Papers

Large language models produce prompt-dependent probabilities over words, whereas scientific systems require uncertainty over meaningful states that can be updated as evidence arrives. We develop an observable framework for determining when language-derived probabilities support such a sequential state representation. Theoretically, we define typed measurable transformations of contextual language, construct a minimal closed representation, and give necessary and sufficient conditions for semantic updates to exist uniquely. We bound irreducible nonclosure and accumulated error, and under average contraction prove existence, uniqueness and stability of an external random recursion on a probability simplex. These results define a stochastic lexical calculus without attributing an internal calculus to the language model. Empirically, frozen experiments test the observable implications. Raw prompt-conditioned probabilities fail the prespecified invariance gate; after prompt-specific calibration, a common three-state representation passes the stability gates and covers 28 of 30 untouched eight-step paths, or 0.933 at nominal level 0.90. Accordingly, language probabilities support a stochastic state only conditionally on verified closure, stability and coverage within a declared operating domain.

In this paper, we establish an explicit boundary-corrected Li–Yau type gradient estimate for positive solutions of the Dirichlet heat equation on the Euclidean half-space. The main idea is to exploit the reflection structure of the Dirichlet heat kernel and introduce a normalized kernel-induced probability measure. Under this representation, logarithmic derivatives of the heat kernel become expectations of explicit kernel quantities. The reflected Gaussian component generates a hyperbolic correction term involving \[ \coth\left(\frac{x_n y_n}{2t}\right), \] which has no analogue in the whole Euclidean heat equation. Using Jensen's inequality and the sharp estimate \[ 0<z\,\csch z<1 , \qquad z>0, \] we prove that every positive solution satisfies \[ \Delta\log w(x,t) \geq -\frac n{2t} -\frac1{x_n^2}. \] The first term represents the classical Euclidean Li–Yau diffusion scaling, while the second term is an explicit inverse-square correction determined by the distance to the Dirichlet boundary. Our approach provides a direct kernel interpretation of the boundary effect and suggests possible extensions to more general domains.

We study one-dimensional particle swarm optimization during stagnation, with two fixed distinct attractors and equal independent uniform acceleration ranges. The position then satisfies a second-order random affine recurrence. For inertia $w$ and acceleration range $c$, we prove that throughout the open mean-square stability region \[ -1<w<1,\qquad c>0,\qquad 12(1-w^2)-c(7-5w)>0, \] no invariant position marginal, and hence no limiting position marginal, can be Gaussian. This solves the open Problem 18 in \cite{ParticleSwarmProblems}. The proof compares the stationary moment equations with the Gaussian moment identities through order eight. A Hermite-polynomial formulation gives explicit fourth- and sixth-order compatibility conditions whose common solutions lie on a degree-$107$ polynomial branch. Exact eighth-order equations exclude every point on that branch. The final certificate is verified using arithmetic modulo $23$ and independently modulo $1{,}000{,}003$. A separate raw-moment implementation produces exact polynomials $Q_4,Q_6,Q_8$ in $(w,c)$ and verifies the same obstruction over the rational numbers. The fourth- and sixth-order curves have a genuine admissible intersection, but the eighth-order condition removes it, showing why low-order Gaussian diagnostics are insufficient. All code, exact polynomials, logs, and plot-validation data are supplied as online resources.

We study the size of the largest monochromatic connected component that must appear in any edge-coloring of a random graph. Let $G\sim G(n,p)$ with $p\gg 1/n$ and $p=o(1)$, and write $np=he^h$. We show that, with high probability, every $2$-edge-coloring of $G$ contains a monochromatic connected component of order at least $n-\Theta(ne^{-h})$. Moreover, we construct colorings showing that this bound is best possible up to constant factors. We extend this result to three colors: for $p\gg 1/n$ and $p=o(1)$, with high probability every $3$-edge-coloring of $G$ contains a monochromatic connected component of size at least $\frac{n}{2}-\Theta(1/p)$, and this estimate is again tight up to constant factors. In the bipartite setting $G\sim G(n,n,p)$, under the same assumptions on $p$, we prove an analogous statement: with high probability, every $2$-edge-coloring contains two monochromatic components whose union covers all but $\Theta(ne^{-h})$ vertices, and this bound is asymptotically sharp. Our approach is elementary and is based on analyzing large connected structures across suitably balanced vertex partitions.

