Optimization and Control
Adaptive Regularization within Trust Region Methods for Stochastic Nonconvex Optimization
Yunsoo Ha, Sara Shashaani and Quoc Tran-dinh
We propose a
stochastic nonconvex optimization
algorithm that achieves almost sure $\tilde{\mathcal{O}}(\epsilon^{-1.5})$ iteration complexity for problems with smooth objective functions and gradients only observable with noise. The mean-zero
stochastic noise is decision-dependent and has unbounded support with subexponential tail, allowing our framework to cover a broad class of problems. The improved almost sure iteration complexity is achieved with a new variant of the adaptive sampling trust-region optimization (ASTRO) augmented with an adaptively regularized local model, which we term Reg-ASTRO. Adaptive sampling ensures that the estimation precision is aligned with a measure of stationarity, so that iterates closer to stationarity trigger higher accuracy requirement for sampling. A key analytical challenge arises because the trust-region radius and regularization are coupled and not determined prior to gradient estimation at each iteration. We further establish an almost sure $\tilde{\mathcal{O}}(\epsilon^{-4.5})$ sample complexity for Reg-ASTRO, which improves to $\tilde{\mathcal{O}}(\epsilon^{-3.5})$ under stronger regularity conditions and use of common random numbers, substantially outperforming first-order methods in theory and numerical experiments.
Read the full abstract ▾
Number Theory
Gaussian Behavior and Geometric Gaps in Decompositions from Recurrences with Zero Coefficients
Sajad Salami
Zeckendorf's
theorem establishes a unique representation for positive integers as sums of non-consecutive Fibonacci numbers. This result has been generalized to Positive Linear Recurrence Sequences (PLRS), where key statistical properties, such as the Gaussian distribution of summands, depend on strictly positive recurrence coefficients. This paper investigates the consequences of relaxing this condition by studying Zero Linear Recurrence Relations (ZLRRs), where the leading coefficient is zero ($c_1=0$). Focusing on the Lagonacci sequence ($Z_{n+1}=Z_{n-1}+Z_{n-2}$) as a primary case study, we demonstrate that while the uniqueness of decompositions is lost, fundamental statistical behaviors persist. We prove that the number of summands in the canonical greedy decomposition converges to a Gaussian distribution and that the distribution of gaps between indices decays geometrically. Furthermore, we utilize the principle of equivalence of ensembles to show these properties are robust for a wide class of ZLRRs. Finally, we quantify the non-uniqueness of these systems, proving that the number of legal decompositions grows exponentially at a rate $\alpha =2$, significantly exceeding the growth of the underlying sequence.
Read the full abstract ▾
Number Theory
Non-standard Zeckendorf decompositions; or, Tribonacci within Fibonacci
Katie Anders, Madeline L. Dawsey and Joseph Vandehey
We study $B(n;k)$, the number of ways of writing $n$ as a sum or difference of the first $k$ Fibonacci numbers. We show that $B(0;k)$ satisfies the Tribonacci-like recurrence $B(0;k+1)=B(0;k)+B(0;k-1)+B(0;k-2)$ and that $B(n;k)$ satisfies a modified version of this recurrence.
Classical Analysis and ODEs
Uncertainty principles and singular potentials
A. Iosevich and C. Park
We establish uncertainty principles on compact Riemannian
manifolds without boundary in the setting of Laplace-Beltrami operators, including the case of real-valued singular potentials. We replace the classical homogeneity assumption by a quantitative spectral condition and obtain corresponding stability versions of uncertainty inequalities. In particular, we prove that \[ (1-\epsilon-\epsilon')^2 \leq \frac{|E|}{|M|}\cdot \# X_S \cdot \sup_{x\in E} \frac{A_S(x)}{\frac{\# X_S}{|M|}}, \] which recovers the classical bound in the homogeneous case, quantifies its deterioration in the presence of spectral inhomogeneity, and is shown to be sharp in general. In dimension one, we show that the homogeneity condition holds automatically, and we complement this
rigidity by incorporating
Fourier-ratio complexity bounds, yielding a quantitative relationship between spectral complexity and spatial support. In higher dimensions, we derive analogous results using pointwise Weyl laws and the eigenfunction restriction estimates on submanifolds.
