Peer Revenue Play credits. Real papers.

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

Mathematics

math Watch Mathematics (sign in to use)

Saturday 18 April

Optimization and Control · Cross-listed from stat.ML

A Mechanism Study of Delayed Loss Spikes in Batch-Normalized Linear Models

Delayed loss spikes have been reported in neural-network training, but existing theory mainly explains earlier non-monotone behavior caused by overly large fixed learning rates. We study one stylized hypothesis: normalization can postpone instability by gradually increasing the effective learning rate during otherwise stable descent. To test this hypothesis at theorem level, we analyze batch-normalized linear models. Our flagship result concerns whitened square-loss linear regression, where we derive explicit no-rising-edge and delayed-onset conditions, bound the waiting time to directional onset, and show that the rising edge self-stabilizes within finitely many iterations. Combined with a square-loss decomposition, this yields a concrete delayed-spike mechanism in the whitened regime. For logistic regression, under highly restrictive active-margin assumptions, we prove only a supporting finite-horizon directional precursor in a knife-edge regime, with an optional appendix-only loss lower bound under an extra non-degeneracy condition. The paper should therefore be read as a stylized mechanism study rather than a general explanation of neural-network loss spikes. Within that scope, the results isolate one concrete delayed-instability pathway induced by batch normalization.
Read the full abstract

All authors Peifeng Gao, Wenyi Fang, Yang Zheng, Difan Zou

Representation Theory

A Non-graded Koszul Duality and Its Applications

Let \(\Lambda\) be a finite-dimensional Koszul algebra with Koszul dual \(\Lambda^!\). We establish derived Koszul dualities at the level of bounded derived categories, both in the graded setting \(\mathsf{D}^{b}(\Lambda\textup{-gmod})\) and in the ungraded setting \(\mathsf{D}^{b}(\Lambda\textup{-mod})\), without imposing finiteness conditions on \(\Lambda^!\). We first prove a graded derived Koszul duality for every finite-dimensional Koszul algebra, with no Noetherian or coherence assumptions on the Koszul dual. We then show that the bounded derived category \(\mathsf{D}^{b}(\Lambda\textup{-mod})\) can be reconstructed from the graded theory as the triangulated hull of an orbit category. This yields a genuinely non-graded derived Koszul duality. We further establish singular and dg refinements of these dualities. For Iwanaga–Gorenstein Koszul algebras, this gives a stable Koszul duality for graded Gorenstein-projective modules and their ungraded counterparts, providing a non-graded form of the Bernstein–Gel'fand–Gel'fand correspondence. As applications, we obtain new descriptions of the bounded derived categories \(\mathsf{D}^{b}(\mathcal{O}_{\lambda})\) for all integral blocks of category \(\mathcal{O}\), including singular blocks, thereby closing a gap left open in the work of Beilinson, Ginzburg, and Soergel. We also establish analogous dualities for certain categories of perverse sheaves arising in geometric representation theory. Finally, we formulate conjectural descriptions of bounded derived and singularity categories of finite-dimensional graded algebras in terms of dg orbit categories.
Read the full abstract

Optimization and Control · Cross-listed from cs.GT

A Stackelberg Game Framework with Drainability Guardrails for Pricing and Scaling in Multi-Tenant GPU Cloud Platforms

Modern Graphics Processing Unit (GPU)-backed services must satisfy strict latency service-level objectives (SLOs) while controlling spare-capacity costs. In multi-tenant GPU cloud platforms, this trade-off is inherently dynamic because workload demand is endogenous; specifically, pricing shapes the submissions of heterogeneous tenants, which subsequently impact congestion and delay. We formulate the joint pricing-and-scaling problem as a large-population Stackelberg game problem, and we derive an explicit equilibrium demand map. The resulting closed-loop model reveals a structural failure mode in which delay-insensitive workloads sustain a residual demand floor, making the backlog undrainable under bounded price and service capacity. This observation motivates a computable drainability guardrail that certifies uniformly negative backlog drift in the residual-demand regime. For any fixed price-capacity pair satisfying the drainability guardrail, we establish global convergence to a unique operating point under a checkable step-size condition. Building on this fixed-pair analysis, we further develop an optimizer-agnostic action shield that provides a negative-drift certificate for shielded execution in the residual regime of the dynamic problem and show empirically that it improves safety and robustness for model-free reinforcement learning (RL) in this setting.
Read the full abstract

