1 / 33

MULTISCALE COMPUTATIONAL METHODS

MULTISCALE COMPUTATIONAL METHODS. Achi Brandt The Weizmann Institute of Science UCLA www.wisdom.weizmann.ac.il/~achi. Poisson equation:. given. Approximating Poisson equation:. given. Solving PDE : Influence of pointwise relaxation on the error. Error of initial guess.

daire
Download Presentation

MULTISCALE COMPUTATIONAL METHODS

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. MULTISCALE COMPUTATIONAL METHODS Achi Brandt The Weizmann Institute of Science UCLA www.wisdom.weizmann.ac.il/~achi

  2. Poisson equation: given Approximating Poisson equation: given

  3. Solving PDE: Influence of pointwise relaxation on the error Error of initial guess Error after 5 relaxation sweeps Error after 10 relaxations Error after 15 relaxations Fast error smoothingslow solution

  4. When relaxation slows down: the error is a sum of low eigen-vectors ELLIPTIC PDE'S (e.g., Poisson equation) the error is smooth The error can be approximated on a coarser grid

  5. h LhUh=Fh LU=F 2h L2hU2h=F2h 4h L4hU4h=F4h

  6. smooth Local Relaxation h LhUh = Fh approximation 2h R2h = ( Fh -Lh ) L2hU2h = F2h L2hV2h = R2h 4h L4hV4h = R4h

  7. h 2h . . . h0/4 h0/2 h0 * * * * interpolation (order m) of corrections Full MultiGrid (FMG) algorithm 1 4 4 4 2 4 4 4 3 multigrid cycle V interpolation (order l+p) to a new grid residual transfer enough sweeps or direct solver relaxation sweeps * algebraic error < truncation error

  8. Multigrid solversCost: 25-100 operations per unknown Linear scalar elliptic equation (~1971)

  9. Multigrid solversCost: 25-100 operations per unknown Linear scalar elliptic equation (~1971)* Nonlinear Grid adaptation General boundaries, BCs* Discontinuous coefficients Disordered: coefficients, grid (FE) AMG Several coupled PDEs* (1980) Non-elliptic: high-Reynolds flow Highly indefinite: waves Many eigenfunctions (N) Near zero modes Gauge topology: Dirac eq. Inverse problems Optimal design Integral equations Full matrix Statistical mechanics Massive parallel processing *Rigorous quantitative analysis (1986) (1977,1982) FAS (1975) Withinonesolver

  10. Multigrid solversCost: 25-100 operations per unknown Linear scalar elliptic equation (~1971)* Nonlinear Grid adaptation General boundaries, BCs* Discontinuous coefficients Disordered: coefficients, grid (FE) AMG Several coupled PDEs* (1980) Non-elliptic: high-Reynolds flow Highly indefinite: waves Many eigenfunctions (N) Near zero modes Gauge topology: Dirac eq. Inverse problems Optimal design Integral equations Statistical mechanics Massive parallel processing *Rigorous quantitative analysis (1986) (1977,1982) FAS (1975) Withinonesolver

  11. h 2h . . . h0/4 h0/2 h0 * * * * interpolation (order l+p) to a new grid interpolation (order m) of corrections residual transfer enough sweeps or direct solver relaxation sweeps * algebraic error < truncation error Full MultiGrid (FMG) algorithm

  12. smooth Local Relaxation h LhUh = Fh approximation Full Approximation scheme (FAS): U2h = + V2h 2h L2hV2h = R2h L2hU2h = F2h Fine-to-coarse defect correction 4h L4hU4h = F4h

  13. Multigrid solversCost: 25-100 operations per unknown Linear scalar elliptic equation (~1971)* Nonlinear Grid adaptation General boundaries, BCs* Discontinuous coefficients Disordered: coefficients, grid (FE) Several coupled PDEs* Non-elliptic: high-Reynolds flow Highly indefinite: waves Many eigenfunctions (N) Near zero modes Gauge topology: Dirac eq. Inverse problems Optimal design Integral equations Statistical mechanics Massive parallel processing *Rigorous quantitative analysis Withinonesolver

  14. h 2h . . . h0/4 h0/2 h0 * * * * interpolation (order l+p) to a new grid interpolation (order m) of corrections residual transfer enough sweeps or direct solver relaxation sweeps * algebraic error < truncation error Full MultiGrid (FMG) algorithm

  15. smooth Local Relaxation h LhUh = Fh approximation Full Approximation scheme (FAS): U2h = + V2h 2h L2hV2h = R2h L2hU2h = F2h Truncation error estimator Fine-to-coarse defect correction 4h L4hU4h = F4h