We present a proof that determines the optimal value of the universal constant in the thin-shell theorem for log-concave distributions in high dimensions. We prove that for any log-concave random vector $X = (X_1,\ldots,X_n)$ in $\mathbb{R}^n$ with mean zero and identity covariance, $$ {\rm Var}( |X|^2 ) \leq 8 n. $$ The constant $8$ is optimal: equality is attained when $X_1,\ldots,X_n$ are independent, identically distributed, standard, centered exponential random variables. Moreover, among isotropic random vectors distributed uniformly on convex bodies in $\mathbb{R}^n$, the quantity ${\rm Var}(|X|^2)$ is maximized by the uniform distribution on a regular simplex. We also provide a corresponding sharp bound on the Hilbert-Schmidt norm of the tensor of $3^{rd}$-moments of isotropic, log-concave distributions. The argument relies on the analysis of a weighted Riemannian manifold associated with log-concave moment measures and the Monge-Ampère equation. This manifold was studied in this context in \cite{lc_moment}. The main improvement over \cite{lc_moment} comes from a concise yet effective analysis of the $3^{rd}$-derivatives tensor of the potential. The proof was found by GPT-5.6 Pro in response to prompts supplied by the first-named author, following general discussions between the two authors concerning log-concave moment measures. The prompts referred to the paper “Logarithmically-concave moment measures I” and suggested bootstrapping a bound on the second trace moment.

We consider a random walk $\{S_n\}$ with a finite positive drift that is stopped at a random time $\tau$ having an intermediate regularly varying distribution. We assume that the jump distribution is lighter-tailed than the distribution of $\tau$. Under these conditions, we show that the tails of the distributions of $S_{\tau}$ and $M_{\tau} = \max_{k\le \tau} S_k$ are asymptotically equivalent and are determined by the tail of $\tau$, while the random walk $\{S_n\}$ contributes only through the law of large numbers.

The classical Grünbaum inequality asserts that the proportion of the volume of a convex body cut off by a halfspace containing its barycenter is at least $1/e$. From its functional counterpart, for any log-concave random variable $X$, one has $\mathbb{P}(X\ge \mathbb{E}X)\ge 1/e$, with equality if and only if $X$ is exponential. Motivated by Grünbaum's inequality for convex bodies and its functional generalizations, we prove analogous inequalities for entropy, with characterizations of the equality cases. We show that if $X$ is a log-concave random variable on $\mathbb{R}$, then $$ h(X)-\frac{e}{e-1}H_2(1/e) \leq h(X|X \leq \mathbb{E}X) \leq h(X), $$ where $h$ is the differential entropy, $H_2(\cdot)$ is the binary entropy function and $X|X\leq \mathbb{E}X$ stands for the distribution of $X$ conditional on $X\leq \mathbb{E}X$. We generalize the upper bound for all Rényi entropies and the lower bound for min-entropy. Our inequalities are sharp and we characterize all equality cases. We discuss potential generalizations in high dimensions and give counterexamples in some directions. As an intermediate step for the proof of the lower bound, we establish a new inequality that we prove using a technique known as degrees of freedom, combined with a standard KKT-type optimization lemma. Along the way, we characterize the equality case in a known comparison inequality between differential and min-entropy, which may be of independent interest.

We establish an exact representation formula for the deficit in Hölder's inequality. Rather than estimating the deficit by auxiliary quantities, we show that it can be represented as an accumulated variance along the natural exponential interpolation connecting the two endpoint densities. More precisely, the logarithmic Hölder deficit is expressed as the integral of the variance of the logarithmic density ratio weighted by the Green kernel associated with the one-dimensional interpolation parameter. This identity reveals that Hölder's inequality is a consequence of the convexity of a logarithmic partition function and provides an intrinsic interpretation of the deficit as an interpolation energy. The representation suggests a broader framework for studying functional inequalities through variance identities.

