi(T )T,
τ(si)T
(i) =
(
1− bi(T )
2)T − bi(T )T
(i).
4. if rw(i, T ) > rw(i + 1, T ) (and cm(i, T ) < cm(i + 1, T )), position
[
∗ i+ 1
i ∗
]
; the formula
is found in case (3) interchanging T and T (i), and using bi(T ) = −bi(T (i)).
To eliminate extra parentheses we will write τ(i, j) for τ((i, j)); where (i, j) is a transposition.
There is a (unique up to constant multiple) positive Hermitian form on Vτ for which τ is
unitary (real orthogonal), that is 〈τ(w)S1, S2〉0 = 〈S1, τ(w)−1S2〉0 = 〈S1, τ(w)∗S2〉0, (S1, S2 ∈
Vτ , w ∈ SN ):
Definition 2.4.
〈
T, T ′
〉
0 := δT,T ′ ×
∏
1≤i<j≤N,
c(i,T )≤c(j,T )−2
(
1−
1
(c(i, T )− c(j, T ))2
)
, T, T ′ ∈ Y(τ).
The verification of the unitary property is based on the relation
〈
T (i), T (i)
〉
0 =
(
1− bi(T )
2)〈T, T 〉0
when 0 < bi(T ) ≤ 12 . Each τ(w) is an orthogonal matrix with respect to the orthonormal
basis
{
〈T, T 〉−1/20 T : T ∈ Y(τ)
}
. The basis vectors T are simultaneous eigenvectors of the
(reverse) Jucys–Murphy elements ωi :=
N∑
j=i+1
(i, j) (with ωN = 0), which commute pairwise
and τ(ωi)T = c(i, T )T, for 1 ≤ i ≤ N (see [11, Lemma 3.6]); as usual, τ is extended to
a homomorphism of the group algebra CSN by τ (
∑
w bww) =
∑
w bwτ(w).
2.2 Vector-valued nonsymmetric Jack polynomials
The main concern of this paper is Pτ = P ⊗ Vτ , the space of Vτ valued polynomials in x, which
is equipped with the SN action:
w
(
xα ⊗ T
)
= (xw)α ⊗ τ(w)T, α ∈ NN0 , T ∈ Y(τ),
extended by linearity to
wp(x) = τ(w)p(xw), p ∈ Pτ .
Definition 2.5. The Dunkl and Cherednik–Dunkl operators are (1 ≤ i ≤ N , p ∈ Pτ )
Dip(x) := ∂ip(x) + κ
∑
j 6=i
τ(i, j)
p(x)− p(x(i, j))
xi − xj
,
Uip(x) := Di(xip(x))− κ
i−1∑
j=1
τ(i, j)p(x(i, j)).
6 C.F. Dunkl
The commutation relations analogous to the scalar case hold, that is,
DiDj = DjDi, UiUj = UjUi, 1 ≤ i, j ≤ N,
wDi = Dw(i)w, ∀w ∈ SN , sjUi = Uisj , j 6= i− 1, i,
siUisi = Ui+1 + κsi, Uisi = siUi+1 + κ, Ui+1si = siUi − κ.
The simultaneous eigenfunctions of {Ui} are called (vector-valued) nonsymmetric Jack polyno-
mials (NSJP). For generic κ these eigenfunctions form a basis of Pτ (we will specify the excluded
rational values in the sequel). They have a triangularity property with respect to the partial
order B. However the structure does not merely rely on leading terms of the type xα ⊗ T . We
need the rank function:
Definition 2.6. For α ∈ NN0 , 1 ≤ i ≤ N
rα(i) := #{j : αj > αi}+ #{j : 1 ≤ j ≤ i, αj = αi},
then rα ∈ SN .
A consequence is that rαα = α+, the nonincreasing rearrangement of α, for any α ∈ NN0 . For
example if α = (1, 2, 1, 4) then rα = [3, 2, 4, 1] and rαα = α+ = (4, 2, 1, 1) (recall wαi = αw−1(i)).
Also rα = I if and only if α is a partition (α1 ≥ α2 ≥ · · · ≥ αN ).
For each α ∈ NN0 and T ∈ Y(τ) there is a NSJP ζα,T with leading term x
α⊗τ
(
r−1α
)
T , that is,
ζα,T = x
α ⊗ τ
(
r−1α
)
T +
∑
αBβ
xβ ⊗ tαβ(κ), tαβ(κ) ∈ Vτ ,
Uiζα,T =
(
αi + 1 + κc(rα(i), T )
)
ζα,T , 1 ≤ i ≤ N.
2.3 The Yang–Baxter graph
The NSJP’s can be constructed by means of a Yang–Baxter graph. The details are in [5]; this
paper has several figures illustrating some typical graphs.
A node consists of
(α, T, ξα.T , rα, ζα,T ),
where α ∈ NN0 , ξα,T is the spectral vector ξα,T (i) = αi + 1 + κc(rα(i), T ), 1 ≤ i ≤ N . The root
is
(
0, T0, [1 + κc(i, T0)]Ni=1, I, 1⊗ T0
)
where T0 is formed by entering N,N − 1, . . . , 1 column-by-
column in the Ferrers diagram, for example τ = (3, 3, 1)
T0 =


7 4 2
6 3 1
5

 , c(·, T0) = [1, 2, 0, 1,−2,−1, 0].
There is an adjacency relation in Y(τ) based on the positions of the pairs {i, i + 1} and an
inversion counter.
Definition 2.7. For T ∈ Y(τ) set
inv(T ) := #
{
(i, j) : i < j, c(i, T )− c(j, T ) ≤ −2
}
.
Recall from Section 2.1 that there are four types of positions of a given pair {i, i + 1} in T ,
and in case (3) it is straightforward to check that inv(T (i)) = inv(T ) + 1.
Orthogonality Measure on the Torus for Vector-Valued Jack Polynomials 7
If αi 6= αi+1 then rsiα = rαsi. The cycle w0 := (123 . . . N) and the affine transformation
Φ(a1, a2, . . . , aN ) := (a2, a3, . . . , aN , a1 + 1)
are fundamental parts of the construction; and rΦα = rαw0 for any α, that is,
rαw0(i) = rα(w0(i)) = rα(i+ 1) = rΦα(i), 1 ≤ i < N,
rαw0(N) = rα(w0(N)) = rα(1) = rΦα(N).
The jumps in the graph, which raise the degree by one, are
(α, T, ξα,T , rα, ζα,T )
Φ
−→
(
Φα, T,Φξα,T , rαw0, xNw
−1
0 ζα,T
)
, (2.1)
ζΦα,T = xNw
−1
0 ζα,T
the leading term is xΦα ⊗ τ
(
w−10 r
−1
α
)
T and w−10 r
−1
α = (rαw0)
−1. For example: α = (0, 3, 5, 0),
rα = [3, 2, 1, 4], Φα = (3, 5, 0, 1), rΦα = [2, 1, 4, 3].
There are two types of steps, labeled by si:
1. If αi < αi+1, then
(α, T, ξα,T , rα, ζα,T )
si−→ (siα, T, siξα,T , rαsi, ζsiα,T ),
ζsiα,T = siζα,T −
κ
ξα,T (i)− ξα,T (i+ 1)
ζα,T .
Observe that this construction is valid provided ξα,T (i) 6= ξα,T (i+ 1), that is, αi+1 − αi 6=
κ(c(rα(i), T ) − c(rα(i + 1), T )). The extreme values of c(·, T ) are τ1 − 1 and 1 − `(τ),
thus |c(rα(i), T ) − c(rα+1(i), T )| ≤ hτ − 1. Furthermore αi+1 − αi ≥ 1 and the step
is valid provided κm /∈ {1, 2, 3, . . . } for m = 1 − hτ , 2 − hτ , . . . , hτ − 1. The bound
−1/(hτ − 1) < κ < 1/(hτ − 1) is sufficient.
2. If αi = αi+1, and the positions of j := rα(i), j + 1 in T are of type (3), that is, c(j, T ) −
c(j + 1, T ) ≥ 2 (the definition of rα implies rα(i+ 1) = j + 1 and sir−1α = r
−1
α sj). Set
b′ =
1
c(j, T )− c(j + 1, T )
=
κ
ξα,T (i)− ξα,T (i+ 1)
;
thus 0 < b′ ≤ 12 , there is a step
(α, T, ξα,T , rα, ζα,T )
si−→
(
α, T (j), siξα,T , rα, ζα,T (j)
)
,
ζα,T (j) = siζα,T − b
′ζα,T ,
(T (j) is the result of interchanging j and j + 1 in T ). The leading term is transformed
si
(
xα ⊗ τ
(
r−1α
)
T
)
= (xsi)α ⊗ τ
(
sir−1α
)
T = xα ⊗ τ
(
r−1α
)
τ(sj)T and τ(sj)T = T (j) + b′T .
There are two other possibilities corresponding to (1) and (2) for the action of si on ζα,T when
αi = αi+1 (note rα(i + 1) = rα(i) + 1): (1) rw(rα(i), T ) = rw(rα(i) + 1, T ), then siζα,T = ζα,T ;
(2) cm(rα(i), T ) = cm(rα(i) + 1, T ), then siζα,T = −ζα,T .
The proofs that these formulae are mutually compatible for different paths in the graph from
the root (0, T0) to a given node (α, T ), use inductive arguments based on the fact that these
paths have the same length. The number of jumps is clearly |α| and the number of steps is
S(α) + inv(T )− inv(T0), where
S(α) :=
1
2
∑
1≤i<j≤N
(|αi − αj |+ |αi − αj + 1| − 1).
