arXiv is now an independent nonprofit! Learn more
License: arXiv.org perpetual non-exclusive license
arXiv:2609.13837v1 [math.QA] 12 Sep 2026

An integral representation of eigenfunctions for the deformed Noumi–Sano operators

Taikei Fujii    Takahiko Nobukawa
Abstract

The deformed Noumi–Sano operator Hn,md​(x,y)H_{n,m}^{d}(x,y) of order dd is a simultaneous extension of the Macdonald operator and the Noumi–Sano operator. In this paper, we construct an integral transform associated with eigenvalue problem of Hn,md​(x,y)H_{n,m}^{d}(x,y). Using the integral transform, we obtain an integral representation of eigenfunctions for Hn,md​(x,y)H_{n,m}^{d}(x,y).

Key words: qq-hypergeometric integral, Macdonald polynomial, qq-difference equation.

2020 Mathematics Subject Classification Numbers: 39A13, 33D52, 44A20.

1 Introduction

We fix qq, t∈ℂt\in\mathbb{C} with 0<|q|,|t|<10<|q|,\,|t|<1. The shift operator Tq,x±μ=∏i=1nTq,xi±μiT_{q,x}^{\pm\mu}=\prod_{i=1}^{n}T_{q,x_{i}}^{\pm\mu_{i}} is defined by (Tq,x±μ​f)​(x1,…,xn)=f⁡(q±μ1​x1,…,q±μn​xn)(T_{q,x}^{\pm\mu}f)(x_{1},\ldots,x_{n})=f(q^{\pm\mu_{1}}x_{1},\ldots,q^{\pm\mu_{n}}x_{n}). Hallnäs, Langmann, Noumi and Rosengren [3] introduced a family of qq-difference operators defined by

Hn,m​(x,y,u,q,t)=Hn,m​(x,y,u)=∑d=0∞ud​Hn,md​(x,y)\displaystyle H_{n,m}(x,y;u;q,t)=H_{n,m}(x,y;u)=\sum_{d=0}^{\infty}u^{d}H_{n,m}^{d}(x,y)
:=∑μ∈(ℤ≥0)n∑I⊂{1,…,m}(t1−n​qm​u)|μ|​(−t​u)|I|​q(|I|2)​Bμ,I​(x,y)​Tq,xμ​Tt,y−I,\displaystyle:=\sum_{\mu\in(\mathbb{Z}_{\geq 0})^{n}}\sum_{I\subset\{1,\ldots,m\}}(t^{1-n}q^{m}u)^{|\mu|}(-tu)^{|I|}q^{\binom{|I|}{2}}B_{\mu,I}(x,y)T_{q,x}^{\mu}T_{t,y}^{-I}, (1.1)
Bμ,I​(x,y)=Δ⁡(qμ​x)Δ⁡(x)​∏i,j=1n(t​xi/xj,q)μi(q​xi/xj,q)μi​∏1≤i,j≤mi∈I,j∉Iyi−q​yjyi−yj\displaystyle B_{\mu,I}(x,y)=\frac{\Delta(q^{\mu}x)}{\Delta(x)}\prod_{i,j=1}^{n}\frac{(tx_{i}/x_{j};q)_{\mu_{i}}}{(qx_{i}/x_{j};q)_{\mu_{i}}}\prod_{\begin{subarray}{c}1\leq i,j\leq m\\ i\in I,j\notin I\end{subarray}}\frac{y_{i}-qy_{j}}{y_{i}-y_{j}}
×∏i=1n(∏j∈I1−xi/(t​yj)1−qμi​xi/yj​∏j∉I1−xi/(q​yj)1−qμi−1​xi/yj),\displaystyle\quad\quad\quad\quad\quad\quad\times\prod_{i=1}^{n}\left(\prod_{j\in I}\frac{1-x_{i}/(ty_{j})}{1-q^{\mu_{i}}x_{i}/y_{j}}\prod_{j\notin I}\frac{1-x_{i}/(qy_{j})}{1-q^{\mu_{i}-1}x_{i}/y_{j}}\right), (1.2)

where |μ|=μ1+⋯+μn|\mu|=\mu_{1}+\cdots+\mu_{n}, qμ​x=(qμ1​x1,…,qμn​xn)q^{\mu}x=(q^{\mu_{1}}x_{1},\ldots,q^{\mu_{n}}x_{n}), Δ⁡(x)=∏i<j(xi−xj)\Delta(x)=\prod_{i<j}(x_{i}-x_{j}) and (a,q)n=∏k=0n−1(1−a​qk)(a;q)_{n}=\prod_{k=0}^{n-1}(1-aq^{k}). The operator Hn,m​(x,y,u)H_{n,m}(x,y;u) is called the deformed Noumi–Sano operator. This operator Hn,m​(x,y,u)H_{n,m}(x,y;u) is a simultaneous extension of the Macdonald operator of type AA [8, 10] and the Noumi–Sano operator [11]. More precisely, H0,m​(y−1,u/t,t,q)H_{0,m}(y^{-1},u/t;t,q) coincides with the Macdonald operator, and Hn,0​(x,u/t,q,t)H_{n,0}(x;u/t;q,t) is the Noumi–Sano operator (see [2, 3, 13] for detailed properties). The operator (1.1) has super-Macdonald polynomials [13] as eigenfunctions. Our aim is to construct an integral transform associated with the eigenvalue problem for the deformed Noumi–Sano operator Hn,md​(x,y)H_{n,m}^{d}(x,y).

We define the function ωn,m;N,M​(x,y,z,w)\omega_{n,m;N,M}(x,y;z,w) of the variables x=(x1,…,xn)x=(x_{1},\ldots,x_{n}), y=(y1,…,ym)y=(y_{1},\ldots,y_{m}), z=(z1,…,zN)z=(z_{1},\ldots,z_{N}) and w=(w1,…,wM)w=(w_{1},\ldots,w_{M}) as

ωn,m;N,M​(x,y,z,w)=xα​yβ​Ψn,m;N,M​(x,y,z,w)​ρ​(x,y),\displaystyle\omega_{n,m;N,M}(x,y;z,w)=x^{\alpha}y^{\beta}\Psi_{n,m;N,M}(x,y;z,w)\rho(x,y), (1.3)

for α,β∈ℂ\alpha,\beta\in\mathbb{C}, where (a,p)∞=∏k=0∞(1−a​pk)(a;p)_{\infty}=\prod_{k=0}^{\infty}(1-ap^{k}) and

ρ⁡(x,y)=(∏1≤i≠j≤n(xi/xj,q)∞/(t​xi/xj,q)∞)​(∏1≤i≠j≤m(yi/yj,t)∞/(q​yi/yj,t)∞)∏i=1n∏j=1m(1−q−1​xi/yj)​(1−t​yj/xi),\displaystyle\rho(x,y)=\frac{\left(\prod_{1\leq i\neq j\leq n}(x_{i}/x_{j};q)_{\infty}/(tx_{i}/x_{j};q)_{\infty}\right)\left(\prod_{1\leq i\neq j\leq m}(y_{i}/y_{j};t)_{\infty}/(qy_{i}/y_{j};t)_{\infty}\right)}{\prod_{i=1}^{n}\prod_{j=1}^{m}(1-q^{-1}x_{i}/y_{j})(1-ty_{j}/x_{i})}, (1.4)
Ψn,m;N,M​(x,y,z,w)\displaystyle\Psi_{n,m;N,M}(x,y;z,w)
=∏i=1n∏j=1N(xi​zj,q)∞(xi​zj/t,q)∞​∏i=1m∏j=1M(q​t​yi​wj,t)∞(t​yi​wj,t)∞​∏i=1n∏j=1M(1−xi​wj)​∏i=1m∏j=1N(1−yi​zj).\displaystyle=\prod_{i=1}^{n}\prod_{j=1}^{N}\frac{(x_{i}z_{j};q)_{\infty}}{(x_{i}z_{j}/t;q)_{\infty}}\prod_{i=1}^{m}\prod_{j=1}^{M}\frac{(qty_{i}w_{j};t)_{\infty}}{(ty_{i}w_{j};t)_{\infty}}\prod_{i=1}^{n}\prod_{j=1}^{M}(1-x_{i}w_{j})\prod_{i=1}^{m}\prod_{j=1}^{N}(1-y_{i}z_{j}). (1.5)

The main result of this paper is as follows.

Theorem 1.1.

Assume qα=t−βq^{\alpha}=t^{-\beta}, r<|q|d​Rr<|q|^{d}R, and |q|<|t||q|<|t|, where α,β∈ℤ\alpha,\beta\in\mathbb{Z} and r,R∈ℝ>0r,R\in\mathbb{R}_{>0}. Suppose that a function φ⁡(x,y)\varphi(x,y) satisfies the following conditions:

  • 1.