All authors Junji Yan, Asrin Efe Yorulmaz, Hanchen Zhou, Tamer Başar

Algebraic Geometry

Implementing p-adic numbers in Macaulay2 using its foreign function interface and FLINT

Macaulay2 is a computer algebra platform widely used by researchers in algebraic geometry and commutative algebra. Using the ForeignFunctions package, it is possible to make calls from Macaulay2 to dynamic libraries such as FLINT. We demonstrate this by introducing a new Macaulay2 package implementing p-adic numbers using FLINT via this interface. We discuss implementation details such as memory allocation, interaction with Macaulay2's garbage collector, and object-oriented design decisions that mirror the existing implementations of the real and complex number fields in Macaulay2.
Read the full abstract

Dynamical Systems

On the well-posedness of linear evolution equations under unbounded nonautonomous perturbations

We study conditions for the well-posedness of nonautonomous perturbation of evolution equations of the form \[ u'(t)=(A+B(t))u(t), \quad t \in [a,b], \] where $A$ generates a $\mathrm{C}_0$-semigroup $\left (T(t)\right )_{t\ge 0}$ with $\| T(t)\| \le Me^{\omega_0 t}$, $t\ge 0$, in a Banach space $\mathbb{X}$ and $B(t)$ are $t$-dependent (unbounded) linear operators in $\mathbb{X}$. The unbounded perturbation operators $B(t)$ are assumed to belong to a normed space (denoted by $\mathcal{GL}_A (\mathbb{X})$) of unbounded linear operators $C$ in $\mathbb{X}$ such that $D(A) \subset D(C)$ with norm \[ \| C\|_A:= (1/M) \sup_{\mu >\omega_0 } \| (\mu-\omega_0) CR(\mu,A)\| <\infty. \] We prove that the above-mentioned evolution equation admits an evolution family if $\| B(\cdot)\|_A$ is continuous in $[a,b]$. The evolution family is unique if $B(\cdot)R(\mu, A)$ as a function $[a,b]\to \mathcal{L}(\mathbb{X})$ is continuously differentiable, and \[ \limsup_{\mu \to\infty} \sup_{t\in [a,b]} \left \| \frac{d}{dt}[B(t)R(\mu,A)]\right \| <\infty. \] Examples are given to illustrate the obtained results.
Read the full abstract

Information Theory · Cross-listed from cs.CV

Generative Semantic Communication via Alternating Dual-Domain Posterior Sampling

Generative semantic communication (SemCom) harnesses pretrained generative priors to improve the perceptual quality of wireless image transmission. Existing generative SemCom receivers, however, rely on maximum a posteriori (MAP) estimation, which fundamentally cannot preserve the data distribution and thus limits achievable perceptual quality. Moreover, current diffusion-based approaches using single-domain guidance face significant limitations: latent-domain guidance is sensitive to channel noise, while image-domain guidance inherits decoder bias. Simply combining both domains simultaneously yields an overconfident pseudo-posterior. In this paper, we formulate semantic decoding as a Bayesian inverse problem and prove that posterior sampling achieves optimal perceptual quality by preserving the data distribution. Building on this insight, we propose alternating dual-domain posterior sampling (ADDPS), a diffusion-based SemCom receiver that alternately enforces latent-domain and image-domain consistency during the sampling process. This alternating strategy decomposes joint posterior sampling into simpler subproblems, avoiding gradient conflicts while retaining the complementary strengths of both domains. Experiments on FFHQ demonstrate that the proposed ADDPS achieves superior perceptual quality compared with existing methods.
Read the full abstract

Probability

Long-Time Behaviors of Branching-Diffusion Processes via Spectral Analysis

We study long-time behaviors for branching-diffusion process corresponding to the drifted Schrödinger operator $\mathcal{L} = \frac{1}{2} \Delta + \langle \nabla V,\nabla \rangle - K$, where $K$ represents the reduction rate of a population dynamics and $\nabla V$ is a given drift term. In particular, we establish exponential convergence rates for the total mass of this process and characterize its quasi-stationary distribution. The proof is based on a novel transformation in spectral analysis, and heat kernel estimates for Schrödinger operators with unbounded potentials. The result is new even in the one-dimensional setting, which especially improves the recent work \cite{CMS}.
Read the full abstract