8 C.F. Dunkl
3 Hermitian forms
For a complex vector space V a Hermitian form is a mapping 〈·, ·〉 : V ⊗ V → C such that
〈u, cv〉 = c〈u, v〉, 〈u, v1 + v2〉 = 〈u, v1〉 + 〈u, v2〉 and 〈u, v〉 = 〈v, u〉 for u, v1, v2 ∈ V , c ∈ C.
The form is positive semidefinite if 〈u, u〉 ≥ 0 for all u ∈ V . The concern of this paper is with
a particular Hermitian form on Pτ which has the properties (for all f, g ∈ Pτ ):
〈
1⊗ T, 1⊗ T ′
〉
=
〈
T, T ′
〉
0, T, T
′ ∈ Y(τ), (3.1)
〈wf,wg〉 = 〈f, g〉, w ∈ SN ,
〈xiDif, g〉 = 〈f, xiDig〉, 1 ≤ i ≤ N.
The commutation Ui = Dixi − κ
∑
j<i
(i, j) = xiDi + 1 + κ
∑
j>i
(i, j) together with 〈(i, j)f, g〉 =
〈f, (i, j)g〉 show that 〈Uif, g〉 = 〈f,Uig〉 for all i. Thus the uniqueness of the spectral vectors dis-
cussed above implies that 〈ζα,T , ζβ,T ′〉 = 0 whenever (α, T ) 6= (β, T ′). In particular polynomials
homogeneous of different degrees are mutually orthogonal, by the basis property of {ζα,T }. We
can deduce contiguity relations corresponding to the steps described above and implied by the
properties of the form. Consider step type (1) with
siζα,T = ζsiα,T + b
′ζα,T , b
′ =
κ
ξα,T (i)− ξα,T (i+ 1)
.
The conditions 〈siζα,T , siζα,T 〉 = 〈ζα,T , ζα,T 〉 and 〈ζα,T , ζsiα,T 〉 = 0 imply
〈ζα,T , ζα,T 〉 =
〈
ζsiα,T + b
′ζα,T , ζsiα,T + b
′ζα,T
〉
= 〈ζsiα,T , ζsiα,T 〉+ b
′2〈ζα,T , ζα,T 〉,
〈ζsiα,T , ζsiα,T 〉 =
(
1− b′2
)
〈ζα,T , ζα,T 〉.
A necessary condition that the form be positive-definite (f 6= 0 implies 〈f, f〉 > 0) is that
−1 < b′ < 1 in each of the possible steps. Since (with j = rα(i) and ` = rα(i+ 1))
1− b′2 =
[αi+1 − αi + (c(`, T )− c(j, T ) + 1)κ][αi+1 − αi + (c(`, T )− c(j, T )− 1)κ]
[αi+1 − αi + (c(`, T )− c(j, T ))κ]2
,
the extreme values of (c(`, T )− c(j, T )± 1) are ±hτ , and αi+1−αi ≥ 1, it follows that −1/hτ <
κ < 1/hτ implies 1− b′2 > 0. Since steps of type (1) link any (α, T ) to (α+, T ) one can obtain
(with ε = ±1)
Eε(α, T ) :=
∏
1≤i<j≤N
αi<αj
(
1 +
εκ
αj − αi + κ(c(rα(j), T )− c(rα(i), T ))
)
, (3.2)
〈ζα,T , ζα,T 〉 =
(
E1(α, T )E−1(α, T )
)−1
〈ζα+,T , ζα+,T 〉.
Similarly the steps of type (2) (with αi = αi+1 and j = rα(i), b′ = 1c(j,T )−c(j+1,T )) imply the
relation
〈
ζα,T (j) , ζα,T (j)
〉
=
(
1− b′2
)
〈ζα,T , ζα,T 〉.
It was shown in [3] (this is a special case of a result of Griffeth [8, Theorem 6.1]) that the
definition for λ ∈ NN,+0
〈ζλ,T , ζλ,T 〉 = 〈T, T 〉0
N∏
i=1
(1 + κc(i, T ))λi
∏
1≤i<j≤N
λi−λj∏
`=1
(
1−
(
κ
`+ κ(c(i, T )− c(j, T ))
)2
)
.
Orthogonality Measure on the Torus for Vector-Valued Jack Polynomials 9
together with formula (3.2) produce a Hermitian form (called the covariant form) satisfying (3.1)
and the additional property 〈xif, g〉 = 〈f,Dig〉 for all f, g ∈ Pτ and 1 ≤ i ≤ N (the bound
−1/hτ < κ < 1/hτ for positivity of this form was found by Etingof and Stoica [7]).
Here we want a Hermitian form for which multiplication by any xi is an isometry, that is,
〈xif, xig〉 = 〈f, g〉 for all f, g ∈ Pτ and 1 ≤ i ≤ N . Heuristically this should involve an integral
over the N -torus. The isometry postulate, and the equations (3.1) determine the form uniquely,
as will be shown. The postulate 〈xNf, g〉 = 〈f,DNg〉 in the covariant form is used to compute the
effect of a jump ζΦα,T = xNw
−1
0 ζα,T (see (2.1)), that is, to evaluate 〈ζΦα,T , ζΦα,T 〉/〈ζα,T , ζα,T 〉.
From the proofs in [3, Appendix, Corollary 5, Theorem 10] and [5] we see that the factor
N∏
i=1
(1 + κc(i, T ))λi arises from ratios of this type. This aspect (here we need the ratio to be 1)
motivates the following:
Definition 3.1. For λ ∈ NN,+0 , α, β ∈ N
N
0 , and T, T
′ ∈ Y(τ) the Hermitian form 〈·, ·〉T on Pτ is
specified by
(α, T ) 6= (β, T ′) =⇒ 〈ζα,T , ζβ,T 〉T = 0,
〈ζλ,T , ζλ,T 〉T = 〈T, T 〉0
∏
1≤i<j≤N
λi−λj∏
`=1
(
1−
(
κ
`+ κ(c(i, T )− c(j, T ))
)2
)
,
〈ζα,T , ζα,T 〉T =
(
E1(α, T )E−1(α, T )
)−1
〈ζα+,T , ζα+,T 〉T;
the form is extended to all of Pτ by linearity in the second variable and Hermitian symmetry,
that is, 〈f, c1g+c2h〉T = c1〈f, g〉T+c2〈f, h〉T and 〈f, g〉T = 〈g, f〉T, for f, g, h ∈ Pτ and c1, c2 ∈ C.
Observe that the formula is invariant when λ is replaced by λ+m 1 = (λ1 +m,λ2 +m, . . .,
λN +m) for any m ∈ N. This follows easily from the commutation (where eN := x1x2 · · ·xN )
Ui
(
emNf
)
= memNf + e
m
NUif, 1 ≤ i ≤ N, m = 1, 2, . . . ,
thus Ui(emNζα,T ) = (m + αi + κc(rα(i), T ))e
m
Nζα,T , and e
m
Nζα,T is a simultaneous eigenfunction
of {Ui} with the same eigenvalues and the same leading term as ζα+m1,T . Hence ζα+m1,T =
emNζα,T . We now extend the structure of NSJP’s to Vτ -valued Laurent polynomials, thereby
producing a basis:
Definition 3.2. Suppose α ∈ ZN then set ζα,T = e−mN ζα+m1,T where m ∈ N0 and satisfies
m ≥ −minj αi. This is valid since α + m1 ∈ NN0 and by the relation ζβ+k1,T = e
k
Nζβ,T for
β ∈ NN0 and k ∈ N0.
The proof that the form satisfies the properties (3.1) with respect to steps is the same as the
one in [3, Propositions 8 and 9], and it suffices to verify the effect of a jump.
Theorem 3.3. Suppose α ∈ NN,+0 then 〈ζΦα,T , ζΦα,T 〉T = 〈ζα,T , ζα,T 〉T.
Proof. We use (2.1) to relate 〈ζΦα,T , ζΦα,T 〉T to 〈ζβ,T , ζβ,T 〉T where β = (Φα)+ = (α1 + 1, α2,
. . . , αN ). The product Eε(Φα, T ) is over the pairs αi = (Φα)i−1 < (Φα)N = α1+1 for 2 ≤ i ≤ N ,
thus
Eε(Φα, T ) =
N∏
i=2
(
1−
εκ
α1 + 1− αi + κ(c(rΦα(N), T )− c(rΦα(i− 1), T ))
)
=
N∏
i=2
(
1−
εκ
α1 + 1− αi + κ(c(1, T )− c(i, T ))
)
.
10 C.F. Dunkl
By definition
〈ζΦα,T , ζΦα,T 〉T =
(
E1(Φa, T )E−1(Φα, T )
)−1
〈ζβ,T , ζβ,T 〉T
=
N∏
i=2
(
1−
(
κ
α1 + 1− αi + κ(c(1, T )− c(i, T ))
)2
)−1
× 〈T, T 〉0
∏
1≤i<j≤N
βi−βj∏
`=1
(
1−
(
κ
`+ κ(c(i, T )− c(j, T ))
)2
)
= 〈T, T 〉0
∏
2≤i<j≤N
αi−αj∏
`=1
(
1−
(
κ
`+ κ(c(i, T )− c(j, T ))
)2
)
×
N∏
j=2
α1−αj∏
`=1
(
1−
(
κ
`+ κ(c(1, T )− c(j, T ))
)2
)
= 〈ζα,T , ζα,T 〉T.
The terms in the product for i = 1 and 2 ≤ j ≤ N , ` = β1 − βj = α1 + 1 − αj are canceled
out. 
We summarize the key results. We say κ is generic if (α, T ) 6= (β, T ′) implies the spectral
vectors ξα,T 6= ξβ,T ′ .