    The function φ⁡(x,y)\varphi(x,y) is a simultaneous eigenfunction of Hn,mkH^{k}_{n,m} (0≤k≤d)(0\leq k\leq d), i.e.

    Hn,mk​(t​x−1,q−1​y−1)​φ​(x,y)=Ek​φ​(x,y),0≤k≤d.\displaystyle H^{k}_{n,m}(tx^{-1},q^{-1}y^{-1})\varphi(x,y)=E_{k}\varphi(x,y),\quad 0\leq k\leq d. (1.6)
  • 2.

    The function φ⁡(x,y)\varphi(x,y) is holomorphic on D={(x,y)∣|q|dR≤|xi|≤R,r≤|yj|≤|t|−1r(1≤i≤n, 1≤j≤m)}D=\{(x,y)\mid|q|^{d}R\leq|x_{i}|\leq R,\,r\leq|y_{j}|\leq|t|^{-1}r\,(1\leq i\leq n,\,1\leq j\leq m)\}.

We put Cd=𝕋Rn×𝕋rmC_{d}=\mathbb{T}^{n}_{R}\times\mathbb{T}^{m}_{r}, where 𝕋ps\mathbb{T}^{s}_{p} denotes the ss-dimensional torus of radius pp. If |zj|<|t|/R,|wj|<1/r|z_{j}|<|t|/R,\,|w_{j}|<1/r, then the function

Ω⁡(z,w)=∫Cdφ⁡(x,y)​ωn,m;N,M​(x,y,z,w)​d​xx​d​yy\displaystyle\Omega(z,w)=\int_{C_{d}}\varphi(x,y)\omega_{n,m;N,M}(x,y;z,w)\frac{dx}{x}\frac{dy}{y} (1.7)

satisfies

HN,Md​(z,w)​Ω​(z,w)=(∑k+l=d(tN−n​qm−M,q)k(q,q)k​(t1−N​qM)k​tβ​l​El)​Ω​(z,w).\displaystyle H^{d}_{N,M}(z,w)\Omega(z,w)=\left(\sum_{k+l=d}\frac{(t^{N-n}q^{m-M};q)_{k}}{(q;q)_{k}}(t^{1-N}q^{M})^{k}t^{\beta l}E_{l}\right)\Omega(z,w). (1.8)
Remark 1.2.

The function ρ⁡(x,y)\rho(x,y) coincides with the function Δn,m​(x,y,q,t)\Delta_{n,m}(x,y;q,t) in [1] with xx and yy replaced by t1/2​xt^{1/2}x and q−1/2yq^{-1/2}y, respectively. The function Ψn,m;N,M​(x,y,z,w)\Psi_{n,m;N,M}(x,y;z,w) is a kernel function of the deformed Noumi–Sano operator [3, Theorem 2.1, Remark 2.2] as

HN,M​(z,w,u)​Ψn,m;N,M​(x,y,z,w)\displaystyle H_{N,M}(z,w;u)\Psi_{n,m;N,M}(x,y;z,w)
=(t1−n​qm​u,q)∞(t1−N​qM​u,q)∞​Hn,m​(x,y,u)​Ψn,m;N,M​(x,y,z,w).\displaystyle=\frac{(t^{1-n}q^{m}u;q)_{\infty}}{(t^{1-N}q^{M}u;q)_{\infty}}H_{n,m}(x,y;u)\Psi_{n,m;N,M}(x,y;z,w). (1.9)
Remark 1.3.

When n=N=0n=N=0, our integral Ω⁡(z,w)\Omega(z,w) coincides with the integral in [9].

Due to the explicit formula [3, equation (36)]

Hn,ml​(x,y)​1=(tn​q−m,q)l(q,q)l​(t1−n​qm)l​1,\displaystyle H^{l}_{n,m}(x,y)1=\frac{(t^{n}q^{-m};q)_{l}}{(q;q)_{l}}(t^{1-n}q^{m})^{l}1, (1.10)

we obtain an eigenfunction of HN,Md​(z,w)H^{d}_{N,M}(z,w) as follows:

Corollary 1.4.

Assume qα=t−βq^{\alpha}=t^{-\beta}, r<|q|d​Rr<|q|^{d}R, and |q|<|t||q|<|t|, where α,β∈ℤ\alpha,\beta\in\mathbb{Z} and r,R∈ℝ>0r,R\in\mathbb{R}_{>0}. If |zj|<|t|/R,|wj|<1/r|z_{j}|<|t|/R,\,|w_{j}|<1/r, then the following equation holds:

HN,Md​(z,w)​∫𝕋Rn×𝕋rmωn,m;N,M​(x,y,z,w)​d​xx​d​yy\displaystyle H^{d}_{N,M}(z,w)\int_{\mathbb{T}^{n}_{R}\times\mathbb{T}^{m}_{r}}\omega_{n,m;N,M}(x,y;z,w)\frac{dx}{x}\frac{dy}{y}
=(∑k+l=d(tN−n​qm−M,q)k​(tn​q−m,q)l(q,q)k​(q,q)l​(t1−N​qM)k​(qm​tβ+1−n)l)\displaystyle=\left(\sum_{k+l=d}\frac{(t^{N-n}q^{m-M};q)_{k}(t^{n}q^{-m};q)_{l}}{(q;q)_{k}(q;q)_{l}}(t^{1-N}q^{M})^{k}(q^{m}t^{\beta+1-n})^{l}\right)
×∫𝕋Rn×𝕋rmωn,m;N,M​(x,y,z,w)​d​xx​d​yy.\displaystyle\quad\quad\quad\quad\times\int_{\mathbb{T}^{n}_{R}\times\mathbb{T}^{m}_{r}}\omega_{n,m;N,M}(x,y;z,w)\frac{dx}{x}\frac{dy}{y}. (1.11)

The contents of this paper are as follows. In Section 2, we review [9]. In Section 3, we give a proof of Theorem 1.1. In Section 4, we summarize the results and discuss related problems. In Appendix A, we construct an integral transform for the eigenvalue problem of the Noumi–Sano operator Hn,0​(x,u)H_{n,0}(x;u).

2 Review of Mimachi–Noumi’s method

In this paper, we construct an integral transform for deformed Noumi–Sano operators using the method of Mimachi–Noumi [9]. In this section, we give a review of this method.

Macdonald introduced a commuting family of qq-difference operators defined by

Dyr=t(r2)​∑I⊂{1,2,…,m}|I|=r∏i∈I,j∉It​yi−yjyi−yj​Tq,yI,\displaystyle D_{y}^{r}=t^{\binom{r}{2}}\sum_{\begin{subarray}{c}I\subset\{1,2,\ldots,m\}\\ |I|=r\end{subarray}}\prod_{i\in I,j\notin I}\frac{ty_{i}-y_{j}}{y_{i}-y_{j}}T_{q,y}^{I}, (2.1)

for y=(y1,…,ym)y=(y_{1},\ldots,y_{m}) and r=1,2,…,mr=1,2,\ldots,m. The generating function of these operators is given by

Dy​(u)=∑r=0m(−u)r​Dyr\displaystyle D_{y}(u)=\sum_{r=0}^{m}(-u)^{r}D_{y}^{r} (2.2)

The operator Dy​(u)D_{y}(u) satisfies the following important kernel identity.

Theorem 2.1 (Macdonald [8], Mimachi–Noumi [9], Noumi [10]).

We set

Πm,M​(y,w)=∏i=1m∏j=1M(t​yi​wj,q)∞(yi​wj,q)∞.\displaystyle\Pi_{m,M}(y;w)=\prod_{i=1}^{m}\prod_{j=1}^{M}\frac{(ty_{i}w_{j};q)_{\infty}}{(y_{i}w_{j};q)_{\infty}}. (2.3)

Then we have

Dy​(u)​Πm,M​(y,w)=(u,t)m−M​Dw​(tm−M​u)​Πm,M​(y,w).\displaystyle D_{y}(u)\Pi_{m,M}(y;w)=(u;t)_{m-M}D_{w}(t^{m-M}u)\Pi_{m,M}(y;w). (2.4)
Proof.

The identity (2.4) is independent of qq. Hence it suffices to prove the case q=tq=t. When q=tq=t, we obtain the following from Cauchy formula.

