Papers

Pairwise dependence measures such as correlation and causality are fundamental to temporal data mining, yet there is still no principled and robust way to quantify dependence between heterogeneous data types, especially between continuous time series and discrete temporal event sequences. Existing approaches rely on ad hoc transformations or mutual-information estimators that are highly sensitive to quantization, repeated values, and event redundancy, leading to biased or unstable results in practice. We propose a nonparametric mutual information estimator that directly measures the dependence between time series and event sequences without data transformation, learning, or ad hoc discretization. Our method models the continuous-discrete duality of real-world time series to handle quantization and repeated-value artifacts and introduces a latent event clustering strategy to mitigate bias from event co-occurrence and redundancy. Together, these yield a robust and unified framework that bridges discrete and continuous mutual information. We evaluate the proposed estimator on four representative tasks: discrete-continuous time-delayed mutual information for causality analysis, global and local temporal repetition discovery, discrete covariate selection for time series forecasting, and continuous feature selection for classification. Experiments on synthetic and real-world datasets show consistent improvements over existing methods in accuracy, robustness, and interpretability, positioning our approach as a general-purpose dependence operator for heterogeneous temporal data, similar to Pearson correlation for homogeneous time series. Code available at: https://github.com/HaojiHu/Multimodal-Temporal-Data-Quantification

Short-horizon accuracy does not ensure that a learned chaotic system has correct long-time dynamics. Trajectory (zeroth-order) matching constrains vector-field values, and Jacobian (first-order) matching constrains local tangent dynamics, but neither determines how the Jacobian varies away from supervised states, so a model can be locally accurate while drifting toward spurious attractors and distorting long-time statistics. We show that second-order supervision mitigates these failures. Because forming full Hessian tensors is computationally prohibitive in high dimensions, we propose model-constrained randomized Jacobian matching, which compares the Jacobians of the true and learned vector fields at randomly perturbed inputs. A Taylor expansion shows that the expected randomized Jacobian loss decomposes into the Jacobian mismatch plus a Hessian mismatch scaled by the noise variance, implicitly enforcing second-order consistency at $O(d^2)$ memory cost without forming the $O(d^3)$ Hessian tensor. In Lorenz 63 with minimal temporal supervision, second-order supervision reduces invariant-measure error and Lyapunov-spectrum MSE, and recovers the constant Hessian norm of the true bilinear field. Across five training seeds, explicit Hessian matching produces catastrophic Lyapunov outliers for four of five seeds, whereas randomized Jacobian matching produces none among 5,000 on-attractor rollouts, and it attains the largest threshold in a directional capture scan. In coupled Lorenz 96, first-order methods enter spurious high-amplitude regimes as forcing increases, while second-order methods retain accurate marginals. Randomized Jacobian matching costs about the same as explicit Hessian matching on Lorenz 63 and 40% less on Lorenz 96, with no reference Hessian evaluations during training.

Strata of holomorphic $1$-forms have an absolute period foliation given by varying $1$-forms while keeping their integrals along closed loops fixed. In this paper, we classify the leaf closures of this foliation when the stratum is connected, outside of a subvariety of high codimension.

For a given subset $S\subseteq V(G)$ of a graph $G$, the pair $(G,S)$ is knitted if for every partition of $S$ into non-empty subsets $S_1, S_2, \ldots, S_t$, there exist pairwise disjoint connected subgraphs $C_1, C_2, \ldots, C_t$ in $G$ such that $S_i\subseteq V(C_i)$ for all $1 \le i \le t$. A graph $G$ is \emph{$\ell$-knitted} if $(G,S)$ is knitted for every subset $S\subseteq V(G)$ of size $\ell$. In this paper, we prove that every $8\ell$-connected graph is $\ell$-knitted. We subsequently apply this result to Hadwiger's Conjecture, which states that every $k$-chromatic graph contains a $K_k$-minor. Specifically, we demonstrate that the vertex connectivity of any minimal counterexample to Hadwiger's Conjecture is at least $\lceil k/8 \rceil$, improving upon the previous lower bound of $\lceil 2k/27 \rceil$ established by Kawarabayashi (2007). Our proof corrects a gap in the argument of Kawarabayashi-Yu~(2013) and establishes the claim stated without proof in Liu–Rolek–Yu~(2019).