  16. Multigrid solversCost: 25-100 operations per unknown Linear scalar elliptic equation (~1971)* Nonlinear Grid adaptation General boundaries, BCs* Discontinuous coefficients Disordered: coefficients, grid (FE) AMG Several coupled PDEs* (1980) Non-elliptic: high-Reynolds flow Highly indefinite: waves Many eigenfunctions (N) Near zero modes Gauge topology: Dirac eq. Inverse problems Optimal design Integral equations Statistical mechanics Massive parallel processing *Rigorous quantitative analysis (1986) (1977,1982) FAS (1975) Withinonesolver

  17. Same fast solver FMG Local patches of finer grids • Each patch may use different coordinate system and anisotropic grid anddifferent physics; e.g. Atomistic

  18. h 2h . . . h0/4 h0/2 h0 * * * * interpolation (order m) of corrections Full MultiGrid (FMG) algorithm 1 4 4 4 2 4 4 4 3 multigrid cycle V interpolation (order l+p) to a new grid residual transfer enough sweeps or direct solver relaxation sweeps * algebraic error < truncation error

  19. Same fast solver FMG, Local patches of finer grids • Same fast solver FMG, FAS • Each level correct the equations of the next coarser level • Each patch may use different coordinate system and anisotropic grid

  20. r , s Finer level with local coordinate transformation With anisotropic further refinement x , y Boundary or tracked layer

  21. Same fast solver FMG, Local patches of finer grids • Same fast solver FMG, FAS • Each level correct the equations of the next coarser level • Each patch may use different coordinate system and anisotropic grid “Quasicontiuum” method [B., 1992] and differet physics; e.g.atomistic • Each patch may use different coordinate system and anisotropic grid anddifferent physics; e.g. Atomistic

  22. Multigrid solversCost: 25-100 operations per unknown Linear scalar elliptic equation (~1971)* Nonlinear Grid adaptation General boundaries, BCs* Discontinuous coefficients Disordered: coefficients, grid (FE) Several coupled PDEs* (1980) Non-elliptic: high-Reynolds flow Highly indefinite: waves Many eigenfunctions (N) Near zero modes Gauge topology: Dirac eq. Inverse problems Optimal design Integral equations Statistical mechanics Massive parallel processing *Rigorous quantitative analysis (1986)

  23. Stokes

  24. Compressible Navier-Stokes(on the viscous scale) h-principal L Central Cauchy-Riemann Central (Navier-) Stokes

  25. Multigrid solversCost: 25-100 operations per unknown Linear scalar elliptic equation (~1971)* Nonlinear Grid adaptation General boundaries, BCs* Discontinuous coefficients Disordered: coefficients, grid (FE) Several coupled PDEs* Non-elliptic: high-Reynolds flow Highly indefinite: waves Many eigenfunctions (N) Near zero modes Gauge topology: Dirac eq. Inverse problems Optimal design Integral equations Statistical mechanics Massive parallel processing *Rigorous quantitative analysis

  26. Multigrid solversCost: 25-100 operations per unknown Linear scalar elliptic equation (~1971)* Nonlinear Grid adaptation General boundaries, BCs* Discontinuous coefficients Disordered: coefficients, grid (FE) Several coupled PDEs* Non-elliptic: high-Reynolds flow Highly indefinite: waves Many eigenfunctions (N) Near zero modes Gauge topology: Dirac eq. Inverse problems Optimal design Integral equations Statistical mechanics Massive parallel processing *Rigorous quantitative analysis

  27. ALGEBRAIC MULTIGRID (AMG) 1982 Ax = b

  28. When relaxation slows down: the error is a sum of low eigen-vectors ELLIPTIC PDE'S the error is smooth DISCRETIZED PDE'S the error is smooth Along characteristics GENERAL SYSTEMS OF LOCAL EQUATIONS The error can be approximated by a far fewer degrees of freedom (coarser grid)

  29. ALGEBRAIC MULTIGRID (AMG) 1982 Ax = b Coarse variables - a subset Criterion: Fast convergence of “compatible relaxation” Relax Ax = 0 Keeping coarse variables = 0

  30. ALGEBRAIC MULTIGRID (AMG) 1982 Ax = b Coarse variables - a subset General procedures for deriving: * Interpolations * Restriction * Coarse-level equations Generalizations: 1. “General” linear systems 2. Variety of graph problems

  31. Graph problems Partition: min cut Clustering bioinformatics Image segmentation VLSI placement Routing Linear arrangement: bandwidth, cutwidth Graph drawing low dimension embedding

More Related