Πm,M​(y,w)=∏i=1m∏j=1M11−yi​wj=∑l⁡(λ)≤min⁡{m,M}sλ​(y)​sλ​(w).\displaystyle\Pi_{m,M}(y;w)=\prod_{i=1}^{m}\prod_{j=1}^{M}\frac{1}{1-y_{i}w_{j}}=\sum_{l(\lambda)\leq\min\{m,M\}}s_{\lambda}(y)s_{\lambda}(w). (2.5)

Suppose M≥mM\geq m, we have

Dy​(u)​sλ​(y)=sλ​(y)​∏i=1m(1−u​tM−i+λi)​∏i=m+1M(1−u​tM−i),\displaystyle D_{y}(u)s_{\lambda}(y)=s_{\lambda}(y)\prod_{i=1}^{m}(1-ut^{M-i+\lambda_{i}})\prod_{i=m+1}^{M}(1-ut^{M-i}),
Dw​(u)​sλ​(w)=sλ​(w)​∏i=1m(1−u​tM−i+λi).\displaystyle D_{w}(u)s_{\lambda}(w)=s_{\lambda}(w)\prod_{i=1}^{m}(1-ut^{M-i+\lambda_{i}}). (2.6)

By (2.5) and (2.6), we obtain the special case q=tq=t of the desired formula. This completes the proof. ∎

The proof of our main results is based on a method similar to that of Mimachi–Noumi [9]. Therefore, we review Mimachi–Noumi’s result and their proof.

Theorem 2.2 (Mimachi–Noumi [9]).

Assume α∈ℤ\alpha\in\mathbb{Z}. Suppose that a function φ⁡(y)\varphi(y) satisfies the following conditions:

  • 1.

    The function φ⁡(y)\varphi(y) is an eigenfunction of Dy−1​(u/qα)D_{y^{-1}}(u/q^{\alpha}), i.e.

    Dy−1​(u/qα)​φ​(y)=E​φ​(y).\displaystyle D_{y^{-1}}(u/q^{\alpha})\varphi(y)=E\varphi(y). (2.7)
  • 2.

    The function φ⁡(y)\varphi(y) is holomorphic on V={y∣|q|≤|yi|≤1​(1≤i≤m)}V=\{y\mid|q|\leq|y_{i}|\leq 1\ (1\leq i\leq m)\}.

If |wj|<1|w_{j}|<1, then the function

Ω⁡(w)=∫𝕋mφ⁡(y)​ωm,M​(y,w)​d​yy\displaystyle\Omega(w)=\int_{\mathbb{T}^{m}}\varphi(y)\omega_{m,M}(y;w)\frac{dy}{y} (2.8)

satisfies

Dw​(u)​Ω​(w)=(u,t)M−m​E​Ω​(w),\displaystyle D_{w}(u)\Omega(w)=(u;t)_{M-m}E\Omega(w), (2.9)

where

ωm,M​(y,w)=yα​Πm,M​(y,w)​ρ​(y),ρ⁡(y)=∏i≠j(yi/yj,q)∞(t​yi/yj,q)∞.\displaystyle\omega_{m,M}(y;w)=y^{\alpha}\Pi_{m,M}(y;w)\rho(y),\quad\rho(y)=\prod_{i\neq j}\frac{(y_{i}/y_{j};q)_{\infty}}{(ty_{i}/y_{j};q)_{\infty}}. (2.10)
Remark 2.3.

In [9, Theorem 1], another integral transform in terms of Jackson integral is constructed. Here we state only the part relevant to the present paper.

The proof of Theorem 2.2 will be given after we prove several lemmas. These lemmas are obtained in [9].

Lemma 2.4.

We put

Dy​(u)=∑I⊂{1,…,m}(−u)|I|​AI​(y)​Tq,yI,AI​(y)=tr⁡(r−1)/2​∏i∈I,j∉It​yi−yjyi−yj.\displaystyle D_{y}(u)=\sum_{I\subset\{1,\ldots,m\}}(-u)^{|I|}A_{I}(y)T_{q,y}^{I},\quad A_{I}(y)=t^{r(r-1)/2}\prod_{i\in I,j\notin I}\frac{ty_{i}-y_{j}}{y_{i}-y_{j}}. (2.11)

Then we have

Tq,y−I​AI​(y)​ρ​(y)=AI​(y−1)​ρ​(y).\displaystyle T_{q,y}^{-I}A_{I}(y)\rho(y)=A_{I}(y^{-1})\rho(y). (2.12)
Proof.

By a simple calculation, we have

Tq,y−I​ρ​(y)ρ⁡(y)=∏i∈I,j∉I1−yi/(q​yj)1−t​yi/(q​yj)​1−t​yj/yi1−yj/yi.\displaystyle\frac{T_{q,y}^{-I}\rho(y)}{\rho(y)}=\prod_{i\in I,j\notin I}\frac{1-y_{i}/(qy_{j})}{1-ty_{i}/(qy_{j})}\frac{1-ty_{j}/y_{i}}{1-y_{j}/y_{i}}. (2.13)

Hence we obtain

Tq,y−I​AI​(y)​ρ​(y)ρ⁡(y)\displaystyle\frac{T_{q,y}^{-I}A_{I}(y)\rho(y)}{\rho(y)} =∏i∈I,j∉I1−yi/(q​yj)1−t​yi/(q​yj)​1−t​yj/yi1−yj/yi​∏i∈I,j∉It​yi/q−yjyi/q−yj\displaystyle=\prod_{i\in I,j\notin I}\frac{1-y_{i}/(qy_{j})}{1-ty_{i}/(qy_{j})}\frac{1-ty_{j}/y_{i}}{1-y_{j}/y_{i}}\prod_{i\in I,j\notin I}\frac{ty_{i}/q-y_{j}}{y_{i}/q-y_{j}}
=∏i∈I,j∉I1−t​yj/yi1−yj/yi=AI​(y−1).\displaystyle=\prod_{i\in I,j\notin I}\frac{1-ty_{j}/y_{i}}{1-y_{j}/y_{i}}=A_{I}(y^{-1}). (2.14)

∎

Lemma 2.5.

Assume α∈ℤ\alpha\in\mathbb{Z} and |t|<|q||t|<|q|. Suppose a function φ⁡(y)\varphi(y) is holomorphic on V={y∣|q|≤|yi|≤1​(1≤i≤m)}V=\{y\mid|q|\leq|y_{i}|\leq 1\ (1\leq i\leq m)\}. If |wi|<1|w_{i}|<1, then

∫𝕋mφ⁡(y)​yα​AI​(y)​(Tq,yI​Π​(y,w))​ρ​(y)​d​yy\displaystyle\int_{\mathbb{T}^{m}}\varphi(y)y^{\alpha}A_{I}(y)\left(T_{q,y}^{I}\Pi(y;w)\right)\rho(y)\frac{dy}{y}
=q−α​|I|​∫𝕋myα​Π​(y,w)​ρ​(y)​AI​(y−1)​(Tq,y−I​φ​(y))​d​yy.\displaystyle=q^{-\alpha|I|}\int_{\mathbb{T}^{m}}y^{\alpha}\Pi(y;w)\rho(y)A_{I}(y^{-1})\left(T_{q,y}^{-I}\varphi(y)\right)\frac{dy}{y}. (2.15)
Proof.

By the change of variables y→q−I​yy\to q^{-I}y, we have

∫𝕋mφ⁡(y)​yα​AI​(y)​(Tq,yI​Π​(y,w))​ρ​(y)​d​yy\displaystyle\int_{\mathbb{T}^{m}}\varphi(y)y^{\alpha}A_{I}(y)\left(T_{q,y}^{I}\Pi(y;w)\right)\rho(y)\frac{dy}{y}
=q−α​|I|​∫qI​𝕋myα​(Tq,y−I​AI​(y)​ρ​(y)​φ​(y))​Π​(y,w)​d​yy.\displaystyle=q^{-\alpha|I|}\int_{q^{I}\mathbb{T}^{m}}y^{\alpha}\left(T_{q,y}^{-I}A_{I}(y)\rho(y)\varphi(y)\right)\Pi(y;w)\frac{dy}{y}. (2.16)

By Lemma 2.4, we get

∫qI​𝕋myα​(Tq,y−I​AI​(y)​ρ​(y)​φ​(y))​Π​(y,w)​d​yy\displaystyle\int_{q^{I}\mathbb{T}^{m}}y^{\alpha}\left(T_{q,y}^{-I}A_{I}(y)\rho(y)\varphi(y)\right)\Pi(y;w)\frac{dy}{y}
=\displaystyle= ∫qI​𝕋myα​(Tq,y−I​φ​(y))​AI​(y−1)​ρ​(y)​Π​(y,w)​d​yy.\displaystyle\int_{q^{I}\mathbb{T}^{m}}y^{\alpha}\left(T_{q,y}^{-I}\varphi(y)\right)A_{I}(y^{-1})\rho(y)\Pi(y;w)\frac{dy}{y}. (2.17)