Functional Analysis

Log-Sobolev and Beckner inequalities and stability of Poincaré inequality with weighted Gaussian measures

We employ a Markov semigroup approach combined with the $\Gamma$-calculus to establish a generalized Beckner inequality associated with weighted Gaussian measures. As a direct consequence, we derive the corresponding Poincaré inequality in the same setting. Subsequently, by means of a duality argument, we investigate gradient and $L^2$ stability estimates of the Poincaré inequality. Furthermore, we formulate a scale-dependent version of the Poincaré inequality for homogeneous Gaussian-type measures and apply it to analyze the stability of the Heisenberg Uncertainty Principle with homogeneous weights. Finally, we establish a Logarithmic Sobolev inequality for weighted Gaussian measures and utilize it to derive the Euclidean Logarithmic Sobolev inequality with homogeneous log-concave weights.
Read the full abstract

Information Theory

Zak-OTFS: A Predictable Physical Layer for Communications and Sensing

This tutorial derives the mathematical foundations of what it means for a carrier waveform to be predictable and non-selective. We focus on Zak-OTFS, where each carrier waveform is a pulse in the delay-Doppler (DD) domain, formally a quasi-periodic localized function with specific periods along delay and Doppler. Viewed in the time domain, the Zak-OTFS carrier is realized as a pulse train modulated by a tone (termed a pulsone). We start by providing physical intuition, describing what it means for the Zak-OTFS carrier waveforms to be geometric modes of the Heisenberg-Weyl (HW) group of discrete delay and Doppler shifts that define the discrete-time communication model. In fact, we show that these geometric modes are common eigenvectors of a maximal commutative subgroup of our discrete HW group. When the channel delay spread is less than the delay period, and the channel Doppler spread is less than the Doppler period, we show that the Zak-OTFS input-output (I/O) relation is predictable and non-selective. Given the I/O response at one DD point in a frame, it is possible to predict the I/O response at all other points, without recourse to some mathematical model of the channel. While it may be intuitive that geometric modes of the HW group are predictable and non-selective wireless carriers, this is not a requirement. We provide a necessary and sufficient condition that depends on the ambiguity properties of the basis of carrier waveforms. In fact, we show that the structure of a pulse train modulated by a Hadamard matrix is common to several families of waveforms proposed for 6G, including Zak-OTFS, AFDM, OTSM and ODDM.
Read the full abstract

All authors Sandesh Rao Mattu, Nishant Mehrotra, Venkatesh Khammammetti, Robert Calderbank

Numerical Analysis

A Structure-Preserving GSAV Exponential Integrator for Smectic-A Liquid Crystals

The modified Landau–de Gennes (mLdG) theory provides a powerful continuum framework for modeling smectic-A (SmA) liquid crystals by coupling the tensorial orientational order parameter $\mathbf{Q}$ with the scalar positional order parameter $u$. In this paper, we develop and analyze a structure-preserving generalized scalar auxiliary variable exponential integrator (GSAV-EI) scheme for the fully coupled mLdG system. The key ingredient is a backward Euler-type reformulation of the exponential integrator, which reveals a coercive discrete structure suitable for energy estimates and error analysis. Moreover, it eliminates the mesh-dependent time-step restriction ($\tau \lesssim h^2$) required in the existing GSAV-EI error analysis and provides a transparent connection between exponential time-differencing and implicit time-stepping methods. Building on this reconstructed structure, we prove unconditional modified-energy stability and establish optimal-order fully discrete error estimates. Numerical experiments in two and three dimensions validate the theoretical convergence rates, verify the discrete energy-dissipation law, and illustrate the self-assembly dynamics of the SmA phase.
Read the full abstract

Analysis of PDEs

Sharp trace inequalities for conformally invariant fractional powers of the sublaplacian on the Heisenberg group and the CR sphere

