
Time Zone Notice. All scheduled times are in U.S. Central Standard Time (CST, UTC−6). Each presentation receives 30 minutes (25-minute talk plus 5-minute discussion); the keynote receives 60 minutes.
In this talk I present joint work with Monika Dörfler (Univ. Vienna) and Henry McNulty (Cognite AS) on data augmentation for audio signals with the help of methods from quantum harmonic analysis, which we briefly review and discuss its connection with time-frequency analysis.
We discuss recent results on sharp norm estimates of the positive Bergman operator on the upper half-plane. We focus on off-diagonal boundedness with change of power of the distance to the boundary.
On the subspaces $\{E_L\}_{L\geq 1}\subset L^2(M)$ generated by eigenfunctions associated to the Dirichlet Laplacian, we study the following Question: For which measure sequence $\{\mu_L\}_{L\geq 1}$, does one have: \begin{equation}\label{Question} \int_M|f|^2d\mu_L \approx \int_M|f|^2dV,\qquad \forall f\in E_L, \end{equation} where the constants are independent of $f$ and $L$. The sequence of measures $\mu=\left\{\mu_{L}\right\}_{L\geq 1}$ is an $L^2$-Carleson sequence if there exists $C>0$, \begin{equation}\label{Carleson} \int_{M}|f_{L}|^{2}d\mu_{L}\leq C\int_{M}|f_{L}|^{2}dV, \qquad \forall f_{L}\in E_{L} \end{equation} The sequence of sets $\mathcal{A}=\left\{A_{L}\right\}_{L} \subset M$ is called a Logvinenko-Sereda(L-S) if the estimates (\ref{Question}) holds for the special measure sequence $d\mu_L=\chi_{A_L}dV$. In this talk, we present both positive and negative results regarding the characterization of Carleson measures and Logvinenko-Sereda(L-S) sets for Dirichlet (Neumann) Laplacian on a compact manifold $M$ with smooth boundary, $\partial M$. These results extend and generalize the corresponding work by J. Marzo and J. Ortega-Cerdá [MO] on sphere, and J. Ortega-Cerda and B. Pridhnani [OP] for Laplacian operator on compact boundaryless manifolds. This talk is based on the results from [Xu1, Xu2, Xu3] and joint work [WXZ] with Xin Wang and Cheng Zhang. [MO] J. Marzo and J. Ortega-Cerdá, Equivalent norms for polynomials on the sphere, Int. Math. Res. Not. IMRN (2008), no. 5, Art. ID rnm 154. [OP] J. Ortega-Cerdá, B. Pridhnani, Carleson measures and Logvinenko-Sereda sets on compact manifolds. Forum Math. 2013, Volume 25-1, 151-172. [WXZ] Xin Wang, Xiangjin Xu and Cheng Zhang, Carleson measures and Logvinenko-Sereda sets on compact manifolds with boundary. Advances in Mathematics Volume 494, June 2026, 110958. [Xu1] Xiangjin Xu, Characterization of Carleson Measures via Spectral Estimates on Compact Manifolds with Boundary, Applied Mathematical Analysis and Computations I. SGMC 2021. Springer Proceedings in Mathematics & Statistics, 2024, Volume 471, 1--23. [Xu2] Xiangjin Xu, Characterization of Carleson Measures via Spectral Estimates on Dirichlet Laplacian. (Preprint) 2026. [Xu3] Xiangjin Xu, Carleson measures and Logvinenko-Sereda sets for spectral clusters on compact manifolds (Preprint) 2026.
The Gleason–Kahane–Żelazko theorem characterizes multiplicative linear functionals on a unital Banach algebra by the identity $\Lambda(x^2)=\Lambda(x)^2$.We present analogues for bounded linear functionals on reproducing kernel Hilbert spaces, where multiplicativity is detected by testing only on functions stemming from kernel functions. We treat two large classes. The first is diagonal holomorphic kernels in several variables, covering weighted Bergman and Dirichlet-type spaces and the Hardy space of the polydisc. The second is diagonal Dirichlet series kernels on half-planes, including the Hardy space of Dirichlet series. These results show that the kernel alone can recover algebraic structure, which opens new links between operator theory and function theory in several and infinitely many variables. If time permits, I will also discuss applications to the theory of composition operators. This is a joint work with T. Bhattacharyya, Jaikishan, and P. Kumar.