Read the full abstract ▾
Probability · Cross-listed from stat.ML
One-Shot Generative Flows: Existence and Obstructions
Panos Tsimpos, Daniel Sharp and Youssef Marzouk
We study dynamic measure transport for generative modeling, focusing on transport maps that connect a source measure $P_0$ to a target measure $P_1$ by integrating a velocity field of the form $v_t(x) = \mathbb{E}[\dot X_t \mid X_t = x]$, where $X_\bullet = (X_t)_t$ is a
stochastic process satisfying $(X_0,X_1)\sim{P_0}\otimes{P_1}$ and $\dot X_t$ is its time derivative. We investigate when $X_\bullet$ induces a straight-line flow: a flow whose pointwise acceleration vanishes and is therefore exactly integrable by any first-order method. First, we develop multiple characterizations of straight-line flows in terms of PDEs involving the conditional statistics of the process. Then, we prove that straight-line flows under endpoint independence exhibit a sharp dichotomy. On the one hand, we construct explicit, computable straight-line processes for arbitrary Gaussian endpoints. On the other hand, we show that straight-line processes do not exist for targets with sufficiently well-separated modes. We demonstrate this obstruction through a sequence of increasingly general impossibility
theorems that uncover a fundamental relationship between the sample-path behavior of a process with independent endpoints and the space-time geometry of this process' flow map. Taken together, these results provide a structural theory of when straight-line generative flows can, and cannot, exist.
Read the full abstract ▾
Mathematical Physics · Cross-listed from hep-th
Local CFTs extremise $F$
Ludo Fraser-Taliente
Many CFTs can be extended to lines of nonlocal CFTs parametrised by the scaling dimension $\Delta$ of the fundamental field appearing in the action. $\Delta=\frac{d}{2}-\zeta$ is set by the exponent of the kinetic term $(-\partial^2)^{\zeta}$, which is nonlocal for noninteger $\zeta$. If $\Delta$ is tuned to $\Delta_\mathrm{local}$, the scaling dimension of the fundamental field in the local CFT, arXiv:1703.05325 showed that we recover the conformal data of that CFT (plus a decoupled sector). One natural question is: how is the local point special on this line of nonlocal CFTs? We prove that these local CFTs lie at the extrema of the (universal part of the) sphere free energy $\tilde{F}(\Delta)=\sin(\frac{\pi d}{2}) \log Z_{S^d}$ of the long-range CFTs: $d\tilde{F}/d\Delta|_{\Delta=\Delta_\mathrm{local}}=0$; and for
unitary CFTs they locally maximise it. The simple proof uses the fact that the derivative of $\tilde{F}$ with respect to $\Delta$ receives contributions only from the nonlocal terms in the action. The nonlocal terms must be absent in the limit $\Delta \to \Delta_\mathrm{local}$, and hence the derivative is zero. Demonstrating maximisation then requires a proof of the generalised $F$-
theorem in conformal perturbation theory to subleading order. We check our result with the O$(N)$ $\phi^4$ and cubic CFTs in the $\epsilon$ expansion and the large-$N$ limit. This result provides a minimal encoding of the scaling dimension of the fundamental field in many CFTs, and also explains the curious derivative structure of the large-$N$ expansion of $\Delta_\phi$ in the O$(N)$ model. Finally, this nonlocal $F$-extremisation can be viewed as a non-supersymmetric version of the known $c,F,a$-extremisation mechanisms.
Read the full abstract ▾
Optimization and Control · Cross-listed from cs.LG
StoSignSGD: Unbiased Structural Stochasticity Fixes SignSGD for Training Large Language Models
Dingzhi Yu, Rui Pan, Yuxing Liu et al. (+2 more)
Sign-based optimization
algorithms, such as SignSGD, have garnered
attention for their performance in distributed learning and training large foundation models. Despite their empirical superiority, SignSGD is known to diverge on non-smooth objectives, which are ubiquitous due to ReLUs, max-pools, and mixture-of-experts. To overcome this limitation, we propose StoSignSGD, an
algorithm that injects structural stochasticity into the sign operator while maintaining an unbiased update step. In the regime of (online) convex optimization, StoSignSGD rigorously resolves the non-convergence issues of SignSGD, achieving a sharp convergence rate matching the lower bound. For the more challenging non-convex non-smooth optimization, we introduce generalized stationary measures that encompass prior definitions, proving that StoSignSGD improves upon the best-known complexity bounds by dimensional factors. Empirically, StoSignSGD is stable and efficient across diverse large
language model (LLM) training regimes. In aggressive low-precision pretraining, which spans both FP8 and the far more demanding FP4 regime where AdamW fails catastrophically, StoSignSGD stays stable and consistently performs the best. It attains a 1.44x to 2.14x speedup over established baselines under FP8. Under 4-bit precision, it improves downstream accuracy on the largest OLMo2-370M model by 1.13 points over the strongest stable baseline, and this advantage grows as both the model size and the data scale up. When fine-tuning 7B LLMs on mathematical
reasoning tasks, StoSignSGD also delivers clear gains over both AdamW and SignSGD. Finally, to explain why it works, we develop a sign conversion framework that turns any general optimizer into its unbiased, sign-based counterpart. Using this framework, we decompose the core components of StoSignSGD and run a comprehensive ablation study to validate our design choices.