We establish sharp Sobolev trace inequalities for conformally invariant fractional powers of the sublaplacian on the Heisenberg group and the CR sphere, extending the corresponding Euclidean results of Einav-Loss, Beckner, and Bez-Machihara-Sugimoto to these non-Euclidean settings. In the limiting case, sharp trace Beckner-Onofri inequalities are also established on the CR sphere. The proofs are based on a duality argument due to Bez-Machihara-Sugimoto, together with the Frank-Lieb sharp form of the Hardy-Littlewood-Sobolev inequalities on the Heisenberg group and the CR sphere. The same approach also yields trace Beckner-Onofri inequalities on the standard sphere.
Read the full abstract

Combinatorics

A Weak Solution of Inverse Treblecross

We study Inverse Treblecross, the impartial avoidance version of the game Treblecross, in which completing a line of three X's is prohibited. We give a weak solution of the game: an empty board of $n$ cells is a $\mathscr{P}$-position (Sprague–Grundy value $0$) if and only if $n\equiv 0,2,3,6,9\pmod{10}$, and has value $1$ if $n\equiv 1,4,5,7,8\pmod{10}$. In particular the empty-board Sprague–Grundy sequence is the purely periodic sequence of period $10$ with values in $\{0,1\}$. The proof introduces two characteristic functions on $\mathbb Z/10\mathbb Z$ and a family of regular positions, and shows by induction that such a position splits, under a reduction move, into a disjunctive sum of regular positions whose sign product is preserved or inverted.
Read the full abstract

Friday 17 April

Dynamical Systems · Cross-listed from cs.LG

Horizon-Constrained Rashomon Sets for Chaotic Forecasting

Predictive multiplicity and chaotic dynamics represent two fundamental challenges in machine learning that have evolved independently despite their conceptual connections. We bridge this gap by introducing horizon-constrained Rashomon sets, a theoretical framework that characterizes how model multiplicity evolves with prediction horizon in chaotic systems. Unlike static prediction tasks where the Rashomon set remains fixed, chaos induces exponential divergence among initially similar models, fundamentally transforming the nature of predictive equivalence. We prove that the effective Rashomon set contracts exponentially with lead time at a rate determined by the maximum Lyapunov exponent and introduce Lyapunov-weighted metrics that provide tighter bounds on predictive disagreement. Leveraging these insights, we develop decision-aligned selection algorithms that choose among near-optimal models based on downstream utility rather than forecast accuracy alone. Extensive experiments on synthetic chaotic systems (Lorenz-96, Kuramoto-Sivashinsky) and real-world applications (wind power, traffic, weather) demonstrate that our framework improves decision quality by 18-34% while maintaining competitive predictive performance. This work establishes the first rigorous connection between chaos theory and predictive multiplicity, providing principled guidance for deploying machine learning in safety-critical chaotic domains.
Read the full abstract

All authors Gauri Kale, Rahul Vishwakarma, Holly Diamond, Ava Hedayatipour, Amin Rezaei

Combinatorics

Causal Edge Rees Algebras for Spatiotemporal Graphs

Understanding the evolution of connectivity in spatiotemporal systems requires mathematical frameworks capable of encoding not only instantaneous interactions but also their cumulative causal structure. In this work, we introduce the Causal Edge Rees Algebra (CERA), a new algebraic construction associated with causal spatiotemporal graphs. Given a temporal filtration induced by causal constraints, we associate a sequence of edge ideals whose Rees algebra encodes the full history of connectivity evolution in a single graded object. This construction establishes a bridge between dynamic graph topology and commutative algebra. In particular, we show that successive quotients of the filtration capture the emergence of new structural connections, allowing the identification of critical edges responsible for the fusion of previously disconnected components. This leads to the definition of temporal bridge modules and to a bridge detection theorem, which relates the dimension of these modules to the reduction in the number of connected components over time. Unlike existing algebraic approaches in topological data analysis, which are primarily based on geometric filtrations, the proposed framework is driven by intrinsic causal constraints. As a result, the CERA captures not only topological features but also their temporal organization and mechanisms of coalescence. The theory provides a new algebraic perspective on causal network dynamics, connecting edge ideals, Rees algebras, and temporal graphs. Beyond its theoretical significance, the framework opens new directions for the analysis of spatiotemporal systems, including epidemic networks, transport systems, and information propagation processes.
Read the full abstract

2604.25948 · 17 Apr · The authors have identified a fundamental methodological issue in the manuscript: the mapping from directed spatiotemporal causal edges to the proposed Rees-algebraic construction is not sufficiently justified. Consequently, the algebraic interpretation and the downstream claims require substantial reformulation. The authors therefore withdraw the paper