We derive the $\mathcal{P}(\Phi)_2$ measure on the two-dimensional unit torus as the rigorous limit of many-body quantum Gibbs states. In particular, the quantum model corresponding to the $\Phi^{2p}_2$ measure is formulated in the grand-canonical ensemble with a general symmetric $p$-body interaction potential which is not required to have a factorized form. Unlike in the $\Phi^4_2$ case investigated by Fröhlich–Knowles–Schlein–Sohinger in \cite{FKSS25}, in the treatment of higher-order interactions, Wick renormalization generates a full hierarchy of lower-order interactions which we organize systematically using a graphical formalism. One key ingredient of our analysis is a logarithmic stability estimate for both the nonlocal Hartree functional and the many-body Hamiltonian that is uniform in the interaction range, allowing us to use the positivity of the leading $p$-body interaction to control all lower-order terms generated by the Wick counterterms.

We study the high-temperature mean-field limit of grand-canonical bosonic Gibbs states on the torus with renormalized nonlocal three-body interactions. In dimensions two and three, we construct the limiting nonlinear classical Gibbs measure and prove convergence of the relative free energy and of the reduced density matrices of every fixed order; in three dimensions, a smallness condition on the interaction is imposed. The proof combines the density-channel representation of the interaction with a coherent-state variational method based on the upper-symbol representation of the free Gibbs state. The same framework also contains, as a special case, the homogeneous positive-type model studied by Lewin, Nam, and Rougerie (2021).

Let $\Pi$ be a $k\times n$ sparse random matrix. For a fixed $r$-dimensional subspace $V\subset{\mathbb R}^n$, let $U_V:{\mathbb R}^r\to{\mathbb R}^n$ denote an isometry from ${\mathbb R}^r$ onto $V$. The product $\Pi U_V$ is a central model in randomized dimension reduction and has been studied primarily through trace and Gaussian comparison inequalities. In this work, we develop an approach to the spectral norm of the matrix product $\Pi U_V$, based on entropy estimates for level sets of vectors $x\in V$. Combining the method with existing estimates, we show the following. Assume that \[ k\ge C\,r(\log\log r)^2,\qquad p\ge (\log k)/k. \] Let $\Pi$ be a $k\times n$ matrix with i.i.d. entries equidistributed with the product $b\,\xi$, where $b$ is a Bernoulli($p$) random variable and $\xi$ is mean-zero, independent of $b$, and satisfies $|\xi|\le1$ almost surely. Then with high probability \[ \|\Pi U_V\|\le C\sqrt{kp}. \] Matching results hold for other random models with negatively associated entries.

The symmetric exclusion process (SEP) is a classical model of interacting particles on a graph, in which multiple particles execute random walks subject to the constraint that no two particles may occupy the same vertex. The version of the process in which the number of particles is fixed, is by now very well understood, through the study of reversible Markov chains. In this paper, we study the non-equilibrium version of the SEP, in which some vertices of the graph interact with external heat baths, held at different temperatures, that mediate transport through the graph. The resulting steady-state distribution is still the stationary distribution of a natural Markov chain, but the chain is no longer reversible. This means that even the steady-state distribution is very hard to describe, and indeed is the subject of a rich literature, which has focused almost exclusively on graphs which are finite subsets of the lattice $Z^d$. Very little is known about other important quantities, notably the mixing time. We derive various quantitative results about the non-equilibrium SEP for arbitrary graphs with arbitrary heat baths. Our main result is a bound on the mixing time in terms of the worst-case hitting time of a single particle to any of the heat baths; this bound is tight up to two logarithmic factors in the size of the graph. We then use this result to derive a simple algorithm, running in time roughly $n^{O(k)}$, that computes the steady-state joint occupation distribution for any set of $k$ vertices; since the steady-state distribution is very elusive, this provides a potentially useful tool to study its $k$-wise correlations for small values of $k$. Finally, we prove that the spectrum of the Markov chain associated with the non-reversible dynamics for the SEP is independent of the heat bath temperatures; thus in particular the spectrum is always real.