Read the full abstract ▾
All authors Dingzhi Yu, Rui Pan, Yuxing Liu, Difan Zou, Tong Zhang
Analysis of PDEs
Estimates to the weak solution of the electro-hydrodynamical boundary value problem for the unit cell of cation-exchange membrane
Yulia Koroleva
We study a model problem on the filtration of a conducting fluid through a porous layer. A porous medium is presented as an assemblage of identical spherical
cells. Each
cell consists of a porous core and
liquid shell. We show the dependence of each flow parameter on the Debye radius which characterizes how far the influence of a charge extends in electrolyte. The common case of finite Debye radius in comparison to the
cell radius is analyzed. We derive apriori estimates for flow characteristics which show the specific behavior of the fluid. The boundedness of velocity field, pressure, electric potential and ion flux densities was proved.
Read the full abstract ▾
Information Theory
Five Constructions of Asymptotically Optimal Aperiodic Doppler Resilient Complementary Sequence Sets with New Parameters
Xuanyu Liu, Pinhui Ke and Zuling Chang
Sequences exhibiting favorable ambiguity function characteristics play a critical role in radar detection systems and modern mobile communication applications. As a newly developed sequence family, Doppler resilient complementary sequence sets (DRCSSs) can effectively suppress ambiguity function sidelobes by coherently combining the ambiguity functions of their constituent subsequences. The objective of this paper is to present five classes of asymptotically optimal aperiodic DRCSSs with novel parameters based on trace functions over finite fields and column orthogonal complex
matrices. Compared with existing asymptotically optimal aperiodic DRCSSs in the literature, the proposed aperiodic DRCSSs deliver superior or novel parameters. Notably, for three families of the constructed aperiodic DRCSSs, the column sequence peak-to-average power ratio (PAPR) is upper bounded by p by selecting suitable column orthogonal complex
matrices.
Read the full abstract ▾
Numerical Analysis · Cross-listed from cs.LG
Python library supporting Discrete Variational Formulations and training solutions with Collocation-based Robust Variational Physics Informed Neural Networks (DVF-CRVPINN)
Tomasz Służalec, Marcin Łoś, Askold Vilkha et al. (+1 more)
We explore the possibility of solving Partial Differential Equations (PDEs) using discrete weak formulations. We propose a programming environment for defining a discrete computational domain, introducing discrete functions defined over a set of points, constructing discrete inner products, and introducing discrete weak formulations employing Kronecker delta test functions. Building on this setup, we propose a discrete neural network representation, training the solution function defined over a discrete set of points and employing discrete finite difference derivatives in the automatic differentiation procedures. As a challenging computational model example, we focus on Stokes equations in two-dimensions, defined over a discrete set of points. We train the solution using the discrete weak residual and the Adamax
algorithm with discrete automatic differentiation of the discrete gradients. Despite introducing the python environment, we also provide a rigorous mathematical formulation based on discrete weak formulations, proving the well-posedness and robustness of the loss function. The solution of the discrete weak formulations is based on neural network training employing a robust loss function that is related to the true error. In this way, we have a robust control of the numerical error during the training of the neural networks. Besides the Stokes formulation, we also explain the functionality of the proposed library using the Laplace problem formulation.
Read the full abstract ▾
All authors Tomasz Służalec, Marcin Łoś, Askold Vilkha, Maciej Paszyński
General Mathematics
A new equivalence to the Riemann Hypothesis by means of the Salem integral equation
Benito J. González and Emilio R. Negrín
This note presents a new equivalence to the Riemann Hypothesis by means of the Salem integral equation.