For classes of self-affine fractal measures, we offer a new analysis of the associated diffusions, Markov processes, semigroups, harmonic analysis, and Dirichlet forms. With view to a new harmonic analysis of pairs of fractal measures, we introduce associated dual-pair systems of operators. A dual-pair analysis consist of an explicit choice of a pair of Hilbert spaces, and two densely defined (unbounded) operators, each one contained in the adjoint of the other.
Suppose \begin{align*} \mathbbm{1}=\begin{bmatrix}1&1&\cdots& 1\\1&1&\cdots&1\\\vdots&\vdots&\ddots&\vdots\\1&1&\cdots&1\end{bmatrix}&\text{ and }& \Phi=\begin{bmatrix}1&\theta_{1,2}&\cdots& \theta_{1,n}\\\theta_{2,1}&1&\cdots&\theta_{2,n}\\\vdots&\vdots&\ddots&\vdots\\\theta_{n,1}&\theta_{n,2}&\cdots&1\end{bmatrix} \end{align*} such that $\lvert\theta_{i,j}\rvert=1$ and $\overline{\theta_{j,i}}=\theta_{i,j}$ for $i\neq j$ . Let the function algebra of the unitary group $U(n)$ is denoted by $ U_{1,\mathbbm{1}}(n).$ The classical unitary group $U(n)$ admits a quantum deformation $U_{q,\mathbbm{1}}(n)$, which forms a compact quantum group in the sense of Woronowicz for $q\in(0,\infty)\setminus{1}$. A two-parameter extension, denoted by $U_{q,\Phi}(2)$, was constructed using the Faddeev–Reshetikhin–Takhtajan (FRT) approach and Hayashi's R-matrix. This construction recovers the standard quantum unitary group when the deformation parameters are trivial and the Connes–Dubois-Violette construction when $q=1$. These results naturally motivate the study of analogous structures in higher dimensions. In this talk, we present a construction of $U_{q,\Phi}(n)$ as a compact quantum group for arbitrary $n$. We then focus on the three-dimensional case, $U_{q,\Phi}(3)$, and classify all irreducible -representations of its associated $C^*$-algebra $C(U_{q,\Phi}(3))$. We establish a connection between these representations and the irreducible representations of a rotational $C^*$-algebra generated by three unitaries. Finally, we investigate the irreducible representations of the crystallized $C^*$-algebra $C(U_{0,\Phi}(3))$, which arises in the limiting case $q=0$.
The numerical radius w(A) of a bounded linear operator A on a complex Hilbert space is a classical quantity in operator theory, closely related to the operator norm, the spectral radius, and the geometry of the numerical range. Beyond its intrinsic interest, it controls the transient growth of non-normal linear dynamics and the convergence of iterative methods, so sharp, computable bounds for it are relevant to the stability of recurrent models and learning algorithms. By Yamazaki’s identity, w(A) is the supremum of the norms of the real parts of the rotations e^{iθ}A. It is therefore built from two ingredients: a real-part map and a norm. Changing either ingredient leads to the generalized, Schatten p-, and weighted numerical radii. In this talk, we give a unified account of these generalizations, with emphasis on the recent framework of Kittaneh and Stojiljković based on the generalized real part Re_{h,g}(A) = hA + gA*. Our main result is an exact formula expressing the square of the generalized numerical radius w^{Re}_{h,g}(A) as a supremum, over unit vectors, of an explicit quadratic expression in ‖Ax‖, ‖A*x‖ and ⟨A²x, x⟩. As consequences, we obtain: the sharp two-sided bound max{|h|,|g|}‖A‖ ≤ w^{Re}{h,g}(A) ≤ (|h|+|g|)‖A‖ for all real h, g, which shows that w^{Re}{h,g} is a norm if and only if (h,g) ≠ (0,0); two-sided bounds in terms of |A|², |A*|² and A²; an exact formula for off-diagonal 2×2 operator matrices, which yields a strict refinement of an inequality of Bani-Domi and Kittaneh. We conclude with open problems and possible directions toward applications in data science.