Let $X = (X_1, \ldots, X_n)$ be a random vector from any Borel probability law on $\mathbb{R}_+^n$. We revisit the problem of deriving a lower confidence bound (LCB) on a scalar parameter of that law. We recast classical work, beginning with Buehler, in purely probabilistic terms to form a more accessible and extensible framework. We then specialize the framework to the case where the components of $X$ are independent. In this context, we prove that Gaffke's bound is Buehler optimal for the order that it induces with respect to the maximum marginal mean parameter: $max_{i \in [n]} E_Q[X_i]$, which reduces to the common mean when the $X_i$ are independent and identically distributed. That is to say, no other valid LCB that orders samples in the same way as Gaffke's bound can improve on it with respect to this parameter.

The Karpelevič theorem describes the complex numbers that occur as eigenvalues of stochastic matrices of a fixed order. We give a self-contained proof based on invariant polygons. Their geometry yields a product equation for an eigenvalue of maximal modulus at a fixed argument. Convexity determines this modulus, explicit stochastic matrices attain it, and a comparison between Farey intervals completes the boundary description. As the matrix order increases, the resulting equation gives a uniform relative asymptotic for the gap $1-|z|$ between a boundary point $z$ and the unit circle, even arbitrarily close to arc endpoints. We also determine the optimal contraction factor for a planar rotation and dilation when vector size is measured by scaling a polygon with at most $N$ vertices.

We investigate a joint optimization problem of dividend payments and capital injections for a surplus process driven by a general diffusion. Dividend payments are assumed to be absolutely continuous in time, with the dividend rate bounded by a nonnegative concave function of the current surplus; while capital injections are modelled by a general nondecreasing process. We first analyze an auxiliary bail-out problem in which capital injections are required to keep the surplus nonnegative at all times. Under a concavity assumption on the drift, we prove that the associated value function is concave and is a classical solution of the corresponding Hamilton-Jacobi-Bellman (HJB) equation. We further characterize an optimal policy as a refraction-reflection strategy: the surplus is reflected at zero by capital injections, while dividends are paid at the maximal admissible rate whenever the surplus exceeds an optimal threshold. Our main contribution establishes that the general optimization problem exhibits a Lokka-Zervos type dichotomy. More precisely, an optimal policy is either a dividend refraction strategy without injections, in which ruin occurs, or a refraction-reflection dividend-injection strategy. This optimal injection decision is characterized through a simple comparison of the auxiliary value functions at the origin.

The Constantin-Iyer formula represents a smooth solution of the incompressible Navier-Stokes equations on $\mathbb{T}^3$ as $u=\mathbb{P}\,\mathbb{E}[(\nabla A_t)^\top(u_0\circ A_t)]$, the projected expectation of a stochastic Weber field. We separate this expectation into the product of the means and a centred covariance, and show that the covariance – together with the second moment of the inverse deformation gradient and the two-point covariance – satisfies a closed deterministic advection-diffusion equation: because every Lagrangian label is driven by the same Brownian path, the quadratic covariation collapses into a Laplacian and no moment hierarchy appears. From these equations we obtain an energy/production balance, pointwise and Holder covariance bounds, and an exact degeneracy of the two-point operator in the separation variable. A second group of results concerns the mean displacement, which solves a forced heat equation along the flow; its $L^2$ norm is controlled unconditionally by the initial energy, while an exact family of Navier-Stokes shear flows shows that no analogous Holder bound can hold uniformly down to $t=0$. We further isolate several obstructions, each backed by an explicit counterexample – the norm of an expected deformation gradient does not control the expected norm, even for genuine stochastic flows of smooth divergence-free drifts; a spatial Holder bound on the mean does not imply parabolic Campanato decay – and we prove a gauge-invariant quotient criterion equivalent to the Serrin norm of the velocity, together with a Liouville-type rigidity theorem for ancient solutions whose mean coordinate is affine of rank at least one. No claim is made regarding global regularity; the missing steps are stated explicitly as open problems.

