ParFUM: High-level Adaptivity Algorithms for Unstructured Meshes

Slides:



Advertisements
Similar presentations
Steady-state heat conduction on triangulated planar domain May, 2002
Advertisements

Lists and the Collection Interface Chapter 4. Chapter Objectives To become familiar with the List interface To understand how to write an array-based.
Collections Chapter Java Collection Frameworks The Java collection framework is a set of utility classes and interfaces. Designed for working with.
Surface Simplification Using Quadric Error Metrics Speaker: Fengwei Zhang September
Resource Management §A resource can be a logical, such as a shared file, or physical, such as a CPU (a node of the distributed system). One of the functions.
Dave Bradley Rick Harper Steve Hunter 4/28/2003 CoolRunnings.
Designing Tensile Structures Using Generic CAD Applications. Structural membranes 2007, Barcelona, September 2007 Javier Sánchez, Tecnun, University.
Section 4: Implementation of Finite Element Analysis – Other Elements
A Bezier Based Approach to Unstructured Moving Meshes ALADDIN and Sangria Gary Miller David Cardoze Todd Phillips Noel Walkington Mark Olah Miklos Bergou.
A Bezier Based Approach to Unstructured Moving Meshes ALADDIN and Sangria Gary Miller David Cardoze Todd Phillips Noel Walkington Mark Olah Miklos Bergou.
CSE351/ IT351 Modeling And Simulation Choosing a Mesh Model Dr. Jim Holten.
Filling Arbitrary Holes in Finite Element Models 17 th International Meshing Roundtable 2008 Schilling, Bidmon, Sommer, and Ertl.
A Load Balancing Framework for Adaptive and Asynchronous Applications Kevin Barker, Andrey Chernikov, Nikos Chrisochoides,Keshav Pingali ; IEEE TRANSACTIONS.
CSE554SimplificationSlide 1 CSE 554 Lecture 7: Simplification Fall 2014.
ParFUM Parallel Mesh Adaptivity Nilesh Choudhury, Terry Wilmarth Parallel Programming Lab Computer Science Department University of Illinois, Urbana Champaign.
Chapter 2 The process Process, Methods, and Tools
Grid Generation.
Parallelization Of The Spacetime Discontinuous Galerkin Method Using The Charm++ FEM Framework (ParFUM) Mark Hills, Hari Govind, Sayantan Chakravorty,
1 Data Structures for Scientific Computing Orion Sky Lawlor charm.cs.uiuc.edu 2003/12/17.
7 th Annual Workshop on Charm++ and its Applications ParTopS: Compact Topological Framework for Parallel Fragmentation Simulations Rodrigo Espinha 1 Waldemar.
Dynamic Meshing Using Adaptively Sampled Distance Fields
Adaptive Mesh Modification in Parallel Framework Application of parFUM Sandhya Mangala (MIE) Prof. Philippe H. Geubelle (AE) University of Illinois, Urbana-Champaign.
Triangular Mesh Decimation
23-Oct-15 Abstract Data Types. 2 Data types A data type is characterized by: a set of values a data representation, which is common to all these values,
Application Paradigms: Unstructured Grids CS433 Spring 2001 Laxmikant Kale.
Supercomputing ‘99 Parallelization of a Dynamic Unstructured Application using Three Leading Paradigms Leonid Oliker NERSC Lawrence Berkeley National Laboratory.
NIH Resource for Biomolecular Modeling and Bioinformatics Beckman Institute, UIUC NAMD Development Goals L.V. (Sanjay) Kale Professor.
CSE554SimplificationSlide 1 CSE 554 Lecture 7: Simplification Fall 2013.
Zibin Zheng DR 2 : Dynamic Request Routing for Tolerating Latency Variability in Cloud Applications CLOUD 2013 Jieming Zhu, Zibin.
Efficient RDF Storage and Retrieval in Jena2 Written by: Kevin Wilkinson, Craig Sayers, Harumi Kuno, Dave Reynolds Presented by: Umer Fareed 파리드.
InterConnection Network Topologies to Minimize graph diameter: Low Diameter Regular graphs and Physical Wire Length Constrained networks Nilesh Choudhury.
CSE554Fairing and simplificationSlide 1 CSE 554 Lecture 6: Fairing and Simplification Fall 2012.
Fault Tolerant Extensions to Charm++ and AMPI presented by Sayantan Chakravorty Chao Huang, Celso Mendes, Gengbin Zheng, Lixia Shi.
1 1 What does Performance Across the Software Stack mean?  High level view: Providing performance for physics simulations meaningful to applications 
Ale with Mixed Elements 10 – 14 September 2007 Ale with Mixed Elements Ale with Mixed Elements C. Aymard, J. Flament, J.P. Perlat.
LLNL-PRES DRAFT This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under contract.
STW Transfer Center for Geometry Processing and CAGD flipper 2.0 An interactive modular geometry processing toolkit.
Partitioning using Mesh Adjacencies  Graph-based dynamic balancing Parallel construction and balancing of standard partition graph with small cuts takes.
Domain decomposition in parallel computing Ashok Srinivasan Florida State University.
Solid Modeling 2002 A Multi-resolution Topological Representation for Non-manifold Meshes Leila De Floriani, Paola Magillo, Enrico Puppo, Davide Sobrero.
BOĞAZİÇİ UNIVERSITY – COMPUTER ENGINEERING Mehmet Balman Computer Engineering, Boğaziçi University Parallel Tetrahedral Mesh Refinement.
Parallelization Strategies Laxmikant Kale. Overview OpenMP Strategies Need for adaptive strategies –Object migration based dynamic load balancing –Minimal.
1 Data Structures for Scientific Computing Orion Sky Lawlor /04/14.
General requirements for BES III offline & EF selection software Weidong Li.
1 Rocket Science using Charm++ at CSAR Orion Sky Lawlor 2003/10/21.
Predictive Load Balancing Using Mesh Adjacencies for Mesh Adaptation  Cameron Smith, Onkar Sahni, Mark S. Shephard  Scientific Computation Research Center.
Scalable Dynamic Adaptive Simulations with ParFUM Terry L. Wilmarth Center for Simulation of Advanced Rockets and Parallel Programming Laboratory University.
Examples (D. Schmidt et al)
Model Optimization Wed Nov 16th 2016 Garrett Morrison.
Designing High Performance BIRT Reports
Subject Name: File Structures
Parallel Graph Algorithms
Interactive Offline Tracking for Color Objects
Parallel Unstructured Mesh Infrastructure
Morphing and Shape Processing
User-Guided Simplification
Unstructured Grids at Sandia National Labs
Performance Evaluation of Adaptive MPI
structures and their relationships." - Linus Torvalds
More Graph Algorithms.
Component Frameworks:
Iterative Optimization
Efficient Document Analytics on Compressed Data: Method, Challenges, Algorithms, Insights Feng Zhang †⋄, Jidong Zhai ⋄, Xipeng Shen #, Onur Mutlu ⋆, Wenguang.
SAMANVITHA RAMAYANAM 18TH FEBRUARY 2010 CPE 691
Yi Wang, Wei Jiang, Gagan Agrawal
Presents: Rally To Java Conversion Suite
Low Order Methods for Simulation of Turbulence in Complex Geometries
Ph.D. Thesis Numerical Solution of PDEs and Their Object-oriented Parallel Implementations Xing Cai October 26, 1998.
Parallel Implementation of Adaptive Spacetime Simulations A
structures and their relationships." - Linus Torvalds
Presentation transcript:

ParFUM: High-level Adaptivity Algorithms for Unstructured Meshes Terry L. Wilmarth, Nilesh Choudhury, Isaac Dooley, Sayantan Chakravorty Parallel Programming Laboratory University of Illinois at Urbana-Champaign

Adaptivity Goals for ParFUM Work with applications scientists to discover realistic adaptivity needs Develop rich set of algorithms for flexible mesh adaptivity interface Easy to use: minimal extra work for user Fast, efficient, scalable Modular, extensible implementation

Example: 1D Wave Propagation

Mesh Sizing Many adaptivity operations strive to achieve a certain mesh sizing Specified in mesh edge length Set at mesh elements Mesh sizing can be used to: Direct refinement and coarsening Maintain regional sizing Maintain a reference mesh

Mesh Sizing, cont'd Setting mesh sizing: FEM_ADAPT_SetElementSizeField(i nt meshID, int elem, double size) FEM_ADAPT_SetElementsSizeField( int meshID, double *sizes) Allows user to set sizing directly, however, adaptivity operations support special sizing techniques

Example: Flexible Sizing

Example: Flexible Sizing

Example: Flexible Sizing

Example: Flexible Sizing

Example: Flexible Sizing

Refinement and Coarsening Simple algorithms guided by mesh sizing requirements and quality measures Refinement: longest edge bisection (Rivara) Coarsening: shortest edge contraction

Refinement and Coarsening, cont'd An element is added to a list of elements to be refined/coarsened if: Its average edge length is not within some tolerance of the sizing field Its quality is low These elements are sorted by: Longest/shortest edge weighted appropriately by quality measure

Refinement and Coarsening, cont'd Adaptation proceeds in passes After a pass, elements are re- evaluated Refinement adds new elements which must be examined in the next pass Coarsening modifies shape of existing elements Both algorithms may skip adapting certain elements (locking, inversion, quality)

Refinement and Coarsening, cont'd API: FEM_ADAPT_Refine(int qm, int method, double factor, double *sizes) FEM_ADAPT_Coarsen(int qm, int method, double factor, double *sizes) Method allows for modifying mesh to a uniform size, to use mesh sizing, to perform scaling, or combinations of these

Example: Remeshing/Repair

Example: Remeshing/Repair

Example: Remeshing/Repair

Example: Remeshing/Repair

Example: Remeshing/Repair

Example: Remeshing/Repair

Mesh Quality These simple algorithms can be used strategically to improve mesh quality: Coarsening removes elements, particularly skinny isosceles triangles Refinement makes flat slivers into skinny isosceles that can be later removed by coarsening

Data Transfer Automatic data transfer built in to low-level mesh modification operations Linear interpolation of nodal data Duplication of element data for bisection Framework for user to override/specify data transfer mechanisms Provides mechanism to push element data to nodes and pull it back again

Ongoing Efforts Translation of current functionality into 3D Remesh/Repair, Smoothing, etc. Element-to-element data transfer Support for adaptivity in “thicker” ghost layers Load balancing and dynamic repartitioning