Dynamical Systems

Spectral Kernel Dynamics for Planetary Surface Graphs: Distinction Dynamics and Topological Conservation

The spectral kernel field equation R[k] = T[k] lacks a conservation-law analog. We prove (i) the fixed-point flow is strictly volume-expanding (tr DF > 0), precluding automatic conservation, and (ii) the conservation deficit per mode equals the Hessian stability margin exactly: D_m = -Delta'. Closing the deficit requires a scene-side compensating contribution, which we formalise as the distinction dynamics equation dc/dt = G[c, h_t], with MaxCal-optimal realisation G_opt. On fixed-topology 3D surface graphs we derive a conditional topology-preserving compression theorem: retaining k >= beta_0 + beta_1 modes (under a spectral-ordering assumption) preserves all Betti-number charges; we include a worked short-cycle counterexample (figure-eight) calibrating when the assumption fails. A triple necessary spectral diagnostic – Fiedler-mode concentration, elevated curl energy, anomalous beta_1 – is derived for planetary drainage networks at O(N) cost. Two internal real-data sequences serve as preliminary consistency checks; full benchmarks and adaptive-topology extensions are deferred.
Read the full abstract

General Mathematics

The Apple Pear Basket Problem: A Combinatorial Exploration

We investigate a combinatorial puzzle in which $N$ apples and $N$ pears are distributed among baskets subject to two constraints: every basket must contain the same number of apples, and every basket must contain a distinct number of pears. We prove that the maximum number of baskets is the largest divisor of $N$ not exceeding $(1 + \sqrt{1+8N})/2$. For the original puzzle with $N = 60$, this yields 10 baskets. The solution reveals a rich interplay between divisibility and combinatorics, leading to a natural classification of integers into perfect values, primes, and highly composite numbers according to their basket-packing efficiency. Computational results for $N$ up to one million confirm the asymptotic growth rate of $\sqrt{2N}$, and a complete tabulation for $N = 1$ to 100 is included.
Read the full abstract

Dynamical Systems

Puzzle Pieces, Bi-accessibility, and Connectivity of the Julia Set for Generalized Blaschke Products

We study the dynamics of a parametric family of rational functions of odd degree, where each function is a generalized Blaschke product that maps the unit circle onto itself. The action of the Blaschke product restricted to the unit circle defines a circle map, and the parameter space of the family exhibits Arnold tongues. As the parameter varies over an Arnold tongue, the action of the circle map changes from a diffeomorphism to a non-injective endomorphisms. Using a combinatorial study of puzzle pieces, we show that for adjacent parameters inside the Arnold tongues, there exist bi-accessible repelling cycles. This topological feature enables us to exclude the presence of multiply connected Fatou components whenever Herman rings are absent. As a result, we obtain a complete characterization of the connectivity of the Julia set for each member of the parametric family.
Read the full abstract

Algebraic Topology

Moment angle complexes and duality for tight manifolds

For a field $\mathbb{F}$ and a triangulated compact $\mathbb{F}$-orientable manifold, consider the homology of the associated Moment-Angle ccomplex $H_*(\mathcal{Z}_{\mathcal{K}})$. We show the total homology rank $\beta(\mathcal{Z}_{\mathcal{K}})$ satisfies the inequality $\beta(\mathcal{Z}_{\mathcal{K}};\mathbb{F})\geq 2^{m-1}(\beta(\mathcal{K};\mathbb{F})-2)+2$, with equality occurring exactly when the triangulation is $\mathbb{F}$-tight. Using Lefschetz duality, we introduce a short exact sequence of functors that, in turn, introduces a new duality theorem in Double Homology for tight manifold triangulations.
Read the full abstract