Proposition 3.4. For generic κ the Hermitian form 〈·, ·〉T satisfies
1) if f , g are homogeneous and deg f 6= deg g then 〈f, g〉T = 0,
2) 〈wf,wg〉T = 〈f, g〉T, f, g ∈ Pτ , w ∈ SN ,
3) 〈xiDif, g〉T = 〈f, xiDig〉T for f, g ∈ Pτ and 1 ≤ i ≤ N ,
4) 〈xif, xig〉T = 〈f, g〉T for f, g ∈ Pτ and 1 ≤ i ≤ N .
Proof. For generic κ the NSJP’s ζα,T with |α| = n form a basis for Pτ,n; this immediately
implies (1). For (2) the fact that 〈siζα,T , siζβ,T ′〉T = 〈ζα,T , ζβ,T ′〉T for 1 ≤ i < N follows from
the corresponding results in [3, Propositions 8 and 9, Corollary 3] when |α| = |β|, otherwise
from Definition 3.1. This suffices for (2) since {si} generates SN . The definition of NSJP’s
implies trivially that 〈Uiζα,T , ζβ,T ′〉T = 〈ζα,T ,Uiζβ,T ′〉T for all i and (α, T ), (β, T ′) because both
sides vanish if (α, T ) 6= (β, T ′), otherwise equal ξα,T (i)〈ζα,T , ζα,T 〉T. The commutation Ui =
Dixi − κ
∑
j<i
(i, j) = xiDi + 1 + κ
∑
j>i
(i, j) together with 〈(i, j)f, g〉T = 〈f, (i, j)g〉T from (2)
show that 〈xiDif, g〉T = 〈f, xiDig〉T for 1 ≤ i ≤ N . For part (4) (recall w0 = (123 . . . N))
ζΦα,T = xNw
−1
0 ζα,T and ζΦβ,T ′ = xNw
−1
0 ζβ,T ′ . By Theorem 3.3 〈ζΦα,T , ζΦβ,T ′〉T = 〈ζα,T , ζβ,T ′〉T
(if (α, T ) 6= (β, T ′) then (Φα, T ) 6= (Φβ, T ′)). Thus for each (α, T ), (β, T ′)
〈
xN
(
w−10 ζα,T
)
, xN
(
w−10 ζβ,T ′
)〉
T = 〈ζα,T , ζβ,T ′〉T =
〈
w−10 ζα,T , w
−1
0 ζβ,T ′
〉
T.
The set
{
w−10 ζα,T : (α, T )
}
is a basis for Pτ thus 〈xNf, xNg〉T = 〈f, g〉T for all f, g ∈ Pτ . For
any i
〈xif, xig〉T = 〈(i,N)xif, (i,N)xig〉T = 〈xN (i,N)f, xN (i,N)g〉T
= 〈(i,N)f, (i,N)g〉T = 〈f, g〉T;
and this completes the proof. 
This lays the abstract foundation for the next developments.
Orthogonality Measure on the Torus for Vector-Valued Jack Polynomials 11
4 Fourier–Stieltjes coefficients on the torus
The torus TN :=
{
x ∈ CN : |xi| = 1, 1 ≤ i ≤ N
}
is a multiplicative compact abelian group with
dual group ZN . We will use this property to find the measure of orthogonality for the NSJP’s
on the torus. First we produce the Fourier–Stieltjes coefficients of the hypothetical measure
and then use a matrix version of a theorem of Bochner to deduce the existence of the measure.
When κ is generic the NSJP’s form a basis for Pτ and it is possible to make the definition
A˜
(
α, β, T, T ′
)
:=
(
〈T, T 〉0
〈
T ′, T ′
〉
0
)−1/2〈
xα ⊗ T, xβ ⊗ T ′
〉
T
for α, β ∈ NN0 and T, T
′ ∈ Y(τ). In effect this uses the orthonormal basis of Vτ . By the
symmetry of the form A˜(α, β, T, T ′) = A˜(β, α, T ′, T ). By Proposition 3.4 |α| 6= |β| implies
A˜(α, β, T, T ′) = 0. Another consequence is A˜(0,0, T, T ′) = δT,T ′ .
Definition 4.1. For each γ ∈ ZN with
∑N
i=1 γi = 0 let γ
pi
i = max(γi, 0) and γ
ν
i = −min(γi, 0) for
1 ≤ i ≤ N ; then γ = γpi−γν and γpi, γν ∈ NN0 . Furthermore |γ
pi| = |γν | and
∑
i |γi| = |γ
pi|+ |γν |
is even.
Introduce the index set ZN and its graded components by
ZN :=
{
α ∈ ZN :
N∑
i=1
αi = 0
}
,
ZN,n :=
{
α ∈ ZN :
N∑
i=1
|αi| = 2n
}
, n = 0, 1, 2, . . . .
Formula (A.1) for #ZN,n is in Appendix A.
Definition 4.2. For γ ∈ ZN the matrix Aγ (of size #Y(τ)×#Y(τ)) is given by
(Aγ)T,T ′ = A˜
(
γpi, γν , T, T ′
)
, T, T ′ ∈ Y(τ), γ ∈ ZN ,
Aγ = 0, γ /∈ ZN .
Proposition 4.3. Suppose α, β ∈ NN0 and T, T
′ ∈ Y(τ) then A˜(α, β, T, T ′) = (Aα−β)T,T ′.
Proof. If |α| 6= |β| then
N∑
i=1
(αi−βi) 6= 0, A˜(α, β, T, T ′) = 0 and Aα−β = 0 by definition. If |α| =
|β| let ζi = min(αi, βi) for 1 ≤ i ≤ N then xζ is a factor of both xα and xβ; by Proposition 3.4
〈xα⊗T, xβ⊗T ′〉T = 〈xα−ζ⊗T, xβ−ζ⊗T ′〉T . By construction (α−β)pi = α−ζ, (α−β)ν = β−ζ.
It follows that
〈
xα ⊗ T, xβ ⊗ T ′
〉
= T ∗Aα−βT
′ =
(
〈T, T 〉0
〈
T ′, T ′
〉
0
)1/2
(Aα−β)T,T ′ . 
For a formal Laurent series h(x) =
∑
α∈ZN
cαxα let CT(h(x)) = c0, the constant term. Then
〈
xα ⊗ T, xβ ⊗ T ′
〉
=
(
〈T, T 〉0
〈
T ′, T ′
〉
0
)1/2
CT
(
x−α
∑
γ∈ZN
(Aγ)T,T ′x
γxβ
)
.
In the next section we investigate analytical properties of the formal series, but first we consider
algebraic properties, that is, those not needing any convergence results.
Theorem 4.4. Suppose γ ∈ ZN and w ∈ SN then A−γ = A∗γ and Awγ = τ(w)Aγτ
(
w−1
)
.
12 C.F. Dunkl
Proof. The relation A˜(α, β, T, T ′) = A˜(β, α, T ′, T ) shows (Aα−β)T,T ′ = (Aβ−α)T ′,T . By defini-
tion
〈
w
(
xα ⊗ T
)
, w
(
xβ ⊗ T ′
)〉
T =
〈
xwα ⊗ τ(w)T, xwβ ⊗ τ(w)T ′
〉
T
=
(
〈T, T 〉0
〈
T ′, T ′
〉
0
)1/2
T ∗τ(w)∗Awα−wβτ(w)T
′
=
〈
xα ⊗ T, xβ ⊗ T ′
〉
T =
(
〈T, T 〉0
〈
T ′, T ′
〉
0
)1/2
T ∗Aα−βT
′
and thus Aγ = τ(w)−1Awγτ(w) (recall τ is real-orthogonal so τ(w)∗ = τ
(
w−1
)
). 
Summing over the graded components ZN,n produces Laurent polynomials with good prop-
erties, such as analyticity in (C\{0})N . The maps a 7→ wα (w ∈ SN ) and α 7→ −α act as
permutations on each ZN,n.
Definition 4.5. For n = 0, 1, 2, . . . let
Hn(x) :=
∑
α∈ZN,n
Aαx
α,
a Laurent polynomial with matrix coefficients.
For complex Laurent polynomials f(x) =
∑
α∈ZN
cαxα (finite sum) define f(x)∗ =
∑
α∈ZN
cαx−α;
if the coefficients {cα} are matrices then f(x)∗ =
∑
α∈ZN
c∗αx
−α. There is a slight abuse of notation
here: if x ∈ TN then (xα) = x−α and f(x)∗ agrees with the adjoint of the matrix f(x).
Proposition 4.6. Suppose n = 0, 1, 2, . . . and w ∈ SN then Hn(xw) = τ(w)−1Hn(x)τ(w) and
Hn(x)∗ = Hn(x).
Proof. Compute
Hn(xw) =
∑
α∈ZN,n
Aα(wx)
α =
∑
α∈ZN,n
Aαx
wα =
∑
β∈ZN,n
Aw−1βx
β
= τ
(
w−1
) ∑
β∈ZN,n
Aβx
βτ(w) = τ(w)−1Hn(x)τ(w).
Also Hn(x)∗ =
∑
α∈ZN,n
A∗αx
−α =
∑
α∈ZN,n
A−αx−α = Hn(x). 
As a consequence we find an important commutation satisfied by a particular point value
of Hn(x) (recall the N -cycle w0 = (1, 2, . . . , N)).
Corollary 4.7. Suppose n = 1, 2, 3, . . . then τ(w0)−1Hn(x0)τ(w0) = Hn(x0), where x0 =(
1, ω, . . . , ωN−1
)
, ω = exp 2piiN .
Proof. By definition x0w0 =
(
ω, . . . , ωN−1, 1
)
= ωx0. Each monomial xα for α ∈ ZN
is homogeneous of degree zero (suppose c ∈ C\{0} then (cx)α = cα1+···+αNxα = xα) thus
Hn(x0w0) = Hn(ωx0) = Hn(x0) and Hn(x0w0) = τ(w0)−1Hn(x0)τ(w0). 