Information Theory
Robust Transmission Design for RIS-Assisted High-Speed Train Communication Coverage Enhancement With Imperfect Cascaded Channels
Changzhu Liu, Ruisi He, Haoxiang Zhang et al. (+4 more)
Reconfigurable intelligent surface (RIS) has recently been gained
attention as an effective technique improving the coverage and performance of communication systems by creating additional communication links. Deployment of RIS is crucial for overcoming signal coverage limitations, especially in high-speed train (HST) scenarios. Considerable research has been performed assuming perfect channel state information (CSI). However, due to the rapidly time-varying fading channels and feedback delays, achieving perfect CSI at the base station (BS) is not feasible in the HST scenarios. To tackle this problem, this paper investigates a robust design strategy for RIS-aided HST communication coverage enhancement, particularly focusing on cascaded BS-RIS-user channels at BS (CBRUB). The study explores the optimization problem under two types distinct of models: centered on minimizing transmit power subject to worst-case rate constraints within the bounded CSI error (BCSIE) model, and the other focusing on outage probability (OP) constraints under the statistical CSI error (SCSIE) model. We use the S-procedure to approximate the non-convex (NC) constraints, converting the worst-case rate constraints into linear
matrix inequalities. Additionally, the Bernstein-type inequality is applied to transform the OP constraints into second-order cone constraints and linear inequalities. The simulation analysis results show that CBRUB errors have a significant effect on system performance compared to direct CSI errors.
Read the full abstract ▾
All authors Changzhu Liu, Ruisi He, Haoxiang Zhang, Jiahui Han, Ruifeng Chen, Bo Ai, Zhangdui Zhong
Group Theory · Cross-listed from cs.FL
On Word Representations and Embeddings in Complex Matrices
Paul C. Bell, George Kenison, Reino Niskanen et al. (+2 more)
Embeddings of word structures into
matrix semigroups provide a natural bridge between combinatorics on words and linear
algebra. However, low-dimensional
matrix semigroups impose strong structural restrictions on possible
embeddings. Certain finitely generated groups admit faithful representations in SL(2, C) and other similar
matrix groups. On the other hand, it is known that the product of two free
semigroups on two generators cannot be embedded into the 2x2 complex
matrices. In this paper we study
embeddings of word structures into low-dimensional
matrix semigroups over the complex numbers and develop new techniques for constructing word representations of the Euclidean Bianchi groups. These representations provide a symbolic framework and a natural first step towards analysing fundamental decision problems in 2x2
matrix semigroups.
Read the full abstract ▾
All authors Paul C. Bell, George Kenison, Reino Niskanen, Igor Potapov, Pavel Semukhin
Information Theory · Cross-listed from quant-ph
Heuristic Search for Minimum-Distance Upper-Bound Witnesses in Quantum APM-LDPC Codes
Kenta Kasai
This paper investigates certified upper bounds on the minimum distance of an explicit family of Calderbank-Shor-Steane
quantum LDPC codes constructed from affine permutation
matrices. All codes considered here have active Tanner graphs of girth eight. Rather than attempting to prove a general lower bound for the full code distance, we focus on constructing low-weight non-stabilizer logical representatives, which yield valid upper bounds once they are verified to lie in the opposite parity-check kernel and outside the stabilizer row space. We develop a unified framework for such witnesses arising from latent row relations, restricted-lift subspaces including block-compressed, selected-fiber, and CRT-stripe constructions, cycle- 8 elementary trapping-set structures, and decoder-failure residuals. In every case, search is used only to generate candidates; the reported bounds begin only after explicit kernel and row-space exclusion tests have been passed. For the latent part, we also identify a block-compression criterion under which the certification becomes exact. Applying these methods to representative APM-LDPC codes sharpens previously reported upper bounds and provides concrete certified values across the explored parameter range.