This paper establishes gradient estimates and Liouville-type theorems for the \(\Phi\)-Laplacian equation \(\Delta_{\Phi}(u) = G(|\nabla u|^2)\) on complete Riemannian manifolds and its parabolic counterpart \(\partial_t u = \Lambda_{\Phi}(u)\) on compact Riemannian manifolds. Using a nonlinear \(\Phi\)-Bochner formula and the Nash-Moser iteration technique, we prove local gradient bounds under the lower bound assumption of Ricci curvature and suitable conditions on \(\Phi\) and \(G\), which leads to Liouville theorems for global solutions. For the parabolic case, we employ the maximum principle to derive gradient estimates on compact Riemannian manifolds, and subsequently obtain Liouville-type results. Our work provides a unified framework that generalizes prior results for \(p\)-harmonic functions and other quasilinear equations.

Let $X$ be a geometrically irreducible smooth projective curve over a field $k$. Yamazaki et al. define a biadditive symmetric pairing $\langle -,-\rangle$ on the torsion subgroup of the Picard group $\mathrm{Pic}(X)$ with values in $k^\times \otimes \mathbb{Q}/\mathbb{Z}$. The intrinsic subgroup $\mathrm{Pic}(X)_\mathrm{tors}^\mathrm{is}$ is the kernel of this pairing. When $X$ is an elliptic curve $E$, we can identify $E \simeq \mathrm{Pic}^0(E)$. We classify $E(k)_\mathrm{tors}^\mathrm{is}$ in purely algebraic terms for many elliptic curves over an arbitrary field $k$. We give a generalization of the analytic methods of Yamazaki et al. from $\mathbb{Q}$ to an arbitrary field $k \subset \mathbb{C}$. Lastly, for $k=\mathbb{Q}$, we describe an algorithm to explicitly compute $E(\mathbb{Q})_\mathrm{tors}^\mathrm{is}$.

We are interested in the asymptotic behavior of solutions toward strong rarefaction waves for a parabolic-hyperbolic system arising from chemotaxis. Suppose that the Riemann problem to the corresponding inviscid system admits rarefaction waves. We show that if the initial data is a small perturbation of an approximate rarefaction wave, then the Cauchy problem has a unique global solution that converges to the rarefaction wave asymptotically in time. The stability results hold regardless of the wave strengths. The waves can be either a single rarefaction wave or a composite of two rarefaction waves. In the composite case, we separate the initial centers of the two approximate profiles to make their interaction error term small while retaining convergence to the prescribed Riemann wave. The proofs are based on the energy method, where the key observations are the monotonicity of the approximate rarefaction waves with respect to both space and time, and the appropriate replacement of spatial derivative of the hyperbolic component with the parabolic component in proper forms.

This short note proposes a polynomial-time algorithm for near-optimal Euclidean estimation of a signal constrained to lie in the unit ball of a symmetric norm, where the symmetry is with respect to a known basis and the norm is accessible through an evaluation oracle. We further extend the method to a random-design, moderate-dimensional linear regression setting, where the regression parameter is likewise assumed to belong to a constraint set defined by a symmetric norm.

In this paper, we compute the Minimal Intersection Radius (MIR) of growing, non-homogeneous ellipsoids in arbitrary ambient dimension. We provide a geometric method to find the MIR using techniques from convex optimization, a secondary method using second-order cone programs, and show that the MIR can be phrased as an LP-type problem, where the computation from convex optimization acts as a certificate. We implement these methods and benchmark them using different convex solvers, and with or without the LP setting. We also provide a comparison with similar but different problems that appeared in the literature, and show that finding the MIR is, in general, not equivalent to finding the minimal enclosing ellipsoid.

In this paper, we establish, for the first time in the literature, the convergence of the practical versions of the Inexact Proximal Point Algorithm (IPPA) and the Inexact Tseng Algorithm (ITA) for computing approximate solutions to monotone inclusions in Hilbert spaces under the the presence of nonsummable errors. Our ap- proach relies on Tikhonov regularization, the contraction property of the associated monotone operators, and the recently developed R-continuity theory. The proposed techniques and results can be extended to analyze various important inexact algorithms in optimization-related problems with nonsummable errors.