The poles of the integrand (2.17) on the right-hand side for y∈(ℂ∖{0})my\in(\mathbb{C}\setminus\{0\})^{m} are as follows:

yi−q−k/wj=0(k∈ℤ≥0, 1≤i≤m, 1≤j≤M),\displaystyle y_{i}-q^{-k}/w_{j}=0\quad\left(k\in\mathbb{Z}_{\geq 0},\ 1\leq i\leq m,\ 1\leq j\leq M\right),
ya−t​qk​yb=0(k∈ℤ≥0, 1≤a,b≤m).\displaystyle y_{a}-tq^{k}y_{b}=0\quad\left(k\in\mathbb{Z}_{\geq 0},\ 1\leq a,b\leq m\right). (2.18)

From assumptions |wj|<1|w_{j}|<1 and |t|<|q||t|<|q|, the integrand (2.17) on the right-hand side is holomorphic on VV. Therefore, we obtain

∫qI​𝕋myα​(Tq,y−I​φ​(y))​AI​(y−1)​ρ​(y)​Π​(y,w)​d​yy\displaystyle\int_{q^{I}\mathbb{T}^{m}}y^{\alpha}\left(T_{q,y}^{-I}\varphi(y)\right)A_{I}(y^{-1})\rho(y)\Pi(y;w)\frac{dy}{y}
=∫𝕋myα​(Tq,y−I​φ​(y))​AI​(y−1)​ρ​(y)​Π​(y,w)​d​yy\displaystyle=\int_{\mathbb{T}^{m}}y^{\alpha}\left(T_{q,y}^{-I}\varphi(y)\right)A_{I}(y^{-1})\rho(y)\Pi(y;w)\frac{dy}{y} (2.19)

From this equation, we obtain the desired equation (2.15). ∎

We give the proof of Theorem 2.2, following [9].

Proof of Theorem 2.2.

By the kernel identity (2.4), we get

Dw​(u)​Ω​(w)\displaystyle D_{w}(u)\Omega(w) =∫𝕋mφ⁡(y)​yα​(Dw​(u)​Π​(y,w))​ρ​(y)​d​yy\displaystyle=\int_{\mathbb{T}^{m}}\varphi(y)y^{\alpha}\left(D_{w}(u)\Pi(y;w)\right)\rho(y)\frac{dy}{y}
=(u,t)M−m​∫𝕋mφ⁡(y)​yα​(Dy​(u)​Π​(y,w))​ρ​(y)​d​yy.\displaystyle=(u;t)_{M-m}\int_{\mathbb{T}^{m}}\varphi(y)y^{\alpha}\left(D_{y}(u)\Pi(y;w)\right)\rho(y)\frac{dy}{y}. (2.20)

By Lemma 2.5, we have

∫𝕋mφ⁡(y)​yα​(Dy​(u)​Π​(y,w))​ρ​(y)​d​yy\displaystyle\int_{\mathbb{T}^{m}}\varphi(y)y^{\alpha}\left(D_{y}(u)\Pi(y;w)\right)\rho(y)\frac{dy}{y}
=∑I⊂{1,…,m}∫𝕋m(−u)|I|​φ​(y)​yα​AI​(y)​(Tq,yI​Π​(y,w))​ρ​(y)​d​yy\displaystyle=\sum_{I\subset\{1,\ldots,m\}}\int_{\mathbb{T}^{m}}(-u)^{|I|}\varphi(y)y^{\alpha}A_{I}(y)\left(T_{q,y}^{I}\Pi(y;w)\right)\rho(y)\frac{dy}{y}
=∑I⊂{1,…,m}∫𝕋mφ⁡(y)​yα​Π​(y,w)​ρ​(y)​(u/qα)|I|​AI​(y−1)​(Tq,y−I​φ​(y))​d​yy.\displaystyle=\sum_{I\subset\{1,\ldots,m\}}\int_{\mathbb{T}^{m}}\varphi(y)y^{\alpha}\Pi(y;w)\rho(y)(u/q^{\alpha})^{|I|}A_{I}(y^{-1})\left(T_{q,y}^{-I}\varphi(y)\right)\frac{dy}{y}. (2.21)

By assumption (2.7), we get

∑I⊂{1,…,m}∫𝕋mφ⁡(y)​yα​Π​(y,w)​ρ​(y)​(u/qα)|I|​AI​(y−1)​(Tq,y−I​φ​(y))​d​yy\displaystyle\sum_{I\subset\{1,\ldots,m\}}\int_{\mathbb{T}^{m}}\varphi(y)y^{\alpha}\Pi(y;w)\rho(y)(u/q^{\alpha})^{|I|}A_{I}(y^{-1})\left(T_{q,y}^{-I}\varphi(y)\right)\frac{dy}{y}
=∫𝕋m(Dy−1​(u/qα)​φ​(y))​ωm,M​(y,w)​d​yy=E​∫𝕋mφ⁡(y)​ωm,M​(y,w)​d​yy.\displaystyle=\int_{\mathbb{T}^{m}}\left(D_{y^{-1}}(u/q^{\alpha})\varphi(y)\right)\omega_{m,M}(y;w)\frac{dy}{y}=E\int_{\mathbb{T}^{m}}\varphi(y)\omega_{m,M}(y;w)\frac{dy}{y}. (2.22)

Hence we have the desired equation (2.9). ∎

3 Integral transform for eigenproblem of deformed Noumi–Sano operator

In this section, we construct an integral transform for deformed Noumi–Sano operators using the method of Mimachi–Noumi [9].

The deformed Noumi–Sano operator Hn,m​(x,y,u)H_{n,m}(x,y;u) satisfies the following kernel identity.

Theorem 3.1 (Hallnäs–Langmann–Noumi–Rosengren [3]).

The following identity holds.

HN,M​(z,w,u)​Ψn,m;N,M​(x,y,z,w)\displaystyle H_{N,M}(z,w;u)\Psi_{n,m;N,M}(x,y;z,w)
=(t1−n​qm​u,q)∞(t1−N​qM​u,q)∞​Hn,m​(x,y,u)​Ψn,m;N,M​(x,y,z,w).\displaystyle=\frac{(t^{1-n}q^{m}u;q)_{\infty}}{(t^{1-N}q^{M}u;q)_{\infty}}H_{n,m}(x,y;u)\Psi_{n,m;N,M}(x,y;z,w). (3.1)

Here Ψn,m;N,M​(x,y,z,w)\Psi_{n,m;N,M}(x,y;z,w) is given in (1.5). In particular, we have

HN,Md​(z,w)​Ψn,m;N,M​(x,y,z,w)\displaystyle H_{N,M}^{d}(z,w)\Psi_{n,m;N,M}(x,y;z,w)
=∑k+l=d(tN−n​qm−M,q)k(q,q)k​(t1−N​qM)k​Hn,ml​(x,y)​Ψn,m;N,M​(x,y,z,w).\displaystyle=\sum_{k+l=d}\frac{(t^{N-n}q^{m-M};q)_{k}}{(q;q)_{k}}(t^{1-N}q^{M})^{k}H_{n,m}^{l}(x,y)\Psi_{n,m;N,M}(x,y;z,w). (3.2)

The identity (3.1) follows from Kajihara’s transformation formula [6]. We omit the proof. For details, see [3]. We note that Ψ\Psi in [3] is defined as

∏i=1n∏j=1N(t​xi​zj,q)∞(xi​zj,q)∞​∏i=1m∏j=1M(q​t​yi​wj,t)∞(t​yi​wj,t)∞​∏i=1n∏j=1M(1−t​xi​wj)​∏i=1m∏j=1N(1−t​yi​zj).\displaystyle\prod_{i=1}^{n}\prod_{j=1}^{N}\frac{(tx_{i}z_{j};q)_{\infty}}{(x_{i}z_{j};q)_{\infty}}\prod_{i=1}^{m}\prod_{j=1}^{M}\frac{(qty_{i}w_{j};t)_{\infty}}{(ty_{i}w_{j};t)_{\infty}}\prod_{i=1}^{n}\prod_{j=1}^{M}(1-tx_{i}w_{j})\prod_{i=1}^{m}\prod_{j=1}^{N}(1-ty_{i}z_{j}). (3.3)

However, this Ψ\Psi does not satisfy the kernel identity (3.1). We slightly modify Ψ\Psi so that it satisfies the kernel identity.

To prove Theorem 1.1, we prove several lemmas.