We turn to the harmonic analysis significance of the matrices {Aα}. For an integrable func-
tion f on TN the Fourier transform (coefficient) is
f̂(α) =
∫
TN
f(x)x−αdm(x), α ∈ ZN ,
Orthogonality Measure on the Torus for Vector-Valued Jack Polynomials 13
where x := (exp(iθ1), . . . , exp(iθN )) and dm(x) = (2pi)−Ndθ1 · · · dθN ; and for a Baire measure µ
on TN the Fourier–Stieltjes transform is
µ̂(α) =
∫
TN
x−αdµ(x), α ∈ ZN .
We will show that there is a matrix-valued measure µ, positive in a certain sense, such that
µ̂(α) = Aα for all α ∈ ZN provided that −1/hτ < κ < 1/hτ . There is a version of a theorem of
Bochner about positive-definite functions on a locally compact abelian group which proves this
claim. The details of the proof and some consequences are in Appendix A.
Let n = dimVτ and identify Vτ with Cn whose elements are considered as column vectors
(in effect we use indices 1 ≤ i ≤ n instead of {T ∈ Y(τ)}). The inner product on Cn is
〈u, v〉 :=
n∑
i=1
uivi, and the norm is |v| =
√
〈v, v〉. Note that 〈u,Av〉 = u∗Av. A positive-definite
matrix P satisfies 〈u, Pu〉 ≥ 0 for all u ∈ Cn (this implies P ∗ = P ).
Definition 4.8. A function F : ZN →Mn(C) is positive-definite if
∑
α,β∈ZN
f(α)∗F (α− β)f(β) ≥ 0
for any finitely supported Cn-valued function f on ZN .
Theorem 4.9. Suppose F is positive-definite then there exist Baire measures {µjk : 1 ≤ j, k ≤ n}
on TN such that
∫
TN
x−αdµjk(x) = F (α)jk, α ∈ ZN , 1 ≤ j, k ≤ n.
Furthermore each µjj is positive and
〈f, g〉F :=
n∑
i,j=1
∫
TN
f(x)ig(x)jdµij(x)
defines a positive-semidefinite Hermitian form on C
(
TN ;Cn
)
(continuous Cn-valued functions
on TN ) satisfying |〈f, g〉F | ≤ B‖f‖∞‖g‖∞ for f, g ∈ C
(
TN ;Cn
)
with B <∞.
The proof is in Appendix A.1 and Theorem A.3. (In general the measures µjk are not
real-valued for j 6= k.) For notational simplicity we introduce
∫
TN
f(x)∗dµ(x)g(x) :=
n∑
i,j=1
∫
TN
f(x)ig(x)jdµij(x). (4.1)
To show that α 7→ Aα is positive-definite let f be a finitely supported Cn-valued function f
on ZN and let p(x) =
∑
α,T
〈T, T 〉−1/20 fT (α)x
α ⊗ T be the associated Laurent polynomial (now we
use the T indices on Cn). Because this is a finite sum there is a nonnegative integer m such
that emNp(x) is polynomial (no negative powers). Then for −1/hτ < κ < 1/hτ
0 ≤
〈
emNp, e
m
Np
〉
T =
∑
α,β∈ZN
∑
T,T ′
(
〈T, T 〉0
〈
T ′, T ′
〉
0
)−1/2
fT (α)fT ′(β)
〈
xα+m1 ⊗ T, xβ+m1 ⊗ T ′
〉
=
∑
α,β∈ZN
∑
T,T ′
fT (α)fT ′(β)(Aα−β)T,T ′ .
Let µ = [µT,T ′ ] be the matrix of measures produced by the theorem, that is,
(Aα)T,T ′ =
∫
TN
x−αdµT,T ′(x), α ∈ Z
N , T, T ′ ∈ Y(τ).
14 C.F. Dunkl
Theorem 4.10. For −1/hτ < κ < 1/hτ there exists a matrix of Baire measures µ = [µT,T ′ ]
on TN such that
〈f, g〉T =
∫
TN
f(x)∗dµ(x)g(x)
for all Laurent polynomials f , g with coefficients in Vτ , in particular for all NSJP’s f , g.
Of course we want more detailed information about these measures. The first step is to apply
an approximate identity, a tool from the convolution structure for measures and functions on
the torus. We consider Cesa`ro summation of the series
∑
αAαx
α based on summing first over
each ZN,n. Set
Sn(x) :=
∑
α∈ZN,n
xα,
a Laurent polynomial, and the corresponding (C, δ)-kernel (for δ > 0) is defined to be (the
Pochhammer symbol is (t)m =
m∏
i=1
(t+ i− 1))
σδn(x) :=
n∑
k=0
(−n)k
(−n− δ)k
Sk(x).
The point is that lim
n→∞
(−n)k
(−n−δ)k
= 1 for fixed k. In terms of convolution
σδn ∗ µ(x) =
∫
TN
σδn
(
xy−1
)
dµ(y),
̂(σδn ∗ µ)(α) =
∫
TN
∫
TN
x−ασδn
(
xy−1
)
dµ(y)dm(x)
=
∫
TN
∫
TN
(xy)−ασδn(x)dµ(y)dm(x) = Aασ̂δn(α),
and σ̂δn(α) =
(−n)k
(−n−δ)k
for α ∈ ZN,k for 0 ≤ k ≤ n and = 0 for |α| > 2n (or α /∈ ZN ). Thus
σδn ∗ µ(x) =
n∑
k=0
(−n)k
(−n−δ)k
Hk(x). In fact σN−1n (x) ≥ 0 for all x ∈ T
N (Corollary 4.12 below)
which implies σN−1n ∗ µ converges to µ in a useful sense (weak-∗, see Theorem 4.17(4)) and
σN−1n ∗µ(x) is a Laurent polynomial all of whose point values are positive-semidefinite matrices.
Also
∥
∥σN−1n
∥
∥
1 :=
∫
TN
∣
∣σN−1n
∣
∣dm = 1.
The complete symmetric polynomial in N variables and degree n is given by
hn(x) :=
∑
{
xα : α ∈ NN0 :
N∑
i=1
αi = n
}
.
Recall
#
{
α ∈ NN0 :
N∑
i=1
αi = m
}
=
(N)m
m!
for m = 0, 1, 2, 3, . . . .
Theorem 4.11. For n ≥ 0
hn
(
1
x1
,
1
x2
, . . . ,
1
xN
)
hn(x1, . . . , xN ) =
(N)n
n!
σN−1n (x). (4.2)
Orthogonality Measure on the Torus for Vector-Valued Jack Polynomials 15
Proof. The product is a sum of terms xα−β with α, β ∈ NN0 and |α| = n = |β|. For example the
term x0 = 1 appears exactly (N)nn! times, because the number of terms in hn is
(N)n
n! . Consider
a fixed γ ∈ ZN,m for some m with 0 ≤ m ≤ n. The term xγ appears in the product for each
pair (α, β) with
α = γpi + α′, β = γν + α′, γ = α− β,
where α′ ∈ NN0 and
N∑
i=1
α
′
i = n−m. Recall γ
pi
i = max(γi, 0) and γ
ν
i = −min(0, γi) = max(0,−γi);
thus
N∑
i=1
γpii = m and
N∑
i=1
αi = n. Therefore the coefficient of xγ is
#
{
α′ ∈ NN0 :
N∑
i=1
α
′
i = n−m
}
=
(N)n−m
(n−m)!
.
Hence
hn
(
1
x1
,
1
x2
, . . . ,
1
xN
)
hn(x1, . . . , xN ) =
N∑
m=0
(N)n−m
(n−m)!
Sm(x).
To finish the proof multiply this relation by n!(N)n and compute
n!
(N)n
(N)n−m
(n−m)!
= (−1)m(−n)m
(N)n−m
(N)n−m(N + n−m)m
= (−1)m
(−n)m
(N + n−m)m
=
(−n)m
(1−N − n)m
. 
Corollary 4.12. σN−1n (x) ≥ 0 for all x ∈ T
N .
Proof. hn(x1, . . . , xN ) = hn
(
1
x1
, 1x2 , . . . ,
1
xN
)
for x ∈ TN . 
Observe that this kernel applies to the quotient space TN/D where
D := {(u, u, . . . , u) : u ∈ C, |u| = 1},
is the diagonal subgroup. That is, each Sn(x) is homogeneous of degree zero, constant on sets
{(ux1, ux2, . . . , uxN ) : |u| = 1} for fixed x ∈ TN .
Here are approximate identity properties of σN−1n ; we use T
N/D to refer to functions homo-
geneous of degree zero. There is a standard formula:
Lemma 4.13. Suppose g, h ∈ C
(
TN
)
and ν is a Baire measure on TN then for h†(x) := h
(
x−1
)
∫
TN
g(x)(h ∗ ν)(x)dm(x) =
∫
TN
(
g ∗ h†
)
(y)dν(y).
Proof. The left side equals
∫
TN
∫
TN
g(x)h
(
xy−1
)
dν(y)dm(x) =
∫
TN
∫
TN
g(x)h†
(
yx−1
)
dm(x)dν(y)
(by Fubini’s theorem) which equals the right side. 
The following is a standard result on approximate identities.
16 C.F. Dunkl
Proposition 4.14. Suppose f ∈ C
(
TN/D
)
then
∥
∥f − f ∗ σN−1n
∥
∥
∞ → 0 as n→∞.