Complex Variables

On the Loewner energy of a welding homeomorphism

To any Jordan curve one may associate a circle homeomorphism $\varphi : \mathbb S^1 \to \mathbb S^1$ via conformal welding. Through this correspondence, the Loewner energy $I^L$, also known as the universal Liouville action, is a Kähler potential for the unique homogeneous Kähler metric on the universal Teichmüller space. Despite this, explicit expressions for $I^L$ in terms of $\varphi$ alone do not seem to be available in the literature. In this paper, we obtain such formulas. For this, we introduce an operator ${\bf \Lambda}_\varphi$ defined using the Fourier coefficients of the function \[ (z,w) \mapsto \log \left|\frac{\varphi(z)-\varphi(w)}{z-w}\right|, \qquad (z,w) \in \mathbb{S}^1 \times \mathbb{S}^1. \] We relate ${\bf \Lambda}_\varphi$ to the single-layer potential and composition operator, and prove an analog of the classical Grunsky inequalities for quasisymmetric $\varphi$. We show moreover that $\varphi$ is Weil–Petersson if and only if ${\bf \Lambda}_\varphi$ is Hilbert–Schmidt, and we express $I^L$ as several related Fredholm determinants as well as a regularized Fredholm determinant. We also treat Schatten classes, and we obtain formulas in terms of Dirichlet integrals involving $\log \varphi'$ and in terms of the composition operator induced by $\varphi$.
Read the full abstract

Optimization and Control

Multitask LQG Control: Performance and Generalization Bounds

We study multitask learning for stochastic and partially observed control systems, focusing on the linear quadratic Gaussian (LQG) problem. Our goal is to learn a common stabilizing controller that generalizes across a distribution of systems and objectives. To this end, we leverage a history-dependent lifting that recasts the multitask LQG problem into an equivalent high-dimensional multitask LQR problem, allowing for the analysis of policy gradient methods. We show that learning a common lifted controller induces a heterogeneity bias which we characterize via a "bisimulation function". We establish performance and generalization guarantees that explicitly depend on such bisimulation-based heterogeneity measures. For model-free, we demonstrate that multitask learning reduces policy gradient estimation variance proportionally to the number of tasks in the training set.
Read the full abstract

All authors Leonardo F. Toso, Kasra Fallah, Charis Stamouli, George J. Pappas, James Anderson

Analysis of PDEs

Spectral structure of the Benjamin-Feir instability in deep-water gravity-capillary Stokes waves

We investigate the Benjamin–Feir instability of small-amplitude gravity-capillary Stokes waves in deep water for the full water-wave equations. While modulational instability has been classically predicted by formal asymptotic approaches, such as nonlinear Schrödinger approximations, a complete spectral description at the level of the Euler equations has remained open. We perform a rigorous Bloch–Floquet spectral analysis of the linearized operator and describe the splitting of the multiple eigenvalues at the origin. In the unstable regime, we identify a pair of eigenvalues with non-zero real part forming the characteristic “figure-eight” pattern in the complex plane. As a consequence, we recover sharp near-origin Benjamin–Feir instability and stability regions in terms of the surface tension parameter, thereby providing a fully rigorous justification of the classical predictions in the gravity-capillary setting.
Read the full abstract

Dynamical Systems · Cross-listed from cs.LG

Late Fusion Neural Operators for Extrapolation Across Parameter Space in Partial Differential Equations

Developing neural operators that accurately predict the behavior of systems governed by partial differential equations (PDEs) across unseen parameter regimes is crucial for robust generalization in scientific and engineering applications. In practical applications, variations in physical parameters induce distribution shifts between training and prediction regimes, making extrapolation a central challenge. As a result, the way parameters are incorporated into neural operator models plays a key role in their ability to generalize, particularly when state and parameter representations are entangled. In this work, we introduce the Late Fusion Neural Operator, an architecture that disentangles learning state dynamics from parameter effects, improving predictive performance both within and beyond the training distribution. Our approach combines neural operators for learning latent state representations with sparse regression to incorporate parameter information in a structured manner. Across four benchmark PDEs including advection, Burgers, and both 1D and 2D reaction-diffusion equations, the proposed method consistently outperforms Fourier Neural Operator and CAPE-FNO. Late Fusion Neural Operators achieve consistently the best performance in all experiments, with an average RMSE reduction of 72.9% in-domain and 71.8% out-domain compared to the second-best method. These results demonstrate strong generalization across both in-domain and out-domain parameter regimes.
Read the full abstract

All authors Eva van Tegelen, Taniya Kapoor, George A.K. van Voorn, Peter van Heijster, Ioannis N. Athanasiadis