Lemma 3.2.
Tq,x−μ​Tt,yI​(Bμ,I​(x,y)​ρ​(x,y))=Bμ,I​(t​x−1,q−1​y−1)​ρ​(x,y).\displaystyle T_{q,x}^{-\mu}T_{t,y}^{I}(B_{\mu,I}(x,y)\rho(x,y))=B_{\mu,I}(tx^{-1},q^{-1}y^{-1})\rho(x,y). (3.4)
Proof.

We put

F0​(x,y)=∏i=1n(∏j∈I1−xi/(t​yj)1−qμi​xi/yj​∏j∉I1−xi/(q​yj)1−qμi−1​xi/yj),\displaystyle F_{0}(x,y)=\prod_{i=1}^{n}\left(\prod_{j\in I}\frac{1-x_{i}/(ty_{j})}{1-q^{\mu_{i}}x_{i}/y_{j}}\prod_{j\notin I}\frac{1-x_{i}/(qy_{j})}{1-q^{\mu_{i}-1}x_{i}/y_{j}}\right), (3.5)
G0​(x,y)=∏i=1n∏j=1m1(1−q−1​xi/yj)​(1−t​yj/xi),\displaystyle G_{0}(x,y)=\prod_{i=1}^{n}\prod_{j=1}^{m}\frac{1}{(1-q^{-1}x_{i}/y_{j})(1-ty_{j}/x_{i})}, (3.6)
F1​(x)=Δ⁡(qμ​x)Δ⁡(x)​∏i,j=1n(t​xi/xj,q)μi(q​xi/xj,q)μi,\displaystyle F_{1}(x)=\frac{\Delta(q^{\mu}x)}{\Delta(x)}\prod_{i,j=1}^{n}\frac{(tx_{i}/x_{j};q)_{\mu_{i}}}{(qx_{i}/x_{j};q)_{\mu_{i}}}, (3.7)
G1​(x)=∏1≤i≠j≤n(xi/xj,q)∞(t​xi/xj,q)∞,\displaystyle G_{1}(x)=\prod_{1\leq i\neq j\leq n}\frac{(x_{i}/x_{j};q)_{\infty}}{(tx_{i}/x_{j};q)_{\infty}}, (3.8)
F2​(y)=∏1≤i,j≤mi∈I,j∉Iyi−q​yjyi−yj,\displaystyle F_{2}(y)=\prod_{\begin{subarray}{c}1\leq i,j\leq m\\ i\in I,j\notin I\end{subarray}}\frac{y_{i}-qy_{j}}{y_{i}-y_{j}}, (3.9)
G2​(y)=∏1≤i≠j≤m(yi/yj,t)∞(q​yi/yj,t)∞.\displaystyle G_{2}(y)=\prod_{1\leq i\neq j\leq m}\frac{(y_{i}/y_{j};t)_{\infty}}{(qy_{i}/y_{j};t)_{\infty}}. (3.10)

By simple calculation, we have

Tq,x−μ​Tt,yI​(F0​(x,y)​G0​(x,y))=F0​(t/x,1/(q​y))​G0​(x,y),\displaystyle T_{q,x}^{-\mu}T_{t,y}^{I}(F_{0}(x,y)G_{0}(x,y))=F_{0}(t/x,1/(qy))G_{0}(x,y), (3.11)
Tq,x−μ​(F1​(x)​G1​(x))=F1​(t/x)​G1​(x),\displaystyle T_{q,x}^{-\mu}(F_{1}(x)G_{1}(x))=F_{1}(t/x)G_{1}(x), (3.12)
Tt,yI​(F2​(y)​G2​(y))=F2​(1/(q​y))​G2​(y).\displaystyle T_{t,y}^{I}(F_{2}(y)G_{2}(y))=F_{2}(1/(qy))G_{2}(y). (3.13)

Since Bμ,I​(x,y)​ρ​(x,y)=F0​(x,y)​G0​(x,y)​F1​(x)​G1​(x)​F2​(y)​G2​(y)B_{\mu,I}(x,y)\rho(x,y)=F_{0}(x,y)G_{0}(x,y)F_{1}(x)G_{1}(x)F_{2}(y)G_{2}(y), we obtain the desired identity (3.4). ∎

Remark 3.3.

The case |I|=|μ|=1|I|=|\mu|=1 of Lemma 3.2 was obtained in [1].

Lemma 3.4.

Assume r<|q|maxi⁡|μi|​Rr<|q|^{\max_{i}{|\mu_{i}|}}R and |q|<|t||q|<|t|, where r,R∈ℝ>0r,R\in\mathbb{R}_{>0}. Suppose a function φ⁡(x,y)\varphi(x,y) is holomorphic on D={(x,y)∣|q|μiR≤|xi|≤R,r≤|yj|≤|t|−1r(1≤i≤n, 1≤j≤m)}D=\{(x,y)\mid|q|^{\mu_{i}}R\leq|x_{i}|\leq R,\,r\leq|y_{j}|\leq|t|^{-1}r\,(1\leq i\leq n,\,1\leq j\leq m)\}. If |zj|<|t|/R,|wj|<1/r|z_{j}|<|t|/R,\,|w_{j}|<1/r, then we have

∫Cμxα​yβ​φ​(x,y)​ρ​(x,y)​Bμ,I​(x,y)​(Tq,xμ​Tt,y−I​Ψn,m;N,M​(x,y,z,w))​d​xx​d​yy\displaystyle\int_{C_{\mu}}x^{\alpha}y^{\beta}\varphi(x,y)\rho(x,y)B_{\mu,I}(x,y)\left(T_{q,x}^{\mu}T_{t,y}^{-I}\Psi_{n,m;N,M}(x,y;z,w)\right)\frac{dx}{x}\frac{dy}{y}
=\displaystyle= q−α​|μ|​tβ​|I|​∫Cμ(xα​yβ​(Tq,x−μ​Tt,yI​φ​(x,y))​ρ​(x,y)​Bμ,I​(t​x−1,q−1​y−1)CLOSE\displaystyle q^{-\alpha|\mu|}t^{\beta|I|}\int_{C_{\mu}}\bigg(x^{\alpha}y^{\beta}\left(T_{q,x}^{-\mu}T_{t,y}^{I}\varphi(x,y)\right)\rho(x,y)B_{\mu,I}(tx^{-1},q^{-1}y^{-1})
×Ψn,m;N,M(x,y;z,w))d​xxd​yy,\displaystyle\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\times\Psi_{n,m;N,M}(x,y;z,w)\bigg)\frac{dx}{x}\frac{dy}{y}, (3.14)

where Cμ=𝕋Rn×𝕋rmC_{\mu}=\mathbb{T}^{n}_{R}\times\mathbb{T}^{m}_{r} and α,β∈ℤ\alpha,\beta\in\mathbb{Z}.

Proof.

By the change of variables x→q−μ​x,y→tI​yx\to q^{-\mu}x,\ y\to t^{I}y, we have

∫Cμxα​yβ​φ​(x,y)​ρ​(x,y)​Bμ,I​(x,y)​(Tq,xμ​Tt,y−I​Ψn,m;N,M​(x,y,z,w))​d​xx​d​yy\displaystyle\int_{C_{\mu}}x^{\alpha}y^{\beta}\varphi(x,y)\rho(x,y)B_{\mu,I}(x,y)\left(T_{q,x}^{\mu}T_{t,y}^{-I}\Psi_{n,m;N,M}(x,y;z,w)\right)\frac{dx}{x}\frac{dy}{y}
=\displaystyle= q−α​|μ|​tβ​|I|​∫qμ​𝕋Rn×t−I​𝕋rm(xα​yβ​(Tq,x−μ​Tt,yI​φ​(x,y)​ρ​(x,y)​Bμ,I​(x,y))CLOSE\displaystyle q^{-\alpha|\mu|}t^{\beta|I|}\int_{q^{\mu}\mathbb{T}^{n}_{R}\times t^{-I}\mathbb{T}^{m}_{r}}\bigg(x^{\alpha}y^{\beta}\left(T_{q,x}^{-\mu}T_{t,y}^{I}\varphi(x,y)\rho(x,y)B_{\mu,I}(x,y)\right)
×Ψn,m;N,M(x,y;z,w))d​xxd​yy.\displaystyle\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\times\Psi_{n,m;N,M}(x,y;z,w)\bigg)\frac{dx}{x}\frac{dy}{y}. (3.15)