Read the full abstract ▾
Combinatorics
Counterexamples to a dimension-free variant of Erdős's diameter conjecture for separated distances
Boon Suan Ho
Erdős asked about the least possible diameter of an $n$-point set in Euclidean space when any two pairwise distances differ by at least $1$. We consider the uniform-in-dimension variant of his question, in which the dimension $d$ may grow with $n$ and the error term is independent of $d$, and answer it in the negative by an explicit construction. For every
prime power $q$, we construct a set $\mathcal X_n$ of $n=q+1$ points in $\mathbb R^{q^2+q}=\mathbb R^{n^2-n}$ such that the $\binom{n}{2}$ pairwise distances are mutually at least $1$ apart, yet $$\operatorname{diam}(\mathcal X_n)\le \Bigl(1-\frac{1}{\pi^2}+o(1)\Bigr)n^2.$$ The construction combines a Singer cyclic difference set with a weighted product of $N=\binom n2$ regular $m$-gons, where $m=q^2+q+1$. The weights are chosen so that the resulting distances increase while the gaps between consecutive distances decrease. After normalizing the smallest gap to $1$, all distinct pairwise distances are separated by at least $1$, while the diameter has asymptotic constant $1-1/\pi^2<1$.
Read the full abstract ▾
Group Theory
Diameter bounds for arbitrary finite groups and applications
Sean Eberhard, Elena Maini, Luca Sabatini et al. (+1 more)
We prove a strong general-purpose bound for the diameter of a finite group depending only on the diameters of its composition factors and the maximal exponent of a normal abelian section. There are a number of notable applications: (1) if $G$ is a finite soluble group of exponent $e$, $\mathrm{diam}(G) \ll e (\log |G|)^8$, (2) anabelian groups with bounded-rank composition factors have polylogarithmic diameter, (3) transitive soluble subgroups of $S_n$ have diameter $\ll n^5$, and (4) Grigorchuk's gap
conjecture holds for any finitely generated group acting faithfully on a bounded-degree rooted tree. Additionally, conditional on Babai's
conjecture, (5) any transitive permutation group of degree $n$ has diameter bounded by a
polynomial in $n$ (a folkloric
conjecture), and (6) Grigorchuk's gap
conjecture holds for residually finite groups, and thus the
conjecture reduces to the simple case.
Read the full abstract ▾
All authors Sean Eberhard, Elena Maini, Luca Sabatini, Gareth Tracey
Mathematical Physics · Cross-listed from quant-ph
Ensembles of random quantum states tunable from volume law to area law
Héloïse Albot and Sebastian Paeckel
A standard approach to generate random pure
quantum states relies on sampling from the Haar measure. However, the
entanglement properties of such states present a fundamental challenge for their general applicability. Here, we introduce the $\sigma$-ensembles $\unicode{x2013}$ a family of random
quantum states with only a single control parameter. Crucially, these states are designed such that they can be tuned between volume-law and area-law behavior, which has been a major obstacle thus far. We construct representatives of this ensemble by imposing a probability distribution on the
eigenvalues of the successive subsystems, and subsequently reconstructing a compatible global state using the
matrix product state (MPS) formalism. Due to their area-law
entanglement, our approach circumvents the intractability of Haar-random pure states in classical simulations of
quantum systems and is more representative of typical
Hamiltonian ground states.
Read the full abstract ▾
Information Theory
Reed–Muller Codes Achieve the Symmetric Capacity on Finite-State Channels
Henry D. Pfister, Navin Kashyap, Jean-Francois Chamberland et al. (+1 more)
We study reliable communication over finite-state channels (FSCs) using Reed–Muller (RM) codes. Building on recent
symmetry-based analyses for memoryless channels, we show that a sequence of binary RM codes (with some random scrambling) can achieve the symmetric capacity (or uniform-input information rate) of a binary-input indecomposable FSC. Our approach has three components. First, we establish a capacity-via-
symmetry theorem for doubly-transitive group codes on discrete memoryless channels (DMCs) with non-binary inputs, under some
symmetry and puncturing conditions. Then, we reduce a binary-input FSC to an almost memoryless non-binary channel by grouping adjacent input bits into blocks and interleaving non-binary codes onto the channel. Finally, we show that the interleaved non-binary codes can be constructed from a single binary RM code.