For a properly embedded Willmore surface $\Sigma$ in $\mathbb R^3$, we prove that if the scale-invariant second fundamental form is sufficiently small near infinity, the surface has finitely many ends. Moreover, if this scale-invariant quantity vanishes at infinity, or if there is only one end, the total $L^2$-norm of the second fundamental form is finite.

Dittert's conjecture gives a sharp upper bound for the Dittert functional on nonnegative matrices whose entries sum to \(n\). It extends the van der Waerden permanent problem from the doubly stochastic polytope to a larger simplex in which row and column sums are allowed to vary. We prove the conjecture for every dimension \(n\ge 17\). The proof combines the Knopp–Sinkhorn lower bound for boundary points of the doubly stochastic polytope with a refined scaling step in the Cheon–Wanless method. The main improvement is a sharper subset-sum estimate for the row and column sums of a near maximizer, which reduces the scalar dilation needed to obtain a doubly superstochastic matrix. This strengthened comparison is sufficient to exclude boundary maximizers in all dimensions \(n\ge 17\), and the known positive-support characterization then identifies the unique maximizer as \(n^{-1}J_n\).

We consider approximation of a Gaussian distribution with a mixture of homoscedastic Gaussians of smaller variance. The solution is obtained by minimising the $L^2$ norm between the original Gaussian and the mixture, which is parameterised to reduce the complexity of the optimisation problem. The developed technique is straightforward, sufficiently robust and yields Gaussian Mixtures that rapidly approach the original function as the number of mixands is increased. The proposed solution is examined for multiple special cases of input parameters resulting in further simplifications. Extension of the proposed method for approximating non-Gaussian distributions is discussed.

In this paper, we develop the Rubio de Francia extrapolation theorem for the multilinear compactness on mixed-norm Lebesgue spaces. More precisely, if a multilinear operator is bounded on weighted product spaces, then its compactness can be extrapolated from unweighted product spaces to the full range of weighted mixed-norm spaces. This result is mainly based on a multilinear interpolation theorem of compactness on weighted mixed-norm spaces, for which we present a characterization of compactness on mixed-norm spaces and a multilinear interpolation theorem of boundedness on multi-mixed-norm spaces. As applications of the extrapolation theorem, we obtain compactness results for several kinds of bi-parameter operators on weighted mixed-norm spaces, including multilinear bi-parameter Calderón-Zygmund operators, multilinear bi-parameter dyadic paraproducts, bilinear bi-parameter continuous paraproducts, and bilinear bi-parameter pseudo-differential operators.

Li and Poon proved that every real square matrix is a real linear combination of four real orthogonal matrices. The resulting question, recorded by Zhan, asks whether four terms are necessary. We prove that three real orthogonal matrices always suffice.

Let $\lambda(n)$ denote the Fourier coefficients of a fixed modular form and $h(n)$ a Steinhaus or Rademacher random multiplicative function. In this paper, we determine, under the generalized Riemann hypothesis, the order of magnitude of $\E|\sum_{n \leq x} h(n)\lambda(n)|^{2q}$ up to factors of size $e^{O(q^2)}$, for all real $x, q$ with $1 \leq q \leq c\log x/\log\log x $ and $c>0$ a small constant.