By Lemma 3.2, we obtain

∫qμ​𝕋Rn×t−I​𝕋rmxα​yβ​(Tq,x−μ​Tt,yI​φ​(x,y)​ρ​(x,y)​Bμ,I​(x,y))​Ψn,m;N,M​(x,y,z,w)​d​xx​d​yy\displaystyle\int_{q^{\mu}\mathbb{T}^{n}_{R}\times t^{-I}\mathbb{T}^{m}_{r}}x^{\alpha}y^{\beta}\left(T_{q,x}^{-\mu}T_{t,y}^{I}\varphi(x,y)\rho(x,y)B_{\mu,I}(x,y)\right)\Psi_{n,m;N,M}(x,y;z,w)\frac{dx}{x}\frac{dy}{y}
=\displaystyle= ∫qμ​𝕋Rn×t−I​𝕋rm(xα​yβ​(Tq,x−μ​Tt,yI​φ​(x,y))​ρ​(x,y)​Bμ,I​(t​x−1,q−1​y−1)CLOSE\displaystyle\int_{q^{\mu}\mathbb{T}^{n}_{R}\times t^{-I}\mathbb{T}^{m}_{r}}\bigg(x^{\alpha}y^{\beta}\left(T_{q,x}^{-\mu}T_{t,y}^{I}\varphi(x,y)\right)\rho(x,y)B_{\mu,I}(tx^{-1},q^{-1}y^{-1})
×Ψn,m;N,M(x,y;z,w))d​xxd​yy.\displaystyle\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\times\Psi_{n,m;N,M}(x,y;z,w)\bigg)\frac{dx}{x}\frac{dy}{y}. (3.16)

The poles of the integrand (3.16) on the right-hand side for (x,y)∈(ℂ∖{0})n+m(x,y)\in(\mathbb{C}\setminus\{0\})^{n+m} are as follows:

xi−q−kt/zj=0(k∈ℤ≥0, 1≤i≤n, 1≤j≤N),\displaystyle x_{i}-q^{-k}t/z_{j}=0\quad(k\in\mathbb{Z}_{\geq 0},\,1\leq i\leq n,\,1\leq j\leq N), (3.17)
yi−t−k−1/wj=0(k∈ℤ≥0, 1≤i≤m, 1≤j≤M),\displaystyle y_{i}-t^{-k-1}/w_{j}=0\quad(k\in\mathbb{Z}_{\geq 0},\,1\leq i\leq m,\,1\leq j\leq M), (3.18)
xi−tqk+μixj=0(k∈ℤ≥0, 1≤i≠j≤n),\displaystyle x_{i}-tq^{k+\mu_{i}}x_{j}=0\quad(k\in\mathbb{Z}_{\geq 0},\,1\leq i\neq j\leq n), (3.19)
yi−qtkyj=0(k∈ℤ≥0, 1≤i≠j≤m),\displaystyle y_{i}-qt^{k}y_{j}=0\quad(k\in\mathbb{Z}_{\geq 0},\,1\leq i\neq j\leq m), (3.20)
xi−qyj=0( 1≤i≤n, 1≤j≤m),\displaystyle x_{i}-qy_{j}=0\quad(\ 1\leq i\leq n,\,1\leq j\leq m), (3.21)
xi−tyj=0( 1≤i≤n, 1≤j≤m),\displaystyle x_{i}-ty_{j}=0\quad(\ 1\leq i\leq n,\,1\leq j\leq m), (3.22)
xi−qμi+1tyj=0( 1≤i≤n,j∈I),\displaystyle x_{i}-q^{\mu_{i}+1}ty_{j}=0\quad(\ 1\leq i\leq n,\,j\in I), (3.23)
xi−qμityj=0( 1≤i≤n,j∉I).\displaystyle x_{i}-q^{\mu_{i}}ty_{j}=0\quad(\ 1\leq i\leq n,\,j\notin I). (3.24)

From assumptions |zj|<|t|/R,|wj|<1/r|z_{j}|<|t|/R,\,|w_{j}|<1/r and |q|<|t||q|<|t|, the integrand (3.16) on the right-hand side is holomorphic on DD. Therefore, we obtain

∫qμ​𝕋Rn×t−I​𝕋rm(xα​yβ​(Tq,x−μ​Tt,yI​φ​(x,y))​ρ​(x,y)​Bμ,I​(t​x−1,q−1​y−1)CLOSE\displaystyle\int_{q^{\mu}\mathbb{T}^{n}_{R}\times t^{-I}\mathbb{T}^{m}_{r}}\bigg(x^{\alpha}y^{\beta}\left(T_{q,x}^{-\mu}T_{t,y}^{I}\varphi(x,y)\right)\rho(x,y)B_{\mu,I}(tx^{-1},q^{-1}y^{-1})
×Ψn,m;N,M(x,y;z,w))d​xxd​yy\displaystyle\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\times\Psi_{n,m;N,M}(x,y;z,w)\bigg)\frac{dx}{x}\frac{dy}{y}
=\displaystyle= ∫𝕋Rn×𝕋rm(xα​yβ​(Tq,x−μ​Tt,yI​φ​(x,y))​ρ​(x,y)​Bμ,I​(t​x−1,q−1​y−1)CLOSE\displaystyle\int_{\mathbb{T}^{n}_{R}\times\mathbb{T}^{m}_{r}}\bigg(x^{\alpha}y^{\beta}\left(T_{q,x}^{-\mu}T_{t,y}^{I}\varphi(x,y)\right)\rho(x,y)B_{\mu,I}(tx^{-1},q^{-1}y^{-1})
×Ψn,m;N,M(x,y;z,w))d​xxd​yy.\displaystyle\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\times\Psi_{n,m;N,M}(x,y;z,w)\bigg)\frac{dx}{x}\frac{dy}{y}. (3.25)

From this equation, we obtain the desired equation (3.14). ∎

Remark 3.5.

The integrand on the right-hand side of (3.16) is holomorphic on DD if we assume R<|q|maxi⁡|μi|+1​|t|​rR<|q|^{\max_{i}{|\mu_{i}|}+1}|t|r instead of r<|q|maxi⁡|μi|​Rr<|q|^{\max_{i}{|\mu_{i}|}}R in Lemma 3.4. Hence, Lemma 3.4 holds with r<|q|maxi⁡|μi|​Rr<|q|^{\max_{i}{|\mu_{i}|}}R replaced by R<|q|maxi⁡|μi|+1​|t|​rR<|q|^{\max_{i}{|\mu_{i}|}+1}|t|r. Therefore, the condition r<|q|d​Rr<|q|^{d}R in Theorem 3.7 below can be replaced by R<|q|d+1​|t|​rR<|q|^{d+1}|t|r.

Remark 3.6.

Because of the poles (3.21)–(3.24), the radii rr and RR of path CμC_{\mu} depend on μ\mu. When m=0m=0, the poles (3.21)–(3.24) do not appear. Therefore, the path can be taken independently of μ\mu. In particular, an integral transform for eigenvalue problem of the generating function Hn,0​(x,u)H_{n,0}(x;u) for Noumi–Sano operator can be constructed. See Appendix A for details.

We restate the main result.

Theorem 3.7.

Assume qα=t−βq^{\alpha}=t^{-\beta}, r<|q|d​Rr<|q|^{d}R, and |q|<|t||q|<|t|, where α,β∈ℤ\alpha,\beta\in\mathbb{Z} and r,R∈ℝ>0r,R\in\mathbb{R}_{>0}. Suppose that a function φ⁡(x,y)\varphi(x,y) satisfies the following conditions:

  • 1.

    The function φ⁡(x,y)\varphi(x,y) is a simultaneous eigenfunction of Hn,mkH^{k}_{n,m} (0≤k≤d)(0\leq k\leq d), i.e.

    Hn,mk​(t​x−1,q−1​y−1)​φ​(x,y)=Ek​φ​(x,y),0≤k≤d.\displaystyle H^{k}_{n,m}(tx^{-1},q^{-1}y^{-1})\varphi(x,y)=E_{k}\varphi(x,y),\quad 0\leq k\leq d. (3.26)
  • 2.

    The function φ⁡(x,y)\varphi(x,y) is holomorphic on D={(x,y)∣|q|dR≤|xi|≤R,r≤|yj|≤|t|−1r(1≤i≤n, 1≤j≤m)}D=\{(x,y)\mid|q|^{d}R\leq|x_{i}|\leq R,\,r\leq|y_{j}|\leq|t|^{-1}r\,(1\leq i\leq n,\,1\leq j\leq m)\}.