Proof. For ε > 0 there exists a Laurent polynomial p on TN/D such that ‖f − p‖∞ < ε. Then
f − f ∗ σN−1n = (f − p) +
(
p− σN−1n ∗ p
)
+ (p− f) ∗ σN−1n ,
and
∥
∥(p − f) ∗ σN−1n
∥
∥
∞ ≤ ‖p − f‖∞
∥
∥σN−1n
∥
∥
1 < ε. Let p(x) =
M∑
m=0
∑
α∈ZN,m
cαxα for some
coefficients cα (and finite M); thus
(
p− σN−1n ∗ p
)
(x) =
M∑
m=0
∑
α∈ZN,m
(
1−
(−n)m
(1−N − n)m
)
cαx
α,
which tends to zero in norm as n→∞. 
Corollary 4.15. Suppose ν is a Baire measure on TN and f ∈ C
(
TN/D
)
then
lim
n→∞
∫
TN
f(x)
(
σN−1n ∗ ν
)
(x)dm(x) =
∫
TN
f(x)dν(x).
Proof. By Lemma 4.13
∫
TN
f(x)
(
σN−1n ∗ ν
)
(x)dm(x) =
∫
TN
(
f ∗ σN−1n
)
(x)dν(x),
since
(
σN−1n
)†
= σN−1n , and f ∗ σ
N−1
n converges uniformly to f as n→∞. 
Definition 4.16. Define the µ-approximating Laurent polynomials
Kn(x) := σ
N−1
n ∗ µ(x) =
n∑
m=0
(−n)m
(1−N − n)m
∑
α∈ZN,n
Aαx
α.
Note (−n)m(1−N−n)m =
(n−m+1)N−1
(n+1)N−1
for 0 ≤ m ≤ n; for example with N = 3, (−n)m(−2−n)m =
(
1 −
m
n+1
)(
1− mn+2
)
, and = 0 for m > n.
Theorem 4.17. For −1/hτ < κ < 1/hτ and n = 1, 2, 3, . . . the following hold:
(1) Kn(x) is positive semi-definite for each x ∈ TN ,
(2) Kn(xw) = τ(w)−1Kn(x)τ(w) for each x ∈ TN , w ∈ SN ,
(3) Kn(x0)τ(w0) = τ(w0)Kn(x0),
(4) lim
n→∞
∫
TN f(x)
∗Kn(x)g(x)dm(x) = 〈f, g〉T for all f, g ∈ Pτ ; the limit exists for any f, g ∈
C
(
TN ;Vτ
)
and defines a ‖ · ‖∞-bounded positive Hermitian form.
Proof. Part (1) is a consequence of Theorem A.4. Parts (2) and (3) follow from the properties
of Hm in Proposition 4.6. For part (4) there is an intermediate step of averaging over the
diagonal group D. Define the operator ρ : C
(
TN
)
→ C
(
TN/D
)
by
ρ(p)(x) :=
1
2pi
∫ pi
−pi
p
(
eiθx
)
dθ, p ∈ C
(
TN
)
.
Orthogonality Measure on the Torus for Vector-Valued Jack Polynomials 17
Clearly ‖ρ(p)‖∞ ≤ ‖p‖∞; in effect ρ is the projection onto Fourier series supported by ZN .
Then
∫
TN
p(x)dµT,T ′(x) =
∫
TN
ρ(p)(x)dµT,T ′(x),
∫
TN
p(x)(Kn(x))T,T ′dm(x) =
∫
TN
ρ(p)(x)(Kn(x))T,T ′dm(x), T, T
′ ∈ Y(τ).
To extend this to the form 〈·, ·〉T express the typical sum
n∑
i,j=1
fiBijgj = tr
(
(f ⊗g∗)∗B
)
where tr
denotes the trace and (f ⊗g∗)ij = figj (1 ≤ i, j ≤ n). Then (ρ is applied to matrices entry-wise)
for f, g ∈ C
(
TN ;Vτ
)
∫
TN
f(x)∗dµ(x)g(x) =
∫
TN
tr
[
(f(x)⊗ g(x)∗)∗dµ(x)
]
=
∫
TN
tr
[
{ρ(f(x)⊗ g(x)∗)}∗dµ(x)
]
,
∫
TN
f(x)∗Kn(x)g(x)dm(x) =
∫
TN
tr
[
(f(x)⊗ g(x)∗)∗Kn(x)
]
dm(x)
=
∫
TN
tr
[
{ρ(f(x)⊗ g(x)∗)}∗Kn(x)
]
dm(x).
The convergence properties of Proposition 4.14 imply part (4). 
5 Recurrence relations
As a simple illustration consider ZN,1 where it suffices to find A1,−1,0...,0. Introduce the unit
basis vectors εi for ZN (with (εi)j = δij), so that (1,−1, 0, . . .) = ε1−ε2. The relation (ε2−ε1) =
(1, 2)(ε1 − ε2) implies Aε2−ε1 = τ(1, 2)Aε1−ε2τ(1, 2) = A
∗
ε1−ε2 . From 〈x1D1f, g〉T = 〈f, x1D1g〉T
(Proposition 3.4(3)) we find
x1D1(x1 ⊗ T ) = x1 ⊗ T + κx1
N∑
j=2
x1 − (x(1, j))1
x1 − xj
τ(1, j)T
= x1 ⊗ T + κx1 ⊗ τ(ω1)T = x1 ⊗ (I + κτ(ω1))T,
x1D1(x2 ⊗ T
′) = κx1
N∑
j=2
x2 − (x(2, j))1
x1 − xj
τ(1, j)T ′ = −κx1τ(1, 2)T
′;
recall the Jucys–Murphy elements ωi :=
N∑
j=i+1
(i, j) and the action τ(ωi)T = c(i, T )T for
T ∈ Y(τ). Next the equation 〈x1D1(x1 ⊗ T ), x2 ⊗ T ′〉T = 〈x1 ⊗ T, x1D1(x2 ⊗ T ′)〉T yields
T ∗(I + κτ(ω1))
∗Aε1−ε2T
′ = −κT ∗A0τ(1, 2)T
′,
and A0 = I. This holds for arbitrary T , T ′, and τ(ω1) is diagonal with the entry at (T, T ) being
c(1, T ) thus
(I + κτ(ω1))Aε1−ε2 = −κτ(1, 2), Aε1−ε2 = −κ(I + κτ(ω1))
−1τ(1, 2),
provided κc(1, T ) 6= −1 for all T ∈ Y(τ).
18 C.F. Dunkl
Lemma 5.1. For α ∈ NN0 , T ∈ Y(τ) and 1 ≤ i ≤ N
xiDi
(
xα ⊗ T
)
= αix
α ⊗ T − κ
∑
αj>αi
αj−αi∑
`=1
xα+`(εi−εj) ⊗ τ(i, j)T
+ κ
∑
αi>αj
αi−αj−1∑
`=0
xα+`(εj−εi) ⊗ τ(i, j)T.
Proof. This follows from performing the division in x
α−x(i,j)α
xi−xj
. 
Proposition 5.2. For α, β ∈ NN0 such that |α| = |β| and 1 ≤ i ≤ N
(αi − βi)Aα−β = κ
∑
αj>αi
αj−αi∑
`=1
τ(i, j)Aα+`(εi−εj)−β − κ
∑
αi>αj
αi−αj−1∑
`=0
τ(i, j)Aα+`(εj−εi)−β
− κ
∑
βj>βi
βj−βi∑
`=1
Aα−`(εi−εj)−βτ(i, j)
+ κ
∑
βi>βj
βi−βj−1∑
`=0
Aα−`(εj−εi)−βτ(i, j). (5.1)
Proof. The statement follows from the equation
〈
xiDi(x
α ⊗ T ), xβ ⊗ T ′
〉
=
〈
xα ⊗ T, xiDi
(
xβ ⊗ T ′
)〉
and the lemma. 
The following is one of the main results of this section. Note that it is important to involve
the multiplicity of the first part of γ.
Theorem 5.3. Suppose γ ∈ ZN,n such that γpi is a partition and γpi1 = γ
pi
m > γm+1 then
(
γ1I + κ
N∑
`=m+1
τ(1, `)
)
Aγ = −κ
N∑
j=m+1, γj≥0
γ1−γj−1∑
`=1
τ(1, j)Aγ+`(εj−ε1)
− κ
N∑
j=m+1, γj<0



τ(1, j)
γ1−1∑
`=1
Aγ+`(εj−ε1) +
−γj∑
`=1
Aγ−`(ε1−εj)τ(1, j)



. (5.2)
Each of the coefficients Aδ appearing on the second line of the equation satisfies δ ∈ ZN,s for
some s < n and for Aδ on the right-hand side of the first line δ = δpi − γν where δpi C γpi.
Proof. The formula follows from equation (5.1) by setting i = 1, β1 = 0 and omitting the case
αj > αi. Suppose γνj = 0 for j ≤ k, and γ
pi
j = 0 for j > k. The typical multi-index in the second
line is
(
γ1 − `, γ
pi
2 , . . . , γ
pi
k ,−γ
ν
k+1, . . . ,−γ
ν
j + `, . . .
)
with 1 ≤ ` ≤ γ1 − 1 or 1 ≤ ` ≤ γνj . If γ1 ≥ ` then the sum of the nonnegative components is
n−` < n; and if γ1 < ` (possible if γνj > γ1) then the sum of the nonnegative components is n−γ1.
In both cases the multi-index is in
⋃
s<n
ZN,s. From [6, Lemma 10.1.3] γpi  (γpi + `(εj − ε1))+
for 1 ≤ ` ≤ γ1 − γpij − 1, thus the multi-indices on the right-hand side of the first line satisfy
δ = δpi − γν with (δpi)+ ≺ γpi, that is, δpi C γpi. 