Suppose $\mathcal{V}$ is a class of stationary integral $n$-varifolds in $B^{n+k}_2(0)\subset\mathbb{R}^{n+k}$ which is closed under weak limits, homotheties, rotations, and disjoint decomposition, and suppose that $\mathcal {V}$ satisfies an $\epsilon$-regularity property near planes of (integer) multiplicity $\leq Q\in \{2,3,\dotsc\}$. This last condition, more precisely, requires that there be a constant $\epsilon = \epsilon({\mathcal V}, Q) \in (0, 1)$ such that if $V\in \mathcal{V}$ is, in the unit cylinder ${\mathbb R}^{k} \times B_{1}^{n}(0)$, $\epsilon$-close as varifolds to the plane $\{0\} \times {\mathbb R}^{n}$ taken with multiplicity $\leq Q$ then, in the half-cylinder ${\mathbb R}^{k} \times B_{1/2}^{n}(0)$, $V$ is represented by the graph of a Lipschitz multi-valued function over $B_{1/2}^{n}(0)$ with uniform quantitative estimates of a $C^{1,\alpha}$ nature. For any varifold in such a class $\mathcal{V}$, we prove that the set of branch points with density $\leq Q$ has Hausdorff dimension $\leq n-2$. By choosing suitable $\mathcal{V}$, a direct consequence of this result and the recently established regularity theorems of the second and third authors (one of which being joint with Becker-Kahn) is that if $V$ is a stationary integral $n$-varifold which is either: (a) represented by the graph of a $2$-valued Lipschitz function; or (b) codimension one, stable, and with no classical singularities of density $<Q$, then the Hausdorff dimension of the density $Q$ branch set ($Q=2$ in (a)) is at most $n-2$. Our proof utilises the planar frequency function introduced by the first and third authors in their work on area minimising currents, and thus does not require the Almgren center manifold for the analysis of branch points except in a single, geometrically canonical case where the center manifold satisfies additional simplifying properties.

We show that the top cohomology of any affine Springer fiber, as a Weyl group representation, contains a large part of the total cohomology of certain Springer fibers. The main ingredient of the proof is the construction of a “perverse filtration” on the pure part of the cohomology of affine Springer fibers.

The classical BFGS algorithm performs excellently for convex optimization problems. However, for non-convex problems, the classical BFGS method may fail to converge reliably. To overcome this limitation, researchers have developed modified BFGS methods that are applicable to both convex and non-convex optimization problems. Among these methods, a robust BFGS algorithm has been shown to achieve global convergence and fast local convergence, with a superlinear convergence rate, for both convex and non-convex nonlinear optimization problems under mild assumptions. In this paper, we propose an arc-search BFGS algorithm that aims to further improve the computational efficiency of the robust BFGS method while preserving its desirable convergence properties. Numerical experiments are carried out, and performance comparisons between the proposed algorithm and state-of-the-art algorithms are reported to demonstrate the advantages of the arc-search BFGS algorithm.

We consider a general class of “inexactly smooth” convex functions, providing a universal model capturing as special cases $L$-smooth, $M$-Lipschitz, and Hölder smooth functions, and any combination thereof. Such functions possess a calculus closely following that of smooth functions. Our main results provide inexactly smooth functions with interpolation theorems that are necessary and sufficient up to modest universal constants. These enable analysis of first-order methods for any inexactly smooth convex problem class via solving convex Performance Estimation Problems (PEPs). Further, these enable the extension of Drori and Taylor's constructive approach to algorithm design. From this, we derive an exactly minimax optimal method for $(\beta,0)$-Hölder smooth problems, methods with the best-known convergence guarantees up to constants for any $(\beta,p)$-Hölder smooth convex minimization, and a new universal fast backtracking method for any inexactly smooth convex problem.

Zeroth-order Newton-type methods are useful when gradients and Hessians are unavailable, but they behave quite differently from first-order gradient-free methods. We develop a kinetic framework for algorithms that estimate both gradient and Hessian from black-box function values. The naive random-direction Hessian estimator turns out to be biased even on quadratics; a Gaussian–Stein correction is needed to estimate the Hessian of the Gaussian-smoothed objective. Linearizing the inverse Hessian exposes two noise channels: gradient noise preconditioned by the inverse Hessian, and Hessian noise transmitted through an inverse-Hessian sandwich. Under a noisy oracle the second channel carries the second-difference factor $\mu_H^{-4}$. A small-mass kinetic lift links the finite-step Newton update to an underdamped phase-space model; the overdamped spatial limit yields a Lyapunov bound that exposes the curvature–variance trade-off between step size, batch sizes, smoothing radii, and regularization. Numerical experiments confirm estimator identities, the gradient and Hessian variance laws, dimension scaling, inverse-perturbation accuracy, and optimization behavior under query-budget and regularization ablations.