We put Cd=𝕋Rn×𝕋rmC_{d}=\mathbb{T}^{n}_{R}\times\mathbb{T}^{m}_{r}, where 𝕋ps\mathbb{T}^{s}_{p} denotes the ss-dimensional torus of radius pp. If |zj|<|t|/R,|wj|<1/r|z_{j}|<|t|/R,\,|w_{j}|<1/r, then the function

Ω⁡(z,w)=∫Cdφ⁡(x,y)​ωn,m;N,M​(x,y,z,w)​d​xx​d​yy\displaystyle\Omega(z,w)=\int_{C_{d}}\varphi(x,y)\omega_{n,m;N,M}(x,y;z,w)\frac{dx}{x}\frac{dy}{y} (3.27)

satisfies

HN,Md​(z,w)​Ω​(z,w)=(∑k+l=d(tN−n​qm−M,q)k(q,q)k​(t1−N​qM)k​tβ​l​El)​Ω​(z,w).\displaystyle H^{d}_{N,M}(z,w)\Omega(z,w)=\left(\sum_{k+l=d}\frac{(t^{N-n}q^{m-M};q)_{k}}{(q;q)_{k}}(t^{1-N}q^{M})^{k}t^{\beta l}E_{l}\right)\Omega(z,w). (3.28)
Proof.

Using formula (3.2), we have

HN,Md​(z,w)​Ω​(z,w)\displaystyle H^{d}_{N,M}(z,w)\Omega(z,w)
=∑k+l=d((tN−n​qm−M,q)k(q,q)k​(t1−N​qM)k​tβ​lCLOSE\displaystyle=\sum_{k+l=d}\bigg(\frac{(t^{N-n}q^{m-M};q)_{k}}{(q;q)_{k}}(t^{1-N}q^{M})^{k}t^{\beta l}
×∫Cdφ(x,y)xαyβ(Hn,ml(x,y)Ψn,m;N,M(x,y;z,w))ρ(x,y)d​xxd​yy).\displaystyle\quad\quad\quad\times\int_{C_{d}}\varphi(x,y)x^{\alpha}y^{\beta}\left(H_{n,m}^{l}(x,y)\Psi_{n,m;N,M}(x,y;z,w)\right)\rho(x,y)\frac{dx}{x}\frac{dy}{y}\bigg). (3.29)

By Lemma 3.4,

∫Cdφ⁡(x,y)​xα​yβ​(Hn,ml​(x,y)​Ψn,m;N,M​(x,y,z,w))​ρ​(x,y)​d​xx​d​yy\displaystyle\int_{C_{d}}\varphi(x,y)x^{\alpha}y^{\beta}\left(H_{n,m}^{l}(x,y)\Psi_{n,m;N,M}(x,y;z,w)\right)\rho(x,y)\frac{dx}{x}\frac{dy}{y}
=∑|μ|+|I|=l∫Cdφ⁡(x,y)​xα​yβ​Bμ,I​(x,y)​(Tq,xμ​Tt,y−I​Ψn,m;N,M​(x,y,z,w))​ρ​(x,y)​d​xx​d​yy\displaystyle=\sum_{|\mu|+|I|=l}\int_{C_{d}}\varphi(x,y)x^{\alpha}y^{\beta}B_{\mu,I}(x,y)\left(T_{q,x}^{\mu}T_{t,y}^{-I}\Psi_{n,m;N,M}(x,y;z,w)\right)\rho(x,y)\frac{dx}{x}\frac{dy}{y}
=∑|μ|+|I|=lq−α​|μ|​tβ​|I|​∫Cd(xα​yβ​(Tq,x−μ​Tt,yI​φ​(x,y))​ρ​(x,y)​Bμ,I​(t​x−1,q−1​y−1)CLOSE\displaystyle=\sum_{|\mu|+|I|=l}q^{-\alpha|\mu|}t^{\beta|I|}\int_{C_{d}}\bigg(x^{\alpha}y^{\beta}\left(T_{q,x}^{-\mu}T_{t,y}^{I}\varphi(x,y)\right)\rho(x,y)B_{\mu,I}(tx^{-1},q^{-1}y^{-1})
×Ψn,m;N,M(x,y;z,w))d​xxd​yy.\displaystyle\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\times\Psi_{n,m;N,M}(x,y;z,w)\bigg)\frac{dx}{x}\frac{dy}{y}. (3.30)

By assumption qα=t−βq^{\alpha}=t^{-\beta}, we obtain

∑|μ|+|I|=dq−α​|μ|​tβ​|I|​∫Cd(xα​yβ​(Tq,x−μ​Tt,yI​φ​(x,y))​ρ​(x,y)​Bμ,I​(t​x−1,q−1​y−1)CLOSE\displaystyle\sum_{|\mu|+|I|=d}q^{-\alpha|\mu|}t^{\beta|I|}\int_{C_{d}}\bigg(x^{\alpha}y^{\beta}\left(T_{q,x}^{-\mu}T_{t,y}^{I}\varphi(x,y)\right)\rho(x,y)B_{\mu,I}(tx^{-1},q^{-1}y^{-1})
×Ψn,m;N,M(x,y;z,w))d​xxd​yy\displaystyle\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\times\Psi_{n,m;N,M}(x,y;z,w)\bigg)\frac{dx}{x}\frac{dy}{y}
=tβ​l​∫Cdxα​yβ​(Hn,ml​(t​x−1,q−1​y−1)​φ​(x,y))​ρ​(x,y)​Ψn,m;N,M​(x,y,z,w)​d​xx​d​yy.\displaystyle=t^{\beta l}\int_{C_{d}}x^{\alpha}y^{\beta}\left(H^{l}_{n,m}(tx^{-1},q^{-1}y^{-1})\varphi(x,y)\right)\rho(x,y)\Psi_{n,m;N,M}(x,y;z,w)\frac{dx}{x}\frac{dy}{y}. (3.31)

This completes the proof of the theorem. ∎

4 Summary and discussion

In this paper, we gave an integral transform associated with eigenvalue problem of Hn,md​(x,y)H_{n,m}^{d}(x,y) in Theorem 1.1. We obtained this result by the kernel identity (3.1), self-adjointness (3.4) and qq-de Rham theory (3.14).

There are many related problems. We mention two of them.

  • •

    In this paper, we gave an integral transform associated with eigenvalue problem of Hn,md​(x,y)H_{n,m}^{d}(x,y). However, we did not obtain an integral transform associated with eigenvalue problem of the generating function Hn,m​(x,y,u)H_{n,m}(x,y;u). It is important to derive an integral transform associated with eigenvalue problem of Hn,m​(x,y,u)H_{n,m}(x,y;u). We note that an integral transform associated with the generating function Hn,0​(x,u)H_{n,0}(x;u) of Noumi–Sano operators can be constructed by the same method. See Appendix A for details.

  • •

    The Ruijsenaars–Macdonald qq-difference operators and the Noumi–Sano qq-difference operators have elliptic analogues [11, 12]. The deformed Noumi–Sano qq-difference operators have also elliptic analogues [4]. It is important to consider an elliptic analogue of integral transform for deformed Noumi–Sano operators. We remark that an integral transform for the elliptic Ruijsenaars operators is given in [7].

Appendix A The case of Noumi–Sano operator

In this appendix, we construct an integral transform for the eigenvalue problem of the Noumi–Sano operator Hn,0​(x,u)H_{n,0}(x;u). We use the same method as in Section 3.

We recall notations for the case of m=0m=0:

Hn,0​(x,u)=∑μ∈(ℤ≥0)n(t1−n​u)|μ|​Bμ,∅​(x)​Tq,xμ,\displaystyle H_{n,0}(x;u)=\sum_{\mu\in(\mathbb{Z}_{\geq 0})^{n}}(t^{1-n}u)^{|\mu|}B_{\mu,\emptyset}(x)T_{q,x}^{\mu}, (A.1)
Bμ,∅​(x)=Δ⁡(qμ​x)Δ⁡(x)​∏i,j=1n(t​xi/xj,q)μi(q​xi/xj,q)μi,\displaystyle B_{\mu,\emptyset}(x)=\frac{\Delta(q^{\mu}x)}{\Delta(x)}\prod_{i,j=1}^{n}\frac{(tx_{i}/x_{j};q)_{\mu_{i}}}{(qx_{i}/x_{j};q)_{\mu_{i}}}, (A.2)
ρ⁡(x)=∏1≤i≠j≤n(xi/xj,q)∞/(t​xi/xj,q)∞,\displaystyle\rho(x)=\prod_{1\leq i\neq j\leq n}(x_{i}/x_{j};q)_{\infty}/(tx_{i}/x_{j};q)_{\infty}, (A.3)
Ψn,0;N,0​(x,z)=∏i=1n∏j=1N(xi​zj,q)∞(xi​zj/t,q)∞,\displaystyle\Psi_{n,0;N,0}(x;z)=\prod_{i=1}^{n}\prod_{j=1}^{N}\frac{(x_{i}z_{j};q)_{\infty}}{(x_{i}z_{j}/t;q)_{\infty}}, (A.4)
ωn,0;N,0​(x,z)=xα​Ψn,0;N,0​(x,z)​ρ​(x).\displaystyle\omega_{n,0;N,0}(x;z)=x^{\alpha}\Psi_{n,0;N,0}(x;z)\rho(x). (A.5)
Lemma A.1.