Read the full abstract ▾
All authors Henry D. Pfister, Navin Kashyap, Jean-Francois Chamberland, Galen Reeves
Dynamical Systems
Borel–Bernstein and Hirst-type Theorems for Nearest-Integer Complex Continued Fractions over Euclidean Imaginary Quadratic Fields
Kangrae Park
For each $d \in {1,2,3,7,11}$, let $T_d$ be the nearest-integer complex continued fraction map associated with the Euclidean ring $\mathcal{O}*d$, and let $(a_n)$ be its digit sequence. We prove two metric results for this five-system family. First, for every sequence $(u_n)*{n\ge 1}$ with $u_n \ge 1$, the set of points for which $|a_n| \ge u_n$ for infinitely many $n$ has full or zero normalized Lebesgue measure according as $\sum_{n=1}^\infty u_n^{-2}$ diverges or converges. This gives a unified Borel–Bernstein
theorem, extending the Hurwitz case $d=1$ to all five Euclidean imaginary quadratic fields. Second, for any infinite set $S \subset \mathcal{O}_d$, if $\tau(S)$ denotes its convergence exponent, then the digit-restricted set $F_d(S)={z:\ a_n(z)\in S\ \text{for all } n,\ |a_n(z)|\to\infty}$ satisfies $\dim_H F_d(S)=\tau(S)/2$. More generally, for any cutoff function $f(n)\to\infty$, the set $F_d(S,f)={z\in F_d(S):\ |a_n(z)|\le f(n)\ \text{for all } n}$ is either empty or has the same Hausdorff dimension $\tau(S)/2$. The proof combines quantitative ergodic properties of the nearest-integer systems with a large-digit conformal iterated function subsystem that is $2$-decaying. We also obtain applications to sparse patterns, shrinking targets, and almost-sure $L'evy$- and Khinchine-type laws.
Read the full abstract ▾
Statistics Theory
Generalization of Pearl's Front-Door Criterion
Carol Wu and Elina Robeva
Pearl's front-door criterion provides a set of sufficient conditions for estimating the total causal effect from observational data in the presence of latent confounding, using the functional P(y | do(x := x*)) = \sum_z P(z | x*) \sum_x P(y | x, z) P(x). An open question is whether these conditions can be generalized to be both necessary and sufficient for the validity of this functional, similar to the generalization achieved for the back-door adjustment criterion by Shpitser. In this paper, we present a new, weakened set of graph-based conditions sufficient for the front-door formula to estimate the total causal effect, expanding the scope of problems amenable to front-door identification.
Read the full abstract ▾
Rings and Algebras
Matrices over Finite Fields of Characteristic 2 as Sums of Diagonalizable and Square-Zero Matrices
Peter Danchev, Esther García and Miguel Gómez Lozano
We investigate the problem asking when any square
matrix whose entries lie in a finite field of characteristic 2 is decomposable into the sum of a diagonalizable
matrix and a nilpotent
matrix with index of nilpotency at most 2 and, as a result, we completely resolve this question in the affirmative for any finite field of characteristic 2 having strictly more than three elements. Our main
theorem of that type, combined with results from our recent publication in Linear
Algebra & Appl. (2026) (see [7]), totally settle this problem for all finite fields different from $\mathbb{F}_2$ and $\mathbb{F}_3$. However, in this paper we also prove that each
matrix over $\mathbb{F}_2$ is expressible as the sum of a potent
matrix with index of potency not exceeding 4 and a nilpotent
matrix with index of nilpotency not exceeding 2, thus substantiating recent examples due to Šter in Linear
Algebra & Appl. (2018) and Shitov in Indag. Math. (2019) (see, respectively, [9] and [8]).
Read the full abstract ▾
Statistics Theory · Cross-listed from stat.ML
Structural interpretability in SVMs with truncated orthogonal polynomial kernels
Víctor Soto-Larrosa, Nuria Torrado and Edmundo J. Huertas
We study post-training interpretability for Support Vector Machines (SVMs) built from truncated orthogonal
polynomial kernels. Since the associated reproducing kernel
Hilbert space is finite-dimensional and admits an explicit tensor-product orthonormal basis, the fitted decision function can be expanded exactly in intrinsic RKHS coordinates. This leads to Orthogonal Representation Contribution Analysis (ORCA), a diagnostic framework based on normalized Orthogonal Kernel Contribution (OKC) indices. These indices quantify how the squared RKHS norm of the classifier is distributed across interaction orders, total
polynomial degrees, marginal coordinate effects, and pairwise contributions. The methodology is fully post-training and requires neither surrogate models nor retraining. We illustrate its diagnostic value on a synthetic double-spiral problem and on a real five-dimensional echocardiogram
dataset. The results show that the proposed indices reveal structural aspects of model complexity that are not captured by predictive accuracy alone.