We develop a framework for quickest detection of distributional change in signals modeled as rough paths. By representing the pre-change and post-change dynamics as two independent rough paths and modeling the observed signal as their concatenation at an unknown random change-point, we formulate quickest detection as an optimal stopping problem using rough paths. We show that the optimal stopping rule takes the form of the first hitting time to a half-space by a linear functional of the rough path signature, and establish that the same structural form arises independently from the intrinsic geometry of the observed path. Statistical guarantees on detection delay and false alarm probability are derived, and the framework is extended to a distributionally robust formulation in which both the pre-change and post-change models are uncertain. The proposed rules are implemented via zeroth-order stochastic approximation over truncated-signature coefficients and evaluated numerically under Brownian and fractional Brownian dynamics, achieving performance comparable to optimal methods in the Brownian setting while outperforming them and remaining robust to adversarial path perturbations in the fractional Brownian setting.

We consider a model for the spread of an infection within an interacting particle system on $\mathbb{Z}^d$, generalizing a framework introduced by Kesten and Sidoravicius to a zero-range process in equilibrium with density $\rho > 0$. In our model, at any time, individuals are either healthy or infected. The infection spreads instantaneously whenever infected and healthy particles occupy the same site, while infected particles heal independently at rate $\delta > 0$. We investigate the extinction-survival phase diagram of this process starting from a configuration where only the particles at the origin are infected. For every fixed positive healing rate, we prove that the infection becomes extinct almost surely if the density $\rho$ is sufficiently small and survives with positive probability if $\rho$ is sufficiently large, establishing the existence of a non-trivial critical density. At sufficiently high densities, survival occurs even under instantaneous healing. We also show that, for every positive density, the infection survives when the healing rate is sufficiently small.

Given a strongly convex function $u$, equip $R^d$ with a Riemannian metric given by the Hessian $\nabla^2 u$. This is a so-called Hessian manifold. Given a probability density $\mu$ one may run a Langevin diffusion intrinsic to the manifold with stationary distribution $\mu$. Such (Hessian) manifold-valued Langevin diffusions are called Mirror Langevin diffusions (MLD) which have recently become popular. One of the questions we explore is whether, given $\mu$, one can choose $u$ to get an exponential convergence to equilibrium for the MLD, especially if $\mu$ is not strongly log-concave. Our results are based on Lyapunov function methods and give sufficient conditions for a Poincaré or a log-Sobolev inequality to hold for the MLD. These, in turn, imply exponential convergence. We also introduce a Markov chain approximation to the MLD given by a two step Gibbs sampler with stationary distribution $\mu$. This Markov chain is a variant of the Sinkhorn Markov chain introduced in arXiv:2307.16421 that is conjectured to converge to a time-inhomogeneous generalization of the MLD. Under suitable assumptions, we prove that the Markov chain has a guaranteed convergence rate in $\chi^2$ that is consistent with the diffusion time scale. Our proofs are based on ideas from entropic optimal transport and strong data processing inequalities.

Clustering of large deviations events in a stationary stochastic process depends critically on the interplay between the tail behavior of the marginal distribution and the strength of temporal dependence. In the class of doubly infinite moving average processes, when the memory is short, previous work has established a sharp contrast in the clustering patterns between light- and heavy-tailed settings, governed by conspiracy and catastrophe principles respectively. It has also been described how long memory interacts with the conspiracy principle to affect clustering in the light-tailed case. This paper addresses the interaction of long memory with the catastrophe principle in the heavy-tailed case. It turns out that long memory generally allows for a wider range of catastrophes to play a role and leads to longer and qualitatively different clustering patterns.