Horvat et al. (J. Math. Chem., 2014) showed that the rank of the $n$-th Hadamard power $D^{(n)}$ of a Euclidean distance matrix satisfies $\operatorname{rank}D^{(n)} \le R_d^n$, and proved that the inequality is strict whenever an annihilating polynomial exists. The converse - that the absence of annihilating polynomials forces $\operatorname{rank}D^{(n)} = R_d^n$ - was left as an open problem. We resolve it by exhibiting a kernel factorisation $D^{(n)} = \Phi_V\, M\, \Phi_V^T$, where $\Phi_V$ is the evaluation matrix on the polynomial space $V$ and $M$ is a universal matrix independent of the point configuration. A trinomial expansion of the kernel reveals that $M$ has a block-diagonal structure whose blocks are sums of Gram matrices with positive coefficients; this yields the non-singularity of~$M$ and completes the rank identity.

Several elegant proofs of Dilworth's theorem for finite posets already exist in the literature. In this note, we present an alternative inductive proof using a local legal merge lemma on chain ends. By combining this lemma with the bound established by Dilworth's theorem, we observe a direct consequence: every chain decomposition can be transformed into a minimal chain decomposition through a finite sequence of such legal merges

We obtain bounds for the $F$-pure threshold of isolated hypersurface singularities over an algebraically closed field of positive characteristic $p$ in terms of classical singularity invariants. For an irreducible plane curve whose value semigroup has first two generators $\beta_0,\beta_1$, we show that the $F$-pure threshold is at most $\frac{1}{\beta_0}+\frac{1}{\beta_1}$ and differs from it by less than $\frac{1}{\gcd(\beta_0,\beta_1)p}$, with equality when certain base-$p$ digits add without carrying, and we give an example showing that higher weighted-order terms can change the $F$-pure threshold. We also conjecture a lower bound in terms of the Milnor number, and prove it for semi-quasi-homogeneous singularities, for all plane curves, and whenever the number of variables exceeds $p$. In particular, it holds in characteristic $2$. As applications, we reprove Igusa's formula for the log canonical threshold of a complex plane branch (with integer coefficients), prove the conjecture of Mustaţă, Takagi and Watanabe for such branches, and bound the log canonical threshold of complex isolated hypersurface singularities.

We introduce temporal matrix scale invariance (tMSI), a mathematical structure for the two-time correlation kernel of a multivariate observable. A kernel $C(t,t')$ satisfies tMSI of order $\alpha$ if $C(kt, kt') = k^{-\alpha}C(t,t')$ for all $k>0$; this condition holds near a tipping point, where the divergence of the coherence time produces temporal scale freedom. By a kernel factorization theorem, every tMSI kernel separates into a power-law envelope $(tt')^{-\alpha/2}$ and a shape function $F(t/t')$ diagonalized by the Mellin transform. This reveals a decoupling of two independent exponents: the dynamical exponent $\alpha$, carried by the envelope, and the spectral relaxation exponent $\beta$, determined by the eigenvalue decay of the finite-dimensional truncation. Their equality $\alpha = \beta$ characterizes a simple critical point; their inequality $\alpha \neq \beta$ is the signature of temporal multicriticality. We provide a classification of tipping points. The Landau quartic coefficient $a_4$ is given exactly by $a_4 = p^2 + q^2 - 2\lambda pq - g^2_{\alpha\alpha\beta}\Gamma(\sigma_\alpha, \sigma_\beta)$, where $\lambda = 2\sqrt{\sigma_\alpha\sigma_\beta}/(\sigma_\alpha+\sigma_\beta) \in (0, 1]$, $g_{\alpha\alpha\beta}$ is the three-point structure constant, and $\Gamma > 0$ is in explicit closed form. The transition is continuous for $a_4 > 0$, tricritical for $a_4 = 0$, and discontinuous for $a_4 < 0$. The simple critical point $\alpha = \beta$ is maximally fragile: any nonzero operator mixing drives $a_4 < 0$, placing the synchronized state generically at the edge of catastrophe. The framework yields a matrix-valued early warning diagnostic, computable from a multivariate time series without knowledge of the underlying equations, that classifies an approaching tipping point as recoverable or catastrophic. Applications to epilepsy and acute myocardial infarction are discussed.