Read the full abstract ▾
Information Theory
Bandwidth Cost of Locally Repairable Convertible Codes in the Global Merge Regime
Saransh Chopra, Shubhransh Singhvi and K.V. Rashmi
Recent studies have shown that distributed storage systems can achieve significant space savings by adapting redundancy levels to varying disk failure rates. This adaptation is performed via code conversion, wherein data encoded under an initial code are transformed to data encoded under a final code. While this process is typically resource-intensive, convertible codes are designed to enable these transformations efficiently while preserving desirable decodability constraints such as repair degree, or the number of nodes accessed during node repair. In this work, we focus on the bandwidth cost of conversion, or the total amount of data transferred during the conversion process. We study fundamental limits on the bandwidth cost of conversion between systematic optimal-distance Locally Repairable Codes (LRCs). We restrict our focus to the global merge regime, in which multiple initial codewords are combined to form a single final codeword while preserving information locality. We focus on stable convertible codes, wherein the number of unchanged nodes is maximized during conversion. We generalize an information-theoretic approach for modeling code conversion to the LRC setting, and derive the first non-trivial lower bounds on the bandwidth cost of conversion in this regime. Notably, our bounds do not rely on any linearity assumptions. Consequently, we show that the constructions of Maturana and Rashmi are bandwidth-optimal across a broad range of parameters in the global merge regime.
Read the full abstract ▾
Statistics Theory · Cross-listed from quant-ph
Cloning is as Hard as Learning for Stabilizer States
Nikhil Bansal, Matthias C. Caro and Gaurav Mahajan
The impossibility of simultaneously cloning non-orthogonal states lies at the foundations of
quantum theory. Even when allowing for approximation errors, cloning an arbitrary unknown pure state requires as many initial copies as needed to fully learn the state. Rather than arbitrary unknown states, modern
quantum learning theory often considers structured classes of states and exploits such structure to develop learning
algorithms that outperform general-state
tomography. This raises the question: How do the sample complexities of learning and cloning relate for such structured classes. We answer this question for an important class of states. Namely, for $n$-
qubit stabilizer states, we show that the optimal sample complexity of cloning is $\Theta(n)$. Thus, also for this structured class of states, cloning is as hard as learning. To prove this result, we relate stabilizer state cloning to a variant of the sample amplification problem for probability distributions that was recently introduced in classical learning theory. This allows us to obtain our cloning lower bounds in analogy to new sample amplification lower bounds for classes of distributions with an underlying linear structure. Our results provide a more fine-grained perspective on No-Cloning
theorems, opening up connections from foundations to
quantum learning theory and
quantum cryptography.
Read the full abstract ▾
Algebraic Topology
Motif-based filtrations for persistent homology: A framework for graph isomorphism and property prediction
Meritxell Vila-Miñana, Robert Jankowski, Aina Ferrà Marcús et al. (+3 more)
Determining whether two graphs are isomorphic is a fundamental problem with practical applications in areas such as molecular chemistry or social network analysis, yet it remains a challenging task, with exact solutions often being computationally expensive. We address this task using persistent homology built on motif-based filtrations of graphs, a method from topological data analysis that summarizes the shape of data by tracking the persistence of structural features along filtrations. Specifically, we use edge-weighting schemes based on the densities of triangles, chordless squares, and chordless pentagons, which have been shown to be effective for detecting network dimensionality. Our cycle-density filtrations distinguish non-isomorphic graphs perfectly or nearly perfectly across four demanding graph families, many of which exhibit
symmetries. We outperform
curvature-based, degree-based, and Vietoris–Rips filtrations, and match or exceed the accuracy of egonet-distance methods while incurring a lower computational cost. The expressive power of our filtrations goes beyond isomorphism testing: because they capture rich structural information from graphs, they consistently achieve top performance on property prediction tasks using real-world data, and exhibit high sensitivity to edge rewiring and removal. Together, these findings establish cycle-density filtrations as an effective and computationally tractable framework for graph comparison and characterization, bridging topological data analysis and network science.
Read the full abstract ▾
All authors Meritxell Vila-Miñana, Robert Jankowski, Aina Ferrà Marcús, Rubén Ballester, M. Ángeles Serrano, Carles Casacuberta
Mathematical Physics · Cross-listed from quant-ph
Computing the free energy of quantum Coulomb gases and molecules via quantum Gibbs sampling
Simon Becker, Cambyse Rouzé and Robert Salzmann
We develop a
quantum Gibbs-sampling approach to free-energy estimation for trapped systems of distinguishable particles with Coulomb interactions in dimensions $d\in\{2,3\}$. The singular infinite-dimensional problem is reduced to a finite-rank low-energy truncation of the interaction, the truncated Gibbs state is prepared by a
quantum Markov
semigroup, and thermodynamic integration gives the free energy. We obtain explicit truncation rates, quantitative spectral-gap estimates, and a finite-dimensional
circuit implementation. In particular, every fixed interaction truncation has a positive gap, while weak truncated interactions admit a particle-number-independent lower bound. The
circuit complexity is conditional on a certified gap lower bound at the accuracy-dependent interaction cutoff.