This paper investigates the discrepancy of a family of random sampling methods obtained by perturbing the grid $\frac{1}{M}\mathbb{Z}^{d}\cap\left[ -1/2,1/2\right)^{d}$, where $M$ is a large positive integer. Parameterized by an arbitrary probability measure $\mu$, this family encompasses several classical methods for evaluating the quality of an $N$-point set in $\mathbb{T}^{d}$, where $N=M^{d}$. We show that all probability measures, except for Dirac measures, behave like the Lebesgue measure in the Monte Carlo discrepancy. This represents a limiting case where the measure $\mu$ depends on $M$. In this latter context, we prove that, up to a constant, the lowest possible discrepancy is achieved when the support of $\mu$ has diameter $\leq c/M$, and that this upper bound is sharp.

The spherical random geometric graph $G(n,d,p)$ is obtained by sampling $n$ independent points uniformly on the unit sphere $\mathbb{S}^{d-1}\subseteq\mathbb{R}^d$ and joining pairs of points which are sufficiently close, where the threshold is chosen so that the edge probability is $p$. The central question related to this model, and to a broad class of other models, is the following: when does the underlying geometry affect the resulting graph in a way which makes it distinguishable from the Erdős–Rényi random graph $G(n,p)$, as measured in total variation distance? The precise answer to this question was conjectured by Bubeck, Ding, Eldan, and Rácz, who predicted that $G(n,d,p)$ and $G(n,p)$ are indistinguishable precisely when $d \gg n^3p^3(\log p^{-1})^3$, and provided a test for distinguishing these models in the low-dimensional regime. Although this conjecture attracted considerable attention from researchers in probability, theoretical computer science, and high-dimensional statistics, it was previously fully proved only in the constant-density case. In this paper, we resolve the distinguishability conjecture in the broad range $1/3 \geq p \geq n^{-1/5} \text{polylog}(n)$. The key ingredient of our proof is a stronger statement which gives a precise asymptotic formula for the probability that $G(n,d,p)$ realizes a prescribed graph $H$: above the conjectured threshold, this probability is at most $(1+o(1))$ times the corresponding probability for $G(n,p)$, with the signed triangle count of $H$ appearing as the leading correction term.

We study Gaussian fluctuations for a lower-triangular system of interacting diffusions in which particle $i$ interacts only with its predecessors. Although the empirical measure of this system converges to the same McKean–Vlasov limit as in the corresponding exchangeable mean-field system, the sequential structure remains visible at the $N^{-1/2}$ fluctuation scale. We introduce the logarithmically weighted fluctuation fields \[ Y_t^{N,n} = \frac{1}{\sqrt{N}} \sum_{i=1}^N \frac{\bigl(\log(N/i)\bigr)^n}{n!} (\delta_{X_t^i}-\bar\rho_t), \qquad n\ge 0, \] and prove joint convergence of the entire family in a countable product of weighted negative Sobolev path spaces to the unique probabilistically strong solution of a linear hierarchy in which $Y^n$ couples to $Y^{n+1}$. In particular, the limit of the empirical fluctuation field $\sqrt{N}(\mu_t^N-\bar\rho_t)=Y_t^{N,0}$ is not governed by the closed fluctuation SPDE arising in the classical exchangeable case. The proof combines conditional-measure replacement, deterministic estimates for the logarithmic weights, a martingale argument, and a weighted Volterra estimate.

Piecewise Deterministic Monte Carlo (PDMC) algorithms utilize continuous time Markov processes to generate samples from continuous distributions, and provide a modern alternative to discrete time Markov chain Monte Carlo algorithms. In this work we provide a survey of recent results on scaling limit arguments to understand the behaviour and efficiency of two often-used Piecewise Deterministic Monte Carlo algorithms: the Bouncy Particle Sampler and the Zig-Zag Process. In particular we discuss a Functional Central Limit Theorem, scaling in the high-dimensional regime, scaling under anisotropy, and scaling in the big data regime. This work is intended as part of the proceedings of the 2024 Isaac Newton Institute Programme "Stochastic systems for anomalous diffusion".