Orthogonality Measure on the Torus for Vector-Valued Jack Polynomials 19
If γ is B-minimal with fixed γν then there are no terms on the right side of the first line
(that is, γj ≥ 0 implies γ1 ≥ γj ≥ γ1 − 1).
Proposition 5.4. Among γ ∈ ZN,n such that γν = β for some fixed β with |β| = n and
such that βj > 0 exactly when j > k the minimal multi-index for the order γ(1)pi B γ(2)pi
is γ(0) =
(
p + 1, . . . ,
(m)
p+ 1, p, . . . ,
(k)
p ,−βk+1, . . . ,−βN
)
where p =
⌊
n
k
⌋
and m = n − kp (so
0 ≤ m < k). For this multi-index the right-hand side of (5.2) contains only Aδ with δ ∈
n−1⋃
s=0
ZN,s.
The proof is technical and is presented as Proposition A.5.
Theorem 5.5. The coefficients Aα are rational functions of κ and are finite provided
κ /∈
{
−
m
c
: m, c ∈ N, 1 ≤ c ≤ τ1 − 1
}
∪
{m
c
: m, c ∈ N, 1 ≤ c ≤ `(τ)− 1
}
.
Also A−α = A∗α and τ(w)
∗Awατ(w) = Aα for all α ∈ ZN , w ∈ SN and permitted values of κ.
Proof. The NSJP ζα,T is a rational function of κ with no poles in −1/hτ < κ < 1/hτ . The
coefficients Aα are defined in terms of all the NSJP’s and are also rational in κ. In equation (5.2)
the operator on the left of Aγ is
(
γ1I + κ
N∑
`=m+1
τ(1, `)
)
= τ(1,m)(γ1I + κτ(ωm))τ(1,m),
where ωm is the Jucys–Murphy element
N∑
`=m+1
(m, `); the action τ(ωm)T = c(m,T )T for all
T ∈ Y(τ) shows that the eigenvalues of the operator are {γ1 + κc(m,T ) : T ∈ Y(τ)} and
the operator is invertible provided κc(m,T ) /∈ {−1,−2,−3, . . .} for 1 ≤ m ≤ N . The set of
values of c(m,T ) is {j ∈ Z : 1 − `(τ) ≤ j ≤ τ1 − 1}. Thus an inductive argument based on n
in ZN,n, the order in Proposition 5.4, and formula (5.2) shows there are unique solutions for {Aα}
provided that the possible poles at n+ κc(i, T ) = 0 are excluded. The relations A−α = A∗α and
τ(w)∗Awατ(w) = Aα hold at least in an interval hence for all κ, excluding the poles. 
The largest interval around 0 without poles is − 1τ1−1 < κ <
1
`(τ)−1 . As illustration we
describe Aγ for γ ∈ ZN,2. Above we showed
Aε1−εj = −κ (I + κτ(ω1))
−1τ(1, j), 2 ≤ j ≤ N.
Next for α = ε1 + ε2 and β = 2εj for 3 ≤ j ≤ N we find
(
I + κ
N∑
i=3
τ(1, i)
)
Aε1+ε2−2εj = −κ(Aε2−ε1 +Aε2−εj )τ(1, j).
For α = ε1 + ε2 and β = εj + εj+1 with 3 ≤ j ≤ N − 1
(
I + κ
N∑
i=3
τ(1, i)
)
Aε1+ε2−εj−εj+1 = −κ
(
Aε2−εjτ(1, j + 1) +Aε2−εj+1τ(1, j)
)
.
For α = 2ε1 and β = 2εN we obtain
(2I + κτ(ω1))A2ε1−2εN
= −κ
{
N−1∑
`=2
τ(1, `)Aε1+ε`−2εN + τ(1, N)Aε1−εN + (Aε1−εN + I)τ(1, N)
}
.
The other coefficients for n = 2 are obtained using the relations
A−α = A
∗
α and τ(w)
∗Awατ(w) = Aα.
20 C.F. Dunkl
6 The differential equation
We will show that µ satisfies a differential system in a distributional sense. Let TNreg :=
TN\
⋃
1≤i<j≤N
{x : xi = xj} (this avoids the singularities of the system) and ∂j := ∂∂xj for
1 ≤ j ≤ N . The system is
xi∂iK(x) = κ
∑
j 6=i
xj
xi − xj
τ(i, j)K(x) + κK(x)
∑
j 6=i
τ(i, j)
xi
xi − xj
, 1 ≤ i ≤ N. (6.1)
Any solution of this system satisfies
N∑
i=1
xi∂iK (x) = 0 and thus is homogeneous of degree zero.
The relation 〈xiDif, g〉T = 〈f, xiDig〉T extends to C(1)
(
TN ;Vτ
)
, the continuously differen-
tiable Vτ -valued functions, because Laurent polynomials are dense in this space. We have shown
lim
n→∞
∫
TN
{
(xiDif(x))
∗Kn(x)g(x)− f(x)
∗Kn(x)xiDig(x)
}
dm(x) = 0.
Suppose p, q ∈ C(1)
(
TN
)
(scalar C-valued) then by periodicity
∫
TN
∂
∂θj
(pq)dm = 0 thus
∫
TN
(
∂
∂θj
p
)
qdm = −
∫
TN
p
(
∂
∂θj
q
)
dm.
Also from ∂∂θj f(x) = ie
iθj∂jf(x) = ixj∂jf(x) and ∂jf∗(x) = −x
−2
j (∂jf)
∗(x) we obtain
∫
TN
(xj∂jf(x))
∗g(x)dm = −
∫
TN
x−1j
(
−x2j
)
∂jf
∗(x)g(x)dm =
∫
TN
f∗(x)xj∂jg(x)dm.
That is, xj∂j is self-adjoint with respect to 〈f, g〉 :=
∫
TN f
∗(x)g(x)dm. The result extends to
C(1)
(
TN ;Vτ
)
.
Specialize to a closed SN -invariant subset E ⊂ TNreg which is the closure of its interior (for
example Eε =
{
x ∈ TN : mini<j |xi − xj | ≥ ε
}
for ε > 0), and let f, g ∈ C(1)
(
TN ;Vτ
)
have
supports contained in E, that is, f(x) = 0 = g(x) for x /∈ E. For fixed f , g, i let
In =
∫
TN
{
(xiDif(x))
∗Kn(x)g(x)− f(x)
∗Kn(x)xiDig(x)
}
dm(x)
=
∫
TN
{
(xi∂if(x))
∗Kn(x)g(x)− f(x)
∗Kn(x)xi∂ig(x)
}
+ κ
∫
TN
∑
j 6=i
(
τ(i, j)xi
f(x)− f(x(i, j))
xi − xj
)∗
Kn(x)g(x)dm(x)
− κ
∫
TN
f(x)∗Kn(x)
∑
j 6=i
(
τ(i, j)xi
g(x)− g(x(i, j))
xi − xj
)
dm(x).
By using
( xi
xi−xj
)∗ = − xjxi−xj , τ(i, j)
∗ = τ(i, j), and rearranging the sums we obtain
In =
∫
TN
{
(xi∂if(x))
∗Kn(x)g(x)− f(x)
∗Kn(x)xi∂ig(x)
}
dm(x)
− κ
∫
TN
f(x)∗
∑
j 6=i
{
xj
xi − xj
τ(i, j)Kn(x) +Kn(x)τ(i, j)
xi
xi − xj
}
g(x)dm(x)
+ κ
∑
j 6=i
∫
TN
{
xjf(x(i, j))
∗τ(i, j)Kn(x)g(x) + xif(x)
∗Kn(x)τ(i, j)g(x(i, j))
} dm(x)
xi − xj
.
Orthogonality Measure on the Torus for Vector-Valued Jack Polynomials 21
Each integral in the third line is finite because f and g vanish on a neighborhood of
⋃
1≤i<j≤N
{x :
xi = xj}. The terms inside {·} are invariant under the change of variable x 7→ x(i, j), because
τ(i, j)Kn(x(i, j)) = Kn(x)τ(i, j), but the denominator (xi−xj) changes sign; thus the integrand
is odd under (i, j) and the integral vanishes.
Because terms like τ(i, j) xixi−xj g(x) are in C
(1)
(
TN ;Vτ
)
(assumption on the support of g) we
can take the limit as n→∞ in the second line. By the adjoint property of xi∂i we find
∫
TN
(xi∂if(x))
∗Kn(x)g(x)dm(x) =
∫
TN
f(x)∗xi∂i{Kn(x)g(x)}dm(x)
=
∫
TN
f(x)∗
{
Kn(x)xi∂ig(x) + (xi∂iKn(x))g(x)
}
dm(x).
Thus the fact that lim
n→∞
In = 0 implies (recall the matrix-valued integral notation (4.1))
lim
n→∞
∫
TN
f(x)∗(xi∂iKn(x))g(x)dm(x)
= κ
∫
TN
f(x)∗
∑
j 6=i
1
xi − xj
{
xjτ(i, j)dµ(x) + dµ(x)τ(i, j)xi
}
g(x). (6.2)
This statement is valid for all f, g ∈ C(1)
(
TN ;Vτ
)
that vanish on a neighborhood of
⋃
1≤i<j≤N
{x :
xi = xj}. The first line of the equation can be written as
lim
n→∞
∫
TN
{
(xi∂if(x))
∗Kn(x)g(x)− f(x)
∗Kn(x)xi∂ig(x)
}
dm(x)
=
∫
TN
{
(xi∂if(x))
∗dµ(x)g(x)− f(x)∗dµ(x)xi∂ig(x)
}
,
which expresses the distributional derivative of dµ. Thus the distribution-sense differential
system is satisfied by dµ on closed subsets of TNreg.
