Searcharxiv⌕ Search

arXiv subjects

Felipe Hernández

Publications and source records attributed to Felipe Hernández.

27 records · Page 2Linked to original sources
arXiv2025

Classical correspondence beyond the Ehrenfest time for open quantum systems with general Lindbladians

Quantum and classical systems evolving under the same formal Hamiltonian $H$ may dramatically differ after the Ehrenfest timescale $t_E \sim \log(\hbar^{-1})$, even as $\hbar \to 0$. Coupling the system to a Markovian environment results in a Lindblad equation for the quantum evolution. Its classical counterpart is given by the Fokker-Planck equation on phase space, which describes Hamiltonian flow with friction and diffusive noise. The quantum and classical evolutions may be compared via the Wigner-Weyl representation. Due to decoherence, they are conjectured to match closely for times far beyond the Ehrenfest timescale as $\hbar \to 0$. We prove a version of this correspondence, bounding the error between the quantum and classical evolutions for any sufficiently regular Hamiltonian $H(x,p)$ and Lindblad functions $L_k(x,p)$. The error is small when the strength of the diffusion $D$ associated to the Lindblad functions satisfies $D \gg \hbar^{4/3}$, which allows vanishing noise in the classical limit. Our method uses a time-dependent semiclassical mixture of variably squeezed Gaussian states. The states evolve according to a local harmonic approximation to the Lindblad dynamics. Both the exact quantum trajectory and its classical counterpart can be expressed as perturbations of this semiclassical mixture, with the errors bounded using Duhamel's principle. We present heuristic arguments suggesting the $4/3$ exponent is optimal and defines a boundary in the sense that asymptotically weaker diffusion permits a breakdown of quantum-classical correspondence at the Ehrenfest timescale. In a shorter companion paper, we treat the special case of Hamiltonians that decompose into kinetic and potential energy with linear Lindblad operators, with explicit bounds that can be applied directly to physical systems.

math-ph↗
arXiv2025

Tail bounds for the Dyson series of random Schrödinger equations

We study Schrödinger equations on $\mathbb{Z}^d$ and $\mathbb{R}^d$, $d\geq 2$ with random potentials of strength $λ$. Our main result gives tail bounds for the terms of the Dyson series that are effective at time scales on the order of $λ^{-2+\varepsilon}$. As corollaries, we obtain estimates on the frequency localization and spatial delocalization of approximate eigenfunctions in the spirit of works by Schlag-Shubin-Wolff and T. Chen. These estimates also apply to Floquet states associated to time-periodic potentials. Our proof is elementary in that we use neither sophisticated harmonic analysis nor diagrammatic arguments. Instead, we use only the noncommutative Khintchine inequality from random matrix theory combined with pointwise dispersive estimates for the free Schrödinger equation.

math-ph↗
arXiv2024

On recurrence for $\mathbb{Z}^d$-Weyl systems

We study the topological recurrence phenomenon of actions of locally compact abelian groups on compact metric spaces. In the case of $\mathbb{Z}^d$-actions we develop new techniques to analyze Bohr recurrence sets. These techniques include finding and manipulating correlations between the coordinates of the set of recurrence. Using this, we show that Bohr recurrence sets are recurrence sets for $\mathbb{Z}^d$-Weyl systems. This family encompasses, for example, all $\mathbb{Z}^d$-affine nilsystems. To our knowledge, this is the first result towards a positive answer to Katznelson question in the case of $\mathbb{Z}^d$-actions.

math.DS↗
arXiv2023

On the multiplicative independence between $n$ and $\lfloor αn\rfloor$

In this article we investigate different forms of multiplicative independence between the sequences $n$ and $\lfloor n α\rfloor$ for irrational $α$. Our main theorem shows that for a large class of arithmetic functions $a, b \colon \mathbb{N} \to \mathbb{C}$ the sequences $(a(n))_{n \in \mathbb{N}}$ and $(b ( \lfloor αn \rfloor))_{n \in \mathbb{N}}$ are asymptotically uncorrelated. This new theorem is then applied to prove a $2$-dimensional version of the Erdős-Kac theorem, asserting that the sequences $(ω(n))_{n \in \mathbb{N}}$ and $(ω( \lfloor αn \rfloor)_{n\in \mathbb{N}}$ behave as independent normally distributed random variables with mean $\log\log n$ and standard deviation $\sqrt{ \log \log n}$. Our main result also implies a variation on Chowla's Conjecture asserting that the logarithmic average of $(λ(n) λ( \lfloor αn \rfloor))_{n \in \mathbb{N}}$ tends to $0$.