Mathematical Physics

Continuum honeycomb Schrödinger operators with incommensurate line defects

We study wave propagation in 2D honeycomb structures with a non-commensurate or “irrational” line defect or edge. Our model is a Schrödinger operator which interpolates, across the edge, between two distinct bulk (asymptotic) Hamiltonians with a common spectral gap about the “Dirac point” of an unperturbed honeycomb operator. We seek edge states, eigenstates that are bounded and oscillatory parallel to the edge, and decaying in the transverse direction. For non-commensurate edges, the rigorous definition of these states is nontrivial due to the lack of translation invariance along the edge. To address this, we exploit quasiperiodicity along the edge by expressing the Hamiltonian as the restriction of a 3D (degenerate elliptic) Hamiltonian describing a 3D medium with a 2D interface within which there is periodicity. Via multiscale analysis, we construct approximate edge states in this 3D setting and obtain by restriction 2D edge states which are quasiperiodic along the irrational edge. These edge states are seeded by eigenfunctions of an effective Dirac operator, which has an infinite block-diagonal structure due to the non-commensurate geometry. A consequence is that infinitely many edge state eigenpairs arise, whose energies are dense in the perturbed bulk spectral gap. In a forthcoming paper, we rigorously construct these gap-filling edge states under a Diophantine condition. The main result here is a key tool in this construction: a resolvent expansion for the 3D Hamiltonian, whose leading term is the resolvent of the block-diagonal Dirac operator. The validity of this expansion requires an omnidirectional non-resonance (no-fold) condition on the dispersion functions of the unperturbed honeycomb Hamiltonian. This condition is satisfied in the strong binding regime. In contrast with earlier works on commensurate edges, the omnidirectional condition is independent of the edge.
Read the full abstract

Dynamical Systems · Cross-listed from eess.SY

Timescale Limits of Linear-Threshold Networks

Linear-threshold networks (LTNs) capture the mesoscale behavior of interacting populations of neurons and are of particular interest to control theorists due to their dynamical richness and relative ease of analysis. The aim of this paper is to advance the study of global asymptotic stability in LTNs with asymmetric neural interactions and heterogeneous dissipation under the structural Lyapunov diagonal stability (LDS) condition. To this end, we introduce a one-parameter family of LTNs that preserves the LDS condition and has a parameter-independent equilibrium set. In the fast limit, this family converges to a projected dynamical system (PDS), while in the slow limit, it converges to a discontinuous hard-selector system (HSS). Under LDS, we prove that the fast PDS limit is globally exponentially stable and that the HSS limit is globally asymptotically stable. This alignment suggests that the limiting systems capture essential mechanisms governing stability across the entire LTN family. Together with numerical evidence, these findings indicate that resolving stability at the fast and slow endpoints provides a promising and structurally grounded path toward establishing global stability for LTNs with biologically plausible recurrence and diagonal dissipation.
Read the full abstract

All authors William Retnaraj, Simone Betteti, Alexander Davydov, Francesco Bullo, Jorge Cortes

Mathematical Physics · Cross-listed from quant-ph

Enabling Lie-Algebraic Classical Simulation beyond Free Fermions

Efficient classical simulation has matured to a critical component of the quantum computing stack, driving hardware validation, algorithm design, benchmarking, and the study of structured quantum dynamics. Lie-algebraic simulation ($\mathfrak{g}$-sim) offers a compelling approach: it represents Heisenberg-picture dynamics in the adjoint space whose dimension is set by the dynamical Lie algebra (DLA) governing the circuit, enabling efficient simulation of expectation values whenever the DLA grows only polynomially with system size. Despite this promise, existing applications of $\mathfrak{g}$-sim have been confined to free-fermionic settings. It has therefore remained unclear if the method can be applied to other structured circuit families, especially when their generators have large Pauli expansions, and hence whether Lie-algebraic simulability presents a genuinely broader paradigm than free fermions. In this work, we resolve this question by identifying additional non-trivial families of polynomial-dimensional DLAs and introducing symmetry-adapted bases that make the required adjoint-space preprocessing tractable. In particular, we develop an explicit Pauli orbit representation for permutation-equivariant dynamics, enabling efficient processing of cubic-dimensional algebras despite exponential Pauli support, and a modified generalized Gell–Mann representation for bounded Hamming-weight ($U(1)$-equivariant) dynamics, yielding polynomial simulation costs on fixed excitation sectors. Together with streamlined routines for free-fermionic algebras, these constructions significantly broaden the practical scope of $\mathfrak{g}$-sim as a unifying simulation tool for structured quantum circuits. Numerical benchmarks confirm favorable preprocessing scaling and validate large-scale proof-of-concept simulations beyond the reach of state-vector simulation.
Read the full abstract