Read the full abstract ▾
Dynamical Systems
Detecting Regime Transitions in Dynamical Systems via the Mixup Euler Characteristic Profile
Sushovan Majhi, Atish Mitra, Santanu Nandi et al. (+2 more)
We develop a framework for detecting regime transitions in dynamical systems using the Mixup Euler Characteristic Profile (Mixup ECP) – the Euler characteristic of the
geometric intersection of ball unions around adjacent delay-embedded trajectory segments, viewed as a function of filtration scale. The Mixup ECP provides a detection statistic with a built-in null and guaranteed stability. We formalize regime detection as a low-side-permutation test, establish its validity and consistency, and introduce a multi-delay extension that automatically selects the most informative dynamical timescale. Complementing the topological signal with Complexity Variance, Higuchi fractal dimension, and a rolling mean baseline, the four-signal combined method achieves $9.50$ days MAE on Indian monsoon onset (Nepal target) – a $32\%$ improvement over the rolling mean baseline and $9\%$ over CUSUM. Validated on the Lorenz system, logistic map, and three monsoon systems spanning both hemispheres (Indian/Nepal, Indian/Kerala, Western North Pacific), plus ENSO and a synthetic
EEG dataset, the framework adds value precisely when the transition is gradual or obscured by noise.
Read the full abstract ▾
All authors Sushovan Majhi, Atish Mitra, Santanu Nandi, Md Nurujjaman, Buddha Nath Sharma
Functional Analysis
Enrique García-Sánchez
These notes are a detailed, self-contained introductory course on convexity and concavity in
Banach lattices, suitable for both experts and beginners. We revisit, from a modern perspective, the classical notions of $(p,q)$-convexity, $(p,q)$-concavity and upper and lower $p$-estimates, and the main relations between these properties, integrating more recent developments in the area. We explain in full detail the $p$-convexification and $p$-concavification techniques. We also provide a comprehensive exposition of the main factorization results for $(p,q)$-convex and $(p,q)$-concave operators, including well-known results from Krivine, Maurey–Nikishin, Pietsch and Pisier, and their applications to the renorming and the representation of convex and concave
Banach lattices.
Read the full abstract ▾
Combinatorics
A matroidal twist on a formula of Brion
Matthias Beck, Caroline Klivans and Dustin Ross
Brion's Formula realizes the Laurent
polynomial of
lattice points in a
lattice polytope P as the sum of rational functions associated to the vertices of P. In this paper, we consider the special case where P is a generalized permutohedron. We study a modification of the rational functions associated to the vertices of P depending on a given matroid M. Upon summing these rational functions, we describe how the resulting Laurent
polynomial Q_M(P) behaves in certain ways like the
lattice points of P, exhibiting natural recursive and reciprocity behaviors. Furthermore, upon evaluating Q_M(P) at 1, we recover the matroid Euler characteristic of Larson, Li, Payne, and Proudfoot, so the combinatorial approach in this paper gives new insight into studying these quantities.
Read the full abstract ▾
Optimization and Control · Cross-listed from eess.SY
Tube-Based Robust Data-Driven Predictive Control
Chi Wang and David Angeli
This paper presents a tractable tube-based robust data-driven predictive control scheme that uses only a single finite noisy input-state trajectory of an unknown discrete-time linear time-invariant (LTI) system. A simplex constraint is imposed on the Hankel coefficient vector, yielding explicit polyhedral bounds on the prediction mismatch induced by bounded measurement noise. Using certified initial and terminal robust positively invariant (RPI) sets, we derive a tube-tightened formulation whose online optimization problem is a strictly convex quadratic program (QP). The resulting controller guarantees recursive feasibility, robust satisfaction of input and state constraints, and practical input-to-state stability of the closed loop with respect to measurement noise. Numerical examples illustrate the effectiveness, robustness, and closed-loop performance of the proposed method.
Read the full abstract ▾