Let $u$ be a solution of the minimal surface equation on a domain containing the closed unit ball $\overline{B^n}\subset\mathbb R^n$. A classical calibration argument gives $|Graph_u\cap B^{n+1}| \leq \frac12 |\mathbb S^n|.$ A basic question is whether this half-sphere bound is sharp for minimal graphs. We show that it is not. More precisely, for every $n\geq2$, there exists an explicit constant $\delta_n>0$, depending only on $n$, such that $ | Graph_u\cap B^{n+1}|\leq \frac12 |\mathbb S^n|-\delta_n$. The proof combines calibration with a canonical spherical filling associated with the subgraph and a quantitative incompatibility between near equality in the calibration estimate and the divergence-free structure of the minimal surface equation. We also give an improved explicit gap and formulate the corresponding sharp extremal problem.

We derive and analyze the exact solutions of the inverted Dirac-Moshinsky oscillator (IDMO) in $(1+1)$ dimensions, obtained from the standard model via the substitution $p \to p + im\omega\beta x$. The upper spinor component satisfies a Weber equation with complex spectral parameter $\lambda = (E^2-m^2)/(2m\omega)+i/2$, whose solutions are parabolic cylinder functions $D_\nu(\xi)$ with complex order $\nu = \lambda - 1/2$. The physical spectrum is purely continuous ($|E|>m$), with no discrete bound states. Three normalization schemes are developed, and the discrete Gamow resonances at $E_n^\pm = \pm\sqrt{m^2+(2n+1)m\omega-im\omega}$ are identified as poles of the resolvent. The negative-energy sector describes antiparticle anti-resonances whose positive imaginary part signals vacuum instability and spontaneous pair production, analogous to the Schwinger effect. The algebraic structure is governed by the principal series of $SU(1,1)$, and the Hamiltonian is $\mathcal{PT}$-symmetric with unbroken symmetry for $|E|>m$.

We provide a simple construction which realizes the Birkhoff center depth at an arbitrary ordinal level and relate it to the Cantor-Bendixson depth.

Under coprimality assumptions on certain submatrices, Rosenbrock's Theorem relates the Smith form of a matrix $P$ over an elementary divisor domain $\mathcal{R}$ to the Smith-McMillan form of a matrix $G$ over the field of fractions of $\mathcal{R}$, where $G$ is a Schur complement in $P$. If $\mathcal{R}$ is not an elementary divisor domain, Rosenbrock's Theorem is not directly applicable in its original form, because not every matrix is unimodularly equivalent to a matrix in Smith form. In this paper, we state an ideal-theoretic version of Rosenbrock's Theorem that is meaningful over any integral domain, and we show that it is equivalent to the classic formulation over an elementary divisor domain. Moreover, we give a characterization of Prüfer domains as those integral domains over which the ideal-theoretic version of Rosenbrock's Theorem holds for every matrix satisfying the assumptions. In particular, the theorem does not hold for every admissible matrix over $\mathbb{C}[x_1,\dots,x_d]$ when $d \geq 2$. However, when $d\leq 3$, it holds for square matrices in a nonempty Zariski-open subset of the coefficient space, having fixed the block sizes and a total degree bound. Finally, we prove that, if $\mathcal{R}$ is an integral domain such that every right invertible matrix can be completed to a unimodular matrix, then every matrix $P$ that satisfies the assumptions of the ideal-theoretic Rosenbrock's Theorem and realizes the same Schur complement $G$ shares the same ideal-theoretic generalization of the Smith form.

In this expository paper, we develop the basic ideas underlying Grothendieck groups and to illustrate their appearance across algebra, topology, representation theory, and homological algebra. Motivated by the universal construction associated to a commutative monoid, we define the Grothendieck groups abelian categories and rings. Along the way we study several fundamental examples, including Euler characteristics, projective modules, and representation rings. We conclude with a discussion of $K$-theory and its applications, indicating how the elementary construction of $K_0$ serves as the first layer of a much richer homotopical theory.