math.NT↗
arXiv2023

Analyzing Populations of Neural Networks via Dynamical Model Embedding

A core challenge in the interpretation of deep neural networks is identifying commonalities between the underlying algorithms implemented by distinct networks trained for the same task. Motivated by this problem, we introduce DYNAMO, an algorithm that constructs low-dimensional manifolds where each point corresponds to a neural network model, and two points are nearby if the corresponding neural networks enact similar high-level computational processes. DYNAMO takes as input a collection of pre-trained neural networks and outputs a meta-model that emulates the dynamics of the hidden states as well as the outputs of any model in the collection. The specific model to be emulated is determined by a model embedding vector that the meta-model takes as input; these model embedding vectors constitute a manifold corresponding to the given population of models. We apply DYNAMO to both RNNs and CNNs, and find that the resulting model embedding spaces enable novel applications: clustering of neural networks on the basis of their high-level computational processes in a manner that is less sensitive to reparameterization; model averaging of several neural networks trained on the same task to arrive at a new, operable neural network with similar task performance; and semi-supervised learning via optimization on the model embedding space. Using a fixed-point analysis of meta-models trained on populations of RNNs, we gain new insights into how similarities of the topology of RNN dynamics correspond to similarities of their high-level computational processes.

cs.LG↗
arXiv2022

Quantum Diffusion via an Approximate Semigroup Property

In this paper we introduce a new approach to the diffusive limit of the weakly random Schrodinger equation, first studied by L. Erdos, M. Salmhofer, and H.T. Yau. Our approach is based on a wavepacket decomposition of the evolution operator, which allows us to interpret the Duhamel series as an integral over piecewise linear paths. We relate the geometry of these paths to combinatorial features of a diagrammatic expansion which allows us to express the error terms in the expansion as an integral over paths that are exceptional in some way. These error terms are bounded using geometric arguments. The main term is then shown to have a semigroup property, which allows us to iteratively increase the timescale of validity of an effective diffusion. This is the first derivation of an effective diffusion equation from the random Schrodinger equation that is valid in dimensions $d\geq 2$.

math-ph↗
arXiv2020

A calibration framework for high-resolution hydrological models using a multiresolution and heterogeneous strategy

Increasing spatial and temporal resolution of numerical models continues to propel progress in hydrological sciences, but, at the same time, it has strained the ability of modern automatic calibration methods to produce realistic model parameter combinations for these models. This paper presents a new reliable and fast automatic calibration framework to address this issue. In essence, the proposed framework, adopting a divide and conquer strategy, first partitions the parameters into groups of different resolutions based on their sensitivity or importance, in which the most sensitive parameters are prioritized with highest resolution in parameter search space, while the least sensitive ones are explored with the coarsest resolution at beginning. This is followed by an optimization based iterative calibration procedure consisting of a series of sub-tasks or runs. Between consecutive runs, the setup configuration is heterogeneous with parameter search ranges and resolutions varying among groups. At the completion of each sub-task, the parameter ranges within each group are systematically refined from their previously estimated ranges which are initially based on a priori information. Parameters attain stable convergence progressively with each run. A comparison of this new calibration framework with a traditional optimization-based approach was performed using a quasi-synthetic double-model setup experiment to calibrate 134 parameters and two well-known distributed hydrological models: the Variable Infiltration Capacity (VIC) model and the Distributed Hydrology Soil Vegetation Model (DHSVM). The results demonstrate statistically that the proposed framework can better mitigate equifinality problem, yields more realistic model parameter estimates, and is computationally more efficient.

stat.AP↗
arXiv2017

Scalable computation of Jordan chains

We present an algorithm to compute the Jordan chain of a nearly defective matrix with a $2\times2$ Jordan block. The algorithm is based on an inverse-iteration procedure and only needs information about the invariant subspace corresponding to the Jordan chain, making it suitable for use with large matrices arising in applications, in contrast with existing algorithms which rely on an SVD. The algorithm produces the eigenvector and Jordan vector with $O(\varepsilon)$ error, with $\varepsilon$ being the distance of the given matrix to an exactly defective matrix. As an example, we demonstrate the use of this algorithm in a problem arising from electromagnetism, in which the matrix has size $212^2\times 212^2$. An extension of this algorithm is also presented which can achieve higher order convergence [$O(\varepsilon^2)$] when the matrix derivative is known.

math.NA↗