arXiv research

A new method learns straight trajectories in one step for optimal flow matching.

This paper improves the efficiency of generative models by optimizing the straightness of Rectified Flow.

Study of straight-line flows on a unique infinite surface.

MFM improves generative model interpolations by learning approximate geodesics on data manifolds.

The automatic reconstruction of three-dimensional particle tracks from Active Target Time Projection Chambers data can be a challenging task, especially in the presence of noise. In this article, we propose a non-parametric algorithm that is based on the idea of clustering point triplets instead of the original points.…

New geometric approach controls motion of a spinning sphere on a plane.

We present two families of knots which have straight number higher than crossing number. In the case of the second family, we have computed the straight number explicitly. We also give a general theorem about alternating knots that states adding an even number of crossings to a twist region will not change whether the …

Jablan and Radović originally defined two invariants called the Meander number and OGC number of knots for certain classes of knots. We generalize these definitions to all knots and name the straight number and contained straight number of a knot, respectively, and prove they are well defined. We answer two questions a…

The study characterizes straight-line flows in dynamic measure transport.

The paper triangulates Heisenberg groups with horizontal and straight simplexes.

In this paper we present the distinguished (d-) Riemannian geometry (in the sense of nonlinear connection, Cartan canonical linear connection, together with its d-torsions and d-curvatures) for a possible Lagrangian inspired by optics in non-uniform media. The corresponding equations of motion are also exposed, and som…

Study on straight-line flows for generative modeling with theoretical obstructions.

We generalise a result of Garofalo and Pauls: a horizontally minimal smooth surface embedded in the Heisenberg group is locally a (straight) ruled surface, i.e. it consists of straight lines tangent to a horizontal vector field along a smooth curve. We show additionally that any horizontally minimal surface is locally …

A novel approach learns goal-conditioned policies for locomotion using batch RL.

Correct method found for drawing precise envelope of straight lines.

Neural Flow Diffusion Models improve diffusion models by learning flexible forward processes.

This paper explores the capability of deep neural networks to capture key characteristics of vehicle dynamics, and their ability to perform coupled longitudinal and lateral control of a vehicle. To this extent, two different artificial neural networks are trained to compute vehicle controls corresponding to a reference…

We study Fredholm properties and index formulas for Dirac operators over complete Riemannian manifolds with straight ends. An important class of examples of such manifolds are complete Riemannian manifolds with pinched negative sectional curvature and finite volume.

Smooth compactness theorem for elasticae, except straight segments.

Automated road infrastructure mapping using connected vehicle data and deep learning.

The paper shows caustics by reflection in projective Finsler metrics have at least four cusps.

The limit of energies of a sequence of harmonic maps as their annular domains approach the boundary of moduli space depends upon the boundary point approached. The infinite energy case is associated with limits of images containing ruled surfaces. The finite energy case yields a limit of images, under a suitable topolo…

To make deep neural networks feasible in resource-constrained environments (such as mobile devices), it is beneficial to quantize models by using low-precision weights. One common technique for quantizing neural networks is the straight-through gradient method, which enables back-propagation through the quantization ma…

Transformer models are compared to curved spacetime in General Relativity.

Topology of the Generic Hamiltonian Dynamical Systems on the Riemann Surfaces given by the real part of the generic holomorphic 1-forms, is studied. Our approach is based on the notion of Transversal Canonical Basis of Cycles (TCB). This approach allows us to present a convenient combinatorial model of the whole topolo…

Reintroduces straight-through estimators for binary neural networks.

Given a real vector space V of finite dimension, together with a particular homogeneous field of bivectors that we call a "field of projective forces", we define a law of dynamics such that the position of the particle is a "ray" i.e. a half-line drawn from the origin of V. The impulsion is a bivector whose support is …

Optimal schedules improve transport map approximation and learning.

In this paper, we investigate the ruled surfaces generated by a straight line according to rotation minimizing frame (RMF). Using this frame of a straight line, we obtained the necessary and sufficient conditions when the ruled surface is developable. Also, we give some new results and theorems related to be the asympt…

The Dirichlet Laplacian in curved tubes of arbitrary cross-section rotating with respect to the Tang frame along infinite curves in Euclidean spaces of arbitrary dimension is investigated. If the reference curve is not straight and its curvatures vanish at infinity, we prove that the essential spectrum as a set coincid…

Paper improves Gumbel-Softmax estimator variance reduction.

New method describes entanglement of straight lines in 3D space.

Proves cup product homomorphism for bounded cohomology on negatively curved manifolds.

Classifies branched Willmore spheres using conformal Gauss maps.

Characterizes corridors in loss surfaces for gradient-based optimization.

New method infers dynamical systems from population data.

New insights into quantized neural networks reveal learning dynamics and generalization errors.

In this paper, on the first, we prove $Δr=2H$ where $Δ$ is the Laplacian operator, $r=\left( r_{1},r_{2},r_{3}\right)$ the position vector field and $H$ is the mean curvature vector field of a surface $\mathcal{S}$ in the 3-dimensional Heisenberg group $H_{3}.$ In the second, we classify the ruled surfaces by straight…

The study identifies unique fluid flow patterns.

Sub-Riemannian geometry connects bike paths to mathematical curves.

A zero mean curvature surface in the Lorentz-Minkowski 3-space is said to be of Riemann-type if it is foliated by circles and at most countably many straight lines in parallel planes. We classify all zero mean curvature surfaces of Riemann-type according to their causal characters, and as a corollary, we prove that if …

New heat trace coefficients reveal curvature effects in polygonal domains.

Study knot diagrams on a sphere without vertical lines, focusing on minimal crossings.

The aim of this paper is to investigate properties preserved and co-preserved by coarsely $n$-to-1 functions, in particular by the quotient maps $X\to X/\sim$ induced by a finite group $G$ acting by isometries on a metric space $X$. The coarse properties we are mainly interested in are related to asymptotic dimension a…

The goal of this paper is to describe all local diffeomorphisms mapping a family of circles, in an open subset of $\r^3$, into straight lines. This paper contains two main results. The first is a complete description of the rectifiable collection of circles in $\r^3$ passing through one point. It turns out that to be r…

Straight lines are a basin of attraction for the elastic flow at least to level 1.9615π.

NM-PPG optimizes adaptive feature acquisition in POMDPs for better predictions.

The straight-line flow on almost every staircase and on almost every square tiled staircase is recurrent. For almost every square tiled staircase the set of periodic orbits is dense in the phase space.