In the scalar case (τ = (N)) the orthogonality weight is known to be (due to [1] for the
symmetric Jack polynomials)
K(x) =
∏
1≤i<j≤N
{
(xi − xj)
(
x−1i − x
−1
j
)}κ
and the differential equation system reduces to (note τ(w) = 1)
xi∂iK(x) = κK(x)
∑
j 6=i
xi + xj
xi − xj
, 1 ≤ i ≤ N.
The differential system (6.1) could be the subject for an article all by itself, but we can sketch
a result about the absolutely continuous part of µ, namely that dµ(x) = L(x)∗BL(x)dm(x),
x ∈ TNreg where B is a locally constant positive matrix and L(x) is a fundamental solution of
∂iL(x) = κL(x)



∑
j 6=i
1
xi − xj
τ(i, j)−
γ
xi
I



, 1 ≤ i ≤ N, (6.3)
γ :=
1
2N
`(τ)∑
i=1
τi(τi − 2i+ 1) =
1
N
N∑
j=1
c(j, T0).
22 C.F. Dunkl
The effect of the term γxi I is to make L(x) homogeneous of degree zero, that is,
N∑
i=1
xi∂iL(x) = 0,
because
∑
1≤i<j≤N
τ(i, j) =



N∑
j=1
c(j, T0)



I
(the sum of the contents in the diagram of τ).
The differential system is Frobenius integrable; this means that in the system ∂iL(x) =
κL(x)Mi(x), 1 ≤ i ≤ N , the two formal differentiations
∂j∂iL(x) = κ
2L(x)Mj(x)Mi(x) + κL(x)∂jMi(x),
∂i∂jL(x) = κ
2L(x)Mi(x)Mj(x) + κL(x)∂iMj(x),
are equal to each other for all i, j (see [2]). The system is analytic and thus any local solution
can be continued analytically to any point in CNreg := (C\{0})
N\
⋃
1≤i<j≤N
{x : xi = xj}. When
restricted to TNreg there are solutions defined on each connected component. This is possible
because L(x) is constant on {ux : |u| = 1} for fixed x ∈ TNreg and each component is homotopic
to a circle. Denote the component containing all the points
{(
eiθ1 , . . . , eiθN
)
: − pi < θ1 < θ2 <
· · · < θN < pi
}
by C0. Since x and ux are in the same component for |u| = 1 we see that x0 ∈ C0
(recall x0 =
(
1, ω, . . . , ωN−1
)
, ω = e2pii/N ). The components are {C0w : w ∈ SN , w(1) = 1}
(corresponding to the (N − 1)! circular permutations of (1, 2, . . . , N)). Now fix the unique
solution L(x) such that L(x0w) = I for each w with w(1) = 1. By differentiating L(x)−1L(x) = I
obtain the system satisfied by L−1:
∂iL(x)
−1 = −κ
{
∑
j 6=i
1
xi − xj
τ(i, j)−
γ
xi
I
}
L(x)−1, 1 ≤ i ≤ N.
The goal here is to replace f , g in formula (6.2) by L−1f , L−1g and deduce the desired result.
By use of xi∂i(L−1∗) = −
(
xi∂iL−1
)∗
we obtain
(
xi∂i
(
L(x)−1∗
))
dµ(x)L(x)−1 + L(x)−1∗dµ(x)
(
xi∂iL(x)
−1)
= κL−1∗
{
∑
j 6=i
−xj
xi − xj
τ(i, j)− γI
}
dµL−1 − κL−1∗dµ
{
∑
j 6=i
xi
xi − xj
τ(i, j)− γI
}
L−1
= −κL−1∗
∑
j 6=i
{
xj
xi − xj
τ(i, j)dµ+ dµτ(i, j)
xi
xi − xj
}
L−1.
Substitute this relation in formula (6.2)
lim
n→∞
∫
TN
f(x)∗L(x)−1∗(xi∂iKn(x))L(x)
−1g(x)dm(x)
+
∫
TN
f(x)∗(xi∂i)
(
L(x)−1∗
)
dµ(x)L(x)−1g(x)
+
∫
TN
f(x)∗L(x)−1∗dµ(x)xi∂iL(x)
−1g(x) = 0.
The formula is valid because L−1f, L−1g ∈ C(1)
(
TNreg;Vτ
)
and vanish for x /∈ E. The first line
of the equation is the distributional derivative of µ so the equation is equivalent to
∫
TN
f(x)∗
[
xi∂i
(
L(x)−1∗dµ(x)L(x)−1
)]
g(x) = 0, 1 ≤ i ≤ N.
Orthogonality Measure on the Torus for Vector-Valued Jack Polynomials 23
Thus all the partial derivatives of the distribution L−1∗dµL−1 vanish and L−1∗dµL−1 = Bdm,
where B is constant on each component of TNreg. We conclude dµ(x) = L(x)
∗BL(x)dm(x) for
x ∈ TNreg. Part (1) of Theorem 4.17 implies B is a positive matrix.
We have to point out that this result provides no information about the behavior of µ on the
singular set
⋃
i<j
{x : xi = xj}. We conjecture that µ does not have a singular part. This question
seems a worthy topic for further investigations. Some of the difficulties in this problem come
from the behavior of the solutions of system (6.3) in neighborhoods of the sets {x : xi = xj};
there are singularities of order |xi − xj |±κ. Both signs appear because the eigenvalues of τ(i, j)
are 1 and −1; a consequence of the assumption that τ is not one-dimensional.
A Appendix
A.1 The matrix Bochner theorem
For a matrix A ∈Mn(C) the operator norm is
‖A‖ := sup{|Av| : |v| = 1} = sup{|〈u,Av〉| : |u| = 1 = |v|}.
Proposition A.1. Suppose F is a positive-definite Mn(C)-valued function on ZN then
(1) F (0) is positive-definite,
(2) F (−α) = F (α)∗ and ‖F (α)‖ ≤ ‖F (0)‖ for all α ∈ ZN .
Proof. Part (1) follows immediately from taking f(0) = u, f(α) = 0 for α 6= 0. For part (2)
fix α 6= 0 and let f(0) = u, f(α) = v and f(β) = 0 otherwise. By definition
u∗F (0)u+ v∗F (0)v + v∗F (α)u+ u∗F (−α)v ≥ 0.
Thus Im(v∗F (α)u + u∗F (−α)v) = 0 for all u, v. For 1 ≤ j, k ≤ n let u = εj , v = cεk, c ∈ C,
then
0 = Im
(
cF (α)kj + cF (−α)jk
)
.
Set c = 1 and c = i to show F (−α)jk = F (α)kj , that is, F (−α) = F (α)∗. Thus u∗F (−α)v =
v∗F (α)u and
−2 Re(v∗F (α)u) ≤ u∗F (0)u+ v∗F (0)v.
Specialize to vectors u, v such that |u| = 1 = |v| and v∗F (α)u = 〈v, F (α)u〉 = −‖F (α)‖.
Since F (0) is positive-definite it follows that u∗F (0)u+ v∗F (0)v ≤ 2‖F (0)‖, and so ‖F (α)‖ ≤
‖F (0)‖. 
Proposition A.2. Suppose α, β ∈ ZN and α, β 6= 0 then
‖F (α)− F (β)‖2 ≤ 2‖F (0)‖{v∗F (0)v − Re(v∗F (α− β)v)},
where |v| = 1 and |(F (α)− F (β))∗v| = ‖F (α)− F (β)‖.
Proof. Assume α 6= β and let f(0) = u, f(α) = w, f(β) = −w, and f(γ) = 0 otherwise. By
definition
u∗F (0)u+ 2w∗F (0)w + w∗(F (α)− F (β))u+ u∗(F (−α)− F (−β))w
− w∗F (α− β)w − w∗F (β − α)w ≥ 0.
24 C.F. Dunkl
Let u, v ∈ Cn satisfy |u| = 1 = |v| and |(F (α)−F (β))∗v| = v∗(F (α)−F (β))u = ‖F (α)−F (β)‖.
Set w = tv with t ∈ R. The inequality becomes
u∗F (0)u+ 2t2v∗F (0t)v + 2t‖F (α)− F (β)‖ − 2t2 Re
(
v∗F (α− β)v
)
≥ 0.
The discriminant of the quadratic polynomial in t must be nonpositive and this implies the
stated inequality, since u∗F (0)u ≤ ‖F (0)‖. 
This is a sort of uniform continuity.
For any fixed u ∈ Cn the scalar function gu(α) = u∗F (α)u is positive-definite in the classical
sense and thus by Bochner’s theorem (see [13, pp. 17–21]) there exists a unique positive Baire
measure µu on TN such that
∫
TN
x−αdµu(x) = u
∗F (α)u, α ∈ ZN .
In particular, for 1 ≤ j ≤ n let u = εj then u∗F (α)u = F (α)jj and set µjj = µεj . For
1 ≤ j, k ≤ n (with j 6= k) let u = εj + εk, v = εj + iεk
∫
TN
x−αdµu(x) = F (α)jj + F (α)jk + F (α)kj + F (α)kk,
∫
TN
x−αdµv(x) = F (α)jj + iF (α)jk − iF (α)kj + F (α)kk.
Thus
∫
TN
x−αd(µu − iµv) = 2F (α)jk + (1− i)
∫
TN
x−αd(µjj + µkk),
and define
µjk =
1
2
(µu − iµv)−
1− i
2
(µjj + µkk),
with the result
∫
TN
x−αdµjk(x) = F (α)jk, α ∈ ZN .
From the above equations we also obtain
∫
TN
x−αd(µu + iµv) = 2F (α)kj + (1 + i)
∫
TN
x−αd(µjj + µkk).