Assume R∈ℝ>0R\in\mathbb{R}_{>0} and α∈ℤ\alpha\in\mathbb{Z}. Suppose a function φ⁡(x)\varphi(x) is holomorphic on D={x∣|q|μi​R≤|xi|≤R⁡(1≤i≤n)}D=\{x\mid|q|^{\mu_{i}}R\leq|x_{i}|\leq R\ (1\leq i\leq n)\}. If |zj|<|t|/R|z_{j}|<|t|/R, then we have

∫Cxα​φ​(x)​ρ​(x)​Bμ,∅​(x)​(Tq,xμ​Ψn,0;N,0​(x,z))​d​xx\displaystyle\int_{C}x^{\alpha}\varphi(x)\rho(x)B_{\mu,\emptyset}(x)\left(T_{q,x}^{\mu}\Psi_{n,0;N,0}(x;z)\right)\frac{dx}{x}
=\displaystyle= q−α​|μ|​∫Cxα​(Tq,x−μ​φ​(x))​ρ​(x)​Bμ,∅​(x−1)​Ψn,0;N,0​(x,z)​d​xx,\displaystyle q^{-\alpha|\mu|}\int_{C}x^{\alpha}\left(T_{q,x}^{-\mu}\varphi(x)\right)\rho(x)B_{\mu,\emptyset}(x^{-1})\Psi_{n,0;N,0}(x;z)\frac{dx}{x}, (A.6)

where C=𝕋RnC=\mathbb{T}^{n}_{R}.

This can be checked in a similar manner as the proof of Lemma 3.4. We remark that conditions of the path are independent from μ\mu. Therefore, we get an integral transform for eigenvalue problem of the generating function Hn,0​(x,u)H_{n,0}(x;u) of the Noumi–Sano operator.

Theorem A.2.

Assume α∈ℤ\alpha\in\mathbb{Z} and R∈ℝ>0R\in\mathbb{R}_{>0}. Suppose that a function φ⁡(x)\varphi(x) satisfies the following conditions:

  • 1.

    The function φ⁡(x)\varphi(x) is an eigenfunction of Hn,0​(x,u)H_{n,0}(x;u), i.e.

    Hn,0​(x−1,q−α​u)​φ​(x)=E​φ​(x).\displaystyle H_{n,0}(x^{-1};q^{-\alpha}u)\varphi(x)=E\varphi(x). (A.7)
  • 2.

    The function φ⁡(x)\varphi(x) is holomorphic on D={x∣|q|μi​R≤|xi|≤R⁡(1≤i≤n)}D=\{x\mid|q|^{\mu_{i}}R\leq|x_{i}|\leq R\ (1\leq i\leq n)\}.

We put C=𝕋RnC=\mathbb{T}^{n}_{R}, where 𝕋ps\mathbb{T}^{s}_{p} denotes the ss-dimensional torus of radius pp. If |zj|<|t|/R|z_{j}|<|t|/R, then the function

Ω⁡(z)=∫Cφ⁡(x)​ωn,0;N,0​(x,z)​d​xx\displaystyle\Omega(z)=\int_{C}\varphi(x)\omega_{n,0;N,0}(x;z)\frac{dx}{x} (A.8)

satisfies

Hn,0​(z,u)​Ω​(z)=(E​(t1−n​u,q)∞(t1−N​u,q)∞)​Ω​(z).\displaystyle H_{n,0}(z;u)\Omega(z)=\left(E\frac{(t^{1-n}u;q)_{\infty}}{(t^{1-N}u;q)_{\infty}}\right)\Omega(z). (A.9)

Due to the explicit formula [11, equation (5.17)]

Hn,0​(x,u)​1=(q1−n​u,q)∞(t​u,q)∞​1,\displaystyle H_{n,0}(x;u)1=\frac{(q^{1-n}u;q)_{\infty}}{(tu;q)_{\infty}}1, (A.10)

we obtain an eigenfunction of HN,0​(z,u)H_{N,0}(z;u) as follows:

Corollary A.3.

Assume α∈ℤ\alpha\in\mathbb{Z} and R∈ℝ>0R\in\mathbb{R}_{>0}. If |zj|<|t|/R|z_{j}|<|t|/R, then the following equation holds:

Hn,0​(z,u)​∫𝕋Rnωn,0;N,0​(x,z)​d​xx\displaystyle H_{n,0}(z;u)\int_{\mathbb{T}^{n}_{R}}\omega_{n,0;N,0}(x;z)\frac{dx}{x}
=((q1−n−α​u,q)∞​(t1−n​u,q)∞(t​q−α​u,q)∞​(t1−N​u,q)∞)​∫𝕋Rnωn,0;N,0​(x,z)​d​xx.\displaystyle=\left(\frac{(q^{1-n-\alpha}u;q)_{\infty}(t^{1-n}u;q)_{\infty}}{(tq^{-\alpha}u;q)_{\infty}(t^{1-N}u;q)_{\infty}}\right)\int_{\mathbb{T}^{n}_{R}}\omega_{n,0;N,0}(x;z)\frac{dx}{x}. (A.11)
Remark A.4.

In the case of the deformed Noumi–Sano operator, the path CdC_{d} in Theorem 3.7 depends on μ\mu, so we do not discuss an integral transform associated with eigenvalue problem of the generating function Hn,m​(x,y,u)H_{n,m}(x,y;u).

Acknowledgements

This work is supported by JSPS KAKENHI Grant Number 26KJ0129.

References

  • [1] F. Atai, M. A. Hallnäs and E. Langmann, Super-Macdonald polynomials: orthogonality and Hilbert space interpretation, Comm. Math. Phys. 388 (2021), no. 1, 435–468.
  • [2] M. Feigin and A. V. Silantyev, Generalized Macdonald-Ruijsenaars systems, Adv. Math. 250 (2014), 144–192.
  • [3] M. A. Hallnäs, E. Langmann, M. Noumi and H. Rosengren, From Kajihara’s transformation formula to deformed Macdonald-Ruijsenaars and Noumi-Sano operators, Selecta Math. (N.S.) 28 (2022), no. 2, Paper No. 24, 36 pp.
  • [4] M. A. Hallnäs, E. Langmann, M. Noumi, H. Rosengren, Higher order deformed elliptic Ruijsenaars operators, Comm. Math. Phys. 392 (2022), no. 2, 659–689.
  • [5] Y. Kajihara and M. Noumi, Multiple elliptic hypergeometric series. An approach from the Cauchy determinant, Indag. Math. (N.S.) 14 (2003), no. 3-4, 395–421.
  • [6] Y. Kajihara, Euler transformation formula for multiple basic hypergeometric series of type AA and some applications, Adv. Math. 187 (2004), no. 1, 53–97.
  • [7] E. Langmann, M. Noumi and J. Shiraishi, Construction of eigenfunctions for the elliptic Ruijsenaars difference operators, Comm. Math. Phys. 391 (2022), no. 3, 901–950.
  • [8] I. G. Macdonald, Symmetric functions and Hall polynomials, second edition, Oxford Mathematical Monographs Oxford Science Publications, Oxford Univ. Press, New York, 1995.
  • [9] K. Mimachi and M. Noumi, An integral representation of eigenfunctions for Macdonald’s qq-difference operators, Tohoku Math. J. (2) 49 (1997), no. 4, 517–525.
  • [10] M. Noumi, Macdonald polynomials—commuting family of qq-difference operators and their joint eigenfunctions, SpringerBriefs in Mathematical Physics, 50, Springer, Singapore, 2023.
  • [11] M. Noumi and A. Sano, An infinite family of higher-order difference operators that commute with Ruijsenaars operators of type AA, Lett. Math. Phys. 111 (2021), no. 4, Paper No. 91, 17 pp.
  • [12] S. N. M. Ruijsenaars, Complete integrability of relativistic Calogero-Moser systems and elliptic function identities, Comm. Math. Phys. 110 (1987), no. 2, 191–213.
  • [13] A. N. Sergeev and A. P. Veselov, Deformed Macdonald-Ruijsenaars operators and super Macdonald polynomials, Comm. Math. Phys. 288 (2009), no. 2, 653–675.

Address:

Taikei Fujii

Department of Mathematics, Ochanomizu University, Japan.

E-mail: fujii.taikei@ocha.ac.jp

Takahiko Nobukawa

Department of Mathematical Sciences, Aoyama Gakuin University, Japan.

E-mail: nobukawa@math.aoyama.ac.jp