James A. Sethian

ORCID: 0000-0002-7250-7789
Publications
Citations
Views
---
Saved
---
About
Contact & Profiles
Research Areas
  • Computer Graphics and Visualization Techniques
  • Advanced Numerical Methods in Computational Mathematics
  • Fluid Dynamics and Heat Transfer
  • Lattice Boltzmann Simulation Studies
  • Fluid Dynamics and Turbulent Flows
  • Medical Image Segmentation Techniques
  • Computational Fluid Dynamics and Aerodynamics
  • Computational Geometry and Mesh Generation
  • Advanced Numerical Analysis Techniques
  • Seismic Imaging and Inversion Techniques
  • 3D Shape Modeling and Analysis
  • Fluid Dynamics and Thin Films
  • Advanced X-ray Imaging Techniques
  • Medical Imaging Techniques and Applications
  • Fluid Dynamics Simulations and Interactions
  • Machine Learning in Materials Science
  • Solidification and crystal growth phenomena
  • Geometric Analysis and Curvature Flows
  • Electron and X-Ray Spectroscopy Techniques
  • Fluid Dynamics and Vibration Analysis
  • Electrohydrodynamics and Fluid Dynamics
  • Scientific Computing and Data Management
  • Meteorological Phenomena and Simulations
  • Computational Physics and Python Applications
  • Surface Modification and Superhydrophobicity

Lawrence Berkeley National Laboratory
2015-2024

University of California, Berkeley
2015-2024

Research Applications (United States)
2016-2023

University of California System
1988-2021

Applied Mathematics (United States)
1991-2021

Office of Science
2021

Alive Hospice
2020

Bay Area Medical Center
2013

New York University
1985-2008

Bureau of Economic Analysis
2008

Shape modeling is an important constituent of computer vision as well graphics research. models aid the tasks object representation and recognition. This paper presents a new approach to shape which retains some attractive features existing methods overcomes their limitations. The authors' techniques can be applied model arbitrarily complex shapes, include shapes with significant protrusions, situations where no priori assumption about object's topology made. A single instance model, when...

10.1109/34.368173 article EN IEEE Transactions on Pattern Analysis and Machine Intelligence 1995-01-01

A fast marching level set method is presented for monotonically advancing fronts, which leads to an extremely scheme solving the Eikonal equation. Level methods are numerical techniques computing position of propagating fronts. They rely on initial value partial differential equation a function and use borrowed from hyperbolic conservation laws. Topological changes, corner cusp development, accurate determination geometric properties such as curvature normal direction naturally obtained in...

10.1073/pnas.93.4.1591 article EN Proceedings of the National Academy of Sciences 1996-02-20

Fast Marching Methods are numerical schemes for computing solutions to the nonlinear Eikonal equation and related static Hamilton--Jacobi equations. Based on entropy-satisfying upwind fast sorting techniques, they yield consistent, accurate, highly efficient algorithms. They optimal in sense that computational complexity of algorithms is O(N log N), where N total number points domain. The use a variety applications, including problems shape offsetting, distances from complex curves surfaces,...

10.1137/s0036144598347059 article EN SIAM Review 1999-01-01

10.1006/jcph.1995.1098 article EN Journal of Computational Physics 1995-05-01

The Fast Marching Method is a numerical algorithm for solving the Eikonal equation on rectangular orthogonal mesh in O ( M log ) steps, where total number of grid points. In this paper we extend to triangulated domains with same computational complexity. As an application, provide optimal time computing geodesic distances and thereby extracting shortest paths manifolds.

10.1073/pnas.95.15.8431 article EN Proceedings of the National Academy of Sciences 1998-07-21

▪ Abstract We provide an overview of level set methods, introduced by Osher and Sethian, for computing the solution to fluid-interface problems. These are computational techniques that rely on implicit formulation interface, represented through a time-dependent initial-value partial-differential equation. discuss essential ideas behind techniques, coupling these finite-difference methods incompressible compressible flow, collection applications including two-phase ship hydrodynamics,...

10.1146/annurev.fluid.35.101101.161105 article EN Annual Review of Fluid Mechanics 2002-12-18

We present a fast algorithm for solving the eikonal equation in three dimensions, based on marching method. The is of order O(N log N), where N total number grid points computational domain. can be used any orthogonal coordinate system and globally constructs solution to each point method unconditionally stable solutions consistent with exact arbitrarily large gradient jumps velocity. In addition, resolves overturning propagation wavefronts. begin mathematical foundation using follow...

10.1190/1.1444558 article EN Geophysics 1999-03-01

We introduce a new limiting system of equations to describe combustion processes at low Mach number in either confined unbounded regions ~d numerically solve these for the case fiame propagating closed vessel.This allows large heat release, substantial temperature and density variations, interaction with hydrodynamic ftow field, including effects turbulence.However, this is much simpler than complete equation~ compressible reacting gas fiow since detailed acoustic waves have been...

10.1080/00102208508960376 article EN Combustion Science and Technology 1985-01-01

10.1007/bf01210742 article EN Communications in Mathematical Physics 1985-12-01

10.1016/0021-9991(92)90229-r article EN Journal of Computational Physics 1992-06-01

We develop a family of fast methods for approximating the solutions to wide class static Hamilton--Jacobi PDEs; these include both semi-Lagrangian and fully Eulerian versions. Numerical problems are typically obtained by solving large systems coupled nonlinear discretized equations. Our techniques, which we refer as "Ordered Upwind Methods" (OUMs), use partial information about characteristic directions decouple systems, greatly reducing computational labor. techniques considered in context...

10.1137/s0036142901392742 article EN SIAM Journal on Numerical Analysis 2003-01-01

Deep convolutional neural networks have been successfully applied to many image-processing problems in recent works. Popular network architectures often add additional operations and connections the standard architecture enable training deeper networks. To achieve accurate results practice, a large number of trainable parameters are required. Here, we introduce based on using dilated convolutions capture features at different image scales densely connecting all feature maps with each other....

10.1073/pnas.1715832114 article EN Proceedings of the National Academy of Sciences 2017-12-26

The Fast Marching Method is a numerical algorithm for solving the Eikonal equation on rectangular orthogonal mesh in O(M log M) steps, where M total number of grid points. scheme relies an upwind finite difference approximation to gradient and resulting causality relationship that lends itself Dijkstra-like programming approach. In this paper, we discuss several extensions technique, including higher order versions unstructured meshes Rn manifolds connections more general static...

10.1073/pnas.090060097 article EN Proceedings of the National Academy of Sciences 2000-05-16

In many physical problems, interfaces move with a speed that depends on the local curvature.Some common examples are flame propagation, crystal growth, and oil-water boundaries.We idealize front as closed, nonintersecting, initial hypersurface flowing along its gradient field curvature.Because explicit solutions seldom exist, numerical approximations often used.In this paper, we review some recent work algorithms for attacking these problems.We show based direct parametrizations of moving...

10.4310/jdg/1214444092 article EN Journal of Differential Geometry 1990-01-01

10.1016/0021-9991(92)90140-t article EN Journal of Computational Physics 1992-02-01

We review recent work on level set methods for following the evolution of complex interfaces. These techniques are based solving initial value partial differential equations functions, using borrowed from hyperbolic conservation laws. Topological changes, corner and cusp development, accurate determination geometric properties such as curvature normal direction naturally obtained in this setting. The methodology results robust, accurate, efficient numerical algorithms propagating interfaces...

10.1017/s0962492900002671 article EN Acta Numerica 1996-01-01
Coming Soon ...