The formula
µkj =
1
2
(µu + iµv)−
1 + i
2
(µjj + µkk),
is consistent with the previous one and it can be directly verified that µ̂kj(−α) = µ̂jk(α) from
the general equation µ̂u(−α) = µ̂u(α). This is a restatement of F (−α) = F (α)∗.
From ‖µu‖ = u∗F (0)u (measure/total variation norm) we obtain ‖µjj‖ = F (0)jj and
‖µjk‖ ≤
1
√
2
(F (0)jj + F (0)kk) + F (0)jj + F (0)kk + |ReF (0)jk|+ | ImF (0)jk|.
Orthogonality Measure on the Torus for Vector-Valued Jack Polynomials 25
In particular, if F (0) = I then ‖µjk‖ ≤ 2 +
√
2. For the space C
(
TN ;Cn
)
(continuous functions
on TN , values in Cn) define an inner product
〈f, g〉F :=
n∑
i,j=1
∫
TN
f(x)ig(x)jdµij(x),
then for α, β ∈ ZN , and 1 ≤ i, j ≤ n (with the standard unit basis vectors εi of Cn)
〈
xαεi, x
βεj
〉
F =
∫
TN
x−αxβdµij(x) = F (α− β)ij .
The following summarizes the above results.
Theorem A.3. The Hermitian form 〈·, ·〉F is bounded on C
(
TN ;Cn
)
, that is |〈f, g〉F | ≤
B‖f‖∞‖g‖∞ for some B < ∞, is positive-semidefinite, and if f , g are finitely supported func-
tions on ZN with values in Cn then for f̂(x) :=
∑
α f(α)x
α and ĝ(x) :=
∑
α g(α)x
α
〈
f̂ , ĝ
〉
F =
∑
α,β
f(α)∗F (α− β)g(β).
Proof. The bound follows from the uniform bound on ‖µjk‖ for all j, k, depending only on F (0).
Suppose f̂ , ĝ are trigonometric (Laurent) polynomials (equivalent to finite support on ZN ), then
〈
f̂ , ĝ
〉
F =
∑
α,β
n∑
i,j=1
∫
TN
f(α)ix
−αg(β)jx
βdµij(x)
=
∑
α,β
n∑
i,j=1
f(α)iF (α− β)ijg(β)j =
∑
α,β
f(α)∗F (α− β)g(β).
By definition
〈
f̂ , f̂
〉
F ≥ 0 and by the density of the trigonometric polynomials in C
(
TN ;Cn
)
it follows that 〈·, ·〉F is a positive semidefinite (it is possible that
〈
f̂ , f̂
〉
F = 0 for some f 6= 0)
Hermitian form. 
The next result is used for the approximate identity arguments.
Theorem A.4. Suppose σ ∈ C
(
TN
)
and σ(x) ≥ 0 for all x ∈ TN then each σ ∗ µij ∈ C
(
TN
)
and [σ ∗ µij(x)]ni,j=1 ∈Mn(C) is positive semidefinite for each x ∈ T
N .
Proof. Suppose (fi)ni=1 ∈ C
(
TN ;Cn
)
. A convolution formula (similar to that in Lemma 4.13)
shows that
n∑
i,j=1
∫
TN
f(x)iσ ∗ µij(x)f(x)jdm(x) =
n∑
i,j=1
∫
TN
∫
TN
σ
(
xy−1
)
f(x)if(x)jdµij(y)dm(x)
=
∫
TN
σ(u)
∫
TN
n∑
i,j=1
f(yu)if(yu)jdµij(y)dm(u) ≥ 0,
(change-of-variable x = yu) because y 7→ f(yu) ∈ C
(
TN ;Cn
)
and the double sum is continuous
in u and nonnegative by the above theorem. Now let f(x) = g(x)v where v ∈ Cn and g ∈ C
(
TN
)
,
g ≥ 0,
∫
TN g
2dm = 1 and g = 0 off an ε-neighborhood of a fixed z ∈ TN (i.e., |x− z| ≥ ε implies
g(x) = 0) then
n∑
i,j=1
∫
TN
f(x)iσ ∗ µij(x)f(x)jdm(x) =
∫
TN
g(x)2
n∑
i,j=1
vi(σ ∗ µij(x))vjdm(x) ≥ 0.
Let ε→ 0 then the integral tends to
n∑
i,j=1
vi(σ ∗ µij(z))vj , and this completes the proof. 
26 C.F. Dunkl
A.2 Some results for the index set
For n = 1, 2, 3, . . .
#ZN,n =
N−1∑
j=1
(
N
j
)(
n− 1
j − 1
)(
N − j + n− 1
n
)
; (A.1)
in each j-subset (there are
(N
j
)
) of [1, 2, . . . , N ] take j-tuples αi1 , . . . , αij with
j∑
`=1
αi` = n and
each αi` ≥ 1 (
(n−1
j−1
)
possibilities) and in the complement take (N − j)-tuples with
N−j∑
`=1
αi` = −n
and each αi` ≤ 0 (
(N−j+n−1
n
)
possibilities). For example when n ≥ 1
#Z2,n = 2, #Z3,n = 6n, #Z4,n = 10n2 + 2, #Z5,n =
5n
3
(
7n2 + 5
)
, . . . .
Proposition A.5. Among γ ∈ ZN,n such that γν = β for some fixed β with |β| = n and
such that βj > 0 exactly when j > k the minimal multi-index for the order γ(1)pi B γ(2)pi
is γ(0) =
(
p + 1, . . . ,
(m)
p+ 1, p, . . . ,
(k)
p ,−βk+1, . . . ,−βN
)
where p =
⌊
n
k
⌋
and m = n − kp (so
0 ≤ m < k).
Proof. The claim is that
(
γ(0)i
)k
i=1 is ≺-minimal among partitions α of length ≤ k and |α| = n.
Argue by induction on the length. The statement is obviously true when k = 1. Suppose it is
true for k and let α be a partition of length ≤ k + 1. Let
n1 =
k∑
i=1
αi, n2 = n1 + αk+1, p1 =
⌊n1
k
⌋
, p2 =
⌊
n2
k + 1
⌋
,
m1 = n1 − kp1, m2 = n2 − (k + 1)p2.
Define γ(1) and γ(2) analogously to the above (γ(s)i = ps + 1 for 1 ≤ i ≤ ms and γ
(1)
i = p1 for
m1 < i ≤ k, γ(2) = p2 for m2 < i ≤ k + 1). By the inductive hypothesis
i∑
j=1
αj ≥
i∑
j=1
γ(1)j . This
implies αk+1 ≤ αk ≤ γ
(1)
k . Thus (k+1)p2 ≤ n1+αk+1 ≤ m1+kp1+αk+1 ≤ m1+(k+1)p1 and p2 ≤
m1
k+1 + p1. Since p1, p2 are integers this implies p2 ≤ p1. If p1 = p2 then m2 = m1 − (p1 − αk+1),
and clearly
i∑
j=1
γ(1)j ≥
i∑
j=1
γ(2)j for 1 ≤ i ≤ k. If p2 < p1 then γ
(2)
j ≤ p2 + 1 ≤ p1 ≤ γ
(1)
j for
1 ≤ j ≤ k. Thus α  γ(2). 
References
[1] Beerends R.J., Opdam E.M., Certain hypergeometric series related to the root system BC, Trans. Amer.
Math. Soc. 339 (1993), 581–609.
[2] Dunkl C.F., Differential-difference operators and monodromy representations of Hecke algebras, Pacific J.
Math. 159 (1993), 271–298.
[3] Dunkl C.F., Symmetric and antisymmetric vector-valued Jack polynomials, Se´m. Lothar. Combin. 64 (2010),
Art. B64a, 31 pages, arXiv:1001.4485.
[4] Dunkl C.F., Vector polynomials and a matrix weight associated to dihedral groups, SIGMA 10 (2014), 044,
23 pages, arXiv:1306.6599.
Orthogonality Measure on the Torus for Vector-Valued Jack Polynomials 27
[5] Dunkl C.F., Luque J.G., Vector-valued Jack polynomials from scratch, SIGMA 7 (2011), 026, 48 pages,
arXiv:1009.2366.
[6] Dunkl C.F., Xu Y., Orthogonal polynomials of several variables, 2nd ed., Encyclopedia of Mathematics and
its Applications, Cambridge University Press, Cambridge, 2014.
[7] Etingof P., Stoica E., Unitary representations of rational Cherednik algebras, Represent. Theory 13 (2009),
349–370, arXiv:0901.4595.
[8] Griffeth S., Orthogonal functions generalizing Jack polynomials, Trans. Amer. Math. Soc. 362 (2010),
6131–6157, arXiv:0707.0251.
[9] James G., Kerber A., The representation theory of the symmetric group, Encyclopedia of Mathematics and
its Applications, Vol. 16, Addison-Wesley Publishing Co., Reading, Mass., 1981.
[10] Lapointe L., Vinet L., Exact operator solution of the Calogero–Sutherland model, Comm. Math. Phys. 178
(1996), 425–452.
[11] Murphy G.E., A new construction of Young’s seminormal representation of the symmetric groups, J. Algebra
69 (1981), 287–297.
[12] Opdam E.M., Harmonic analysis for certain representations of graded Hecke algebras, Acta Math. 175
(1995), 75–121.
[13] Rudin W., Fourier analysis on groups, Interscience Tracts in Pure and Applied Mathematics, Vol. 12, Inter-
science Publishers, New York – London, 1962.
[14] Vershik A.M., Okunkov A.Yu., A new approach to representation theory of symmetric groups. II, J. Math.
Sci. 131 (2005), 5471–5494, math.RT/0503040.