Algebraic Geometry

Weighted blowups and 3d Poisson desingularizations

We establish existence of functorial orbifold reductions of singularities for Poisson subvarieties in smooth Poisson threefolds. Namely, we show that with enough weighted blowups, one can reduce the singularities of such Poisson subvarieties to certain simple, explicit, local normal forms: Du Val surface singularities where the Poisson structure is locally Jacobian, and plane curves lying in the vanishing locus of a particular linear Poisson structure. The proof combines Abramovich–Temkin–Włodarczyk and McQuillan's recent approach to resolution of singularities for varieties via weighted blowups with some new normal forms for three-dimensional Poisson brackets derived via Poisson cohomology. Along the way, we describe necessary and sufficient conditions for a polyvector field to lift to the weighted blowup of an orbifold along a suborbifold, generalizing criteria of Polishchuk for unweighted blowups of Poisson structures on smooth varieties.
Read the full abstract

All authors Simon Lapointe, Mykola Matviichuk, Brent Pym, Boris Zupancic

Differential Geometry

Left-invariant harmonic spinors on three-dimensional Lie groups

We study the existence of left-invariant harmonic spinors on three-dimensional Lie groups equipped with a left-invariant pseudo-Riemannian metric. An existing formula for the spin Dirac operator acting on left-invariant spinors in the Riemannian setting is revised and specialised to our cases, in particular to almost Abelian Lie algebras. Focussing on dimension two and three, we find equivalent conditions for the Lie groups to admit left-invariant harmonic spinors in terms of constraints on the structure equations of the corresponding Lie algebras. We then identify those metrics (up to automorphism) carrying left-invariant harmonic spinors in each case.
Read the full abstract

Information Theory

Maximal quadrics over finite fields and minimal codewords of projective Reed-Muller codes

We study the classification of minimal codewords of projective Reed-Muller codes of order $2$. This problem is equivalent to identifying quadrics over finite fields whose set of rational points is maximal with respect to the inclusion. We prove that except one particular case over $\mathbb{F}_2$, any two absolutely irreducible quadrics whose sets of rational points are contained within one another should be equal as projective varieties. We deduce a precise characterisation of the minimal codewords of projective Reed-Muller codes of order $2$ and further give their exact number for each possible weight.
Read the full abstract

Optimization and Control

Distance characteristics for incremental quantities

We derive distance relay characteristics in terms of incremental phasors. We use a circuit model of the network to estimate the incremental remote current. If we assume that all sources are stationary, i.e., remain periodic shortly after a fault, then the incremental remote current, and thus the characteristics, do not depend on the real-time voltages or current injections of the sources.
Read the full abstract

Statistics Theory

Horseshoe Predictive Inference

Predictive inference in the sparse Gaussian sequence model has received considerably less attention than its non-sparse, finite-sample counterpart. Existing work has largely been confined to discrete mixture priors. In this paper, we study predictive inference under a widely used continuous mixture prior, the Horseshoe. We provide new theoretical results establishing exact asymptotic minimax optimality of the predictive Bayes estimator when the sparsity level is known. Furthermore, through a Gaussian-mixture representation of the posterior predictive density (which we term Horseshoe spectroscopy), the phase-transition in the local shrinkage scale is inherited by the predictive mechanism, producing behavior similar to that of previous thresholding/switching estimators. When sparsity is unknown, we adopt a fully Bayesian approach using a hierarchical Horseshoe prior and show that it performs adaptive, as opposed to manual, switching. Under a theta-min condition, the resulting predictive risk admits an upper bound over a restricted parameter class that is sharper than the minimax rate over the full class. We demonstrate the practical value of predictive Horseshoe shrinkage on data such as images and time series that can be naturally modeled as sparse Gaussian sequences. We illustrate this approach on facial recognition across varying facial expressions and study region-wise atypical brain lateralization in autism spectrum disorder.
Read the full abstract