The classical Plancherel identity shows that the Fourier transform on $\mathbb R^d$ is an isometry between the $L^2$-spaces. We show that for certain unbounded subsets $\Omega$ of $\mathbb R^d$, one can associate a ``dual subset'' $\tilde \Omega$ with a renormalized Lebesgue measure $\mu$ on $\tilde\Omega$, so that the restriction of the Fourier transform to $L^2(\Omega)$ is an isometric isomorphism into $L^2(\tilde \Omega, \mu)$, and thus we have Plancherel identities for the pair $(\Omega,\tilde\Omega)$.
We present a functional model for isometric pairs that satisfy an automorphic covariant relation. The special case when the automorphism is the identity function, the model coincides with the one obtained by Bercovici--Douglas--Foias in their series of papers in the early 2000s. As an application, we obtain a set of unitary invariants for such isometric pairs. The results appear to be new even in the case when the automorphism is a rotaion.
Compactness of Hankel operators on Bergman spaces reflects the interaction between a symbol and the analytic structure of the boundary. In the several complex variables setting, we explain the role of the $\bar\partial$-Neumann problem and discuss geometric criteria on convex domains, including Hankel operators acting on square-integrable $\bar\partial$-closed forms. We then consider extensions beyond convexity and a recent characterization on convex Reinhardt domains in $\mathbb C^2$ using normalized Bergman kernels and the Berezin transform. We end with questions concerning higher form levels and more general domains.
Shift-invariant spaces provide a classical framework for signal representation on the real line. In this talk, we discuss their counterparts on finite graphs, where translations are replaced by graph shifts. We first introduce graph shift-invariant spaces and their connections with bandlimited spaces. We then examine how repeated eigenvalues affect these relationships, using spectral projections to characterize shift-invariant spaces and distinguish polynomial filters from general shift-invariant filters. Finally, motivated by the limitations of exact shift invariance for low-dimensional signal representation, we introduce a data-adaptive approach based on clustering nearby eigenvalues. This leads to approximately shift-invariant spaces that balance approximation accuracy and shift invariance.
We study perturbed exponential systems and unions of shifted frequency lattices through shifted-sinc operators and finite-dimensional fiberizations. For the auxiliary binary local-shift model, we identify the exact pattern-dependent invariant $$ \rho_{\delta,s}=\|P_-T_{2\delta}P_+\|, $$ which gives the optimal squared synthesis bounds $1\pm\rho_{\delta,s}$. For periodic patterns, the synthesis operator reduces exactly to a single constant Vandermonde-type matrix, yielding finite-dimensional basis criteria and optimal bounds. We further prove that the worst binary lower bound is the infimum of its periodic counterparts; at the critical value $\delta=1/4$, a step pattern is governed by a half-shifted Hilbert matrix and has vanishing lower bound, although every fixed periodic pattern remains invertible. We also combine exact bounds for structured patterns with a support-independent perturbation estimate, obtaining explicit stability neighborhoods. For finite unions of bounded intervals and finitely many shifted lattices, folding modulo one yields finitely many rectangular phase matrices whose singular values and ranks give the exact bounds and complete Riesz-sequence, frame, and Riesz-basis criteria. This framework also yields finite Hadamard constraints for periodic spectra and an exact reduction of rationally commensurate multirate sampling.
The Geometric Satake Equivalence--formulated originally by Ginzburg and later proven by Mirković-Vilonen--gives a bridge to compute the geometry of affine Grassmannians of a complex reductive group G in terms of the representation theory of the Langlands dual group. Later work by Nadler and Gaitsgory generalized the equivalence to the cases of real reductive groups and their associated symmetric spaces. In this talk, we will apply these equivalences to compute the cohomologies of the spaces in terms of their Langlands dual groups.
An overdetermined boundary-value problem for the Laplace operator on an unbounded domain will typically not admit solutions, other than for the complements of balls and half-spaces. In two dimensions, the overdetermined problem with constant Dirichlet and constant Neumann data is also supported on a third type of domain, and the solution has an intriguing connection to minimal geometry. The talk will describe this solution and its relation to the Robertson-Walker-Friedmann metric in general relativity.
In Euclidean spaces, every closed, bounded, convex set can be characterized by two equivalent notions of separation properties. This is not true in general for arbitrary Banach spaces. In this work, we present a ball separation characterization for spaces where the unit ball is dentable. We also explore related properties.