We prove a martingale representation theorem for cylindrical martingale-valued measures defined on a separable Banach space. The main tool for establishing the theorem, is a new theory of non-radonifying stochastic integration in reflexive Banach spaces. A second one is the study and characterization of cylindrical white noise measure processes. As consequences of our representation theorem, we prove analogous versions for Hilbert space-valued measures and for cylindrical square integrable martingales. Finally, we apply the results to characterize the solutions to the weak martingale problem for SDEs driven by cylindrical white noise measures.

We study two interacting urns with power reinforcement $W(n)=n^\alpha$, for $\alpha>1$. Using Qin's stochastic-approximation theorem as input, we reduce the phase transition problem for domination to the sign analysis of a single analytic function $\Phi_p$. This deterministic reduction identifies the sharp threshold and proves that the two natural critical parameters for domination coincide. We determine the first-order behavior of the critical interaction parameter as $\alpha\downarrow1$ and show that it is not nondecreasing in the reinforcement exponent, answering two questions posed by Qin.

This paper studies a multi-player non-zero-sum optimal stopping game driven by a spider Brownian bridge. We begin by analyzing the two-player setting, in which the underlying process is a skew Brownian bridge, and establish sufficient conditions for the existence of a Nash equilibrium. We derive sufficient conditions for the uniqueness of the Nash equilibrium in the symmetric case, (where the skew Brownian bridge reduces to a standard Brownian bridge and both players have the same payoff structures). We also give examples which show that there can be multiple Nash equilibria. We then extend the framework to the multi-player setting and provide sufficient conditions for the existence of a Nash equilibrium in this more general setting with a spider Brownian bridge.

We consider an irreducible, supercritical multi-type general branching process (Crump-Mode-Jagers process) with finite type space, counted with random characteristic. We prove that, under certain second moment assumptions, a central limit theorem holds for the process. Our result extends the single-type central limit theorem obtained recently in [Ann. Probab. 52 (2024), no. 4, 1538-1606] and unifies various limiting results for specific branching processes.

We prove that there exist positive constants $A$ and $V$ such that the high-dimensional critical percolation $k$-point function is given by \[ T_{p_c}(x_1,x_2,\ldots,x_{k}) \sim V^{k-2} A^{2k-3} \sum_{T\in \mathsf{Tr}(k)} \sum_{\substack{\Phi:V(T)\to \mathbb{Z}^d \\ \Phi(i)=x_i \forall 1\leq i \leq k}} \prod_{\substack{u,v\in V(T)\\u\sim v}}G(\Phi(u),\Phi(v)) \] as $\min_{i\neq j}\|x_i-x_j\|\to \infty$, where $\mathsf{Tr}(k)$ is a set of isomorphism class representatives of trees with $k$ labelled leaves $\{1,\ldots,k\}$ and unlabelled internal vertices all of which have degree $3$ and $G$ is the lattice Green's function. This verifies a conjecture of Aizenman and Newman (1984) subject to the usual perturbative conditions needed for convergence of the lace expansion. It follows from this theorem that the law of the cluster of the origin, considered as the counting measure on its range, converges under rescaling to the canonical measure of the integrated super-Brownian excursion. By computing the asymptotics of various more complicated variations on the $k$-point function, we also prove the stronger result that the cluster converges as an embedded metric-measure space to the continuum random tree equipped with its Brownian embedding into $\mathbb{R}^d$. This convergence holds simultaneously with respect to the chemical distance, pivotal distance, and resistance distance on the cluster, which we prove are asymptotic to constant multiples of each other. This resolves conjectures of Hara and Slade (1998) and van der Hofstad and Slade (2003). As a corollary of our results we prove that there exists a positive constant $C$ such that $\mathbb{P}_{p_c}(0\leftrightarrow \mathbb{Z}^d \setminus [-r,r]^d)\sim C r^{-2}$, answering a question of Heydenreich and van der Hofstad (2017).

For the heat equations driven by a stochastic measure on $[-L, L]$ with the Dirichlet boundary condition, we prove that solutions tend to the solution of the heat equations defined on ${\mathbb R}$ as $L\to \infty$. The estimate of the convergence rate is obtained. For a stochastic measure, we assume the $\sigma$-additivity in probability only.