Catalog description: A unified introduction to the theory and practice of modern, near linear-time, numerical methods for large-scale partial differential and integral equations. Topics include: preconditioned iterative methods; generalized Fast Fourier Transform and other butterfly-based methods; multiresolution approaches including multigrid algorithms, hierarchical low-rank matrix decompositions, and low and high frequency Fast Multipole Methods. Example applications include: aircraft design, cardiovascular system modeling, electronic structure computation, and tomographic imaging.
Lectures: Monday/Wednesday 9:30-11:00 am in 2-136.
Office Hours: Monday and Wednesday at 4-5pm in 2-238C.
Prerequisites: This course covers advanced techniques for discretizing and solving PDEs. Some familiarity with ordinary differential equations, partial differential equations, Fourier transforms, linear algebra, and basic numerical methods for PDEs is assumed. It is strongly recommended that you have taken a previous course on basic numerical methods, such as 2.096/6.336/16.910, 2.097/6.339/16.920, 18.085, or 6.337/18.335. Problem sets will involve extensive coding: they may be completed in the language of your choice. Solutions and class demos will be in Julia.
Textbooks & Other Reading: Recommended reading will be posted as the class progresses. There is no textbook for the course, but the following books may be useful:
- Strauss "Partial Differential Equations: An Introduction". An advanced undergrad introduction to PDEs.
- Boyd "Chebyshev and Fourier Spectral Methods". Very readable and available online.
- Martinsson "Fast Direct Solvers for Elliptic PDEs". Modern and concise.
- LeVeque "Finite difference methods for ordinary and partial differential equations". (MIT access through SIAM.)
Grading: 50% problem sets (approximately biweekly), 50% final project report and presentation. Unless previous arrangements are made, late problem sets will be accepted for 1 week after the initial due date with a 50% penalty.
Collaboration Policy: Make a strong effort to solve problems on your own before discussing with any classmates. You must write up your own code and solutions, and indicate your collaborators on your assignments.
Problem sets will be posted here (approximately biweekly) and should be submitted on Gradescope.
- Problem Set 1 is due on Friday February 23, 2024 at 6pm (solutions).
- Problem Set 2 is due on Monday March 18, 2024 at 6pm (solutions).
- Problem Set 3 is due on Friday April 5, 2024 at 6pm (solutions).
- Problem Set 4 is due on Friday April 19, 2024 at 6pm.
The final project is a 10-12 page paper and a 15 minute presentation during the last week of classes. The final project is broad in scope, but must include the implementation of a fast algorithm in Python or Julia along with performance and error analyses. The project can take the form of a "literature review project" discussing a published algorithm or a "research project" attempting to implement a new solver for a problem of your choice.
One possibility is to review and implement an algorithm that was mentioned briefly or not covered in the course. Such a project should follow one or several published research papers describing the algorithm, along with a new implementation. Possible topics and suggested papers include:
- Advanced optimizations for production-level FFT implementations.
- Hierarchical Poincare-Steklov schemes using Chebyshev polynomials.
- Rectangular collocation with aliasing analysis.
- Alternating direction implicit scheme for multidimensional Chebyshev Poisson solvers.
- Bivariate orthogonal polynomials on triangles.
- Non-uniform Fast Fourier Transforms.
- Sparse spectral elements.
Another option is to use the methods covered in class to implement a fast solver for a research problem in your field. This should include:
- A discussion of the scientific problem and a brief derivation of the model PDE.
- Mention of the current commonly used methods in the field for the problem.
- A fast implementation of a new solver for the problem, or a related first-step.
- Discussion of the prospects and limitations of fast/high-order techniques for this problem in the future.
The goal should be producing a functional solver matching or improving on existing techniques in certain cases. However, it is understood that this may not turn out to be feasible (or even possible), and that's what makes it research! In that case, an implementation for a related toy model is expected, along with an analysis of the barriers to making a solver for the original problem.
Your report should be written in the style a SIAM article, using the SIAM article LaTeX templates. Your report should be between 10 and 12 pages. The SIAM layout has generous margins so this is not a lot of space. Make sure your presentation and notation is concise, but also has enough content and is not just padded out with images!
Along with your report, you must submit the code implementing your algorithms, preferably in the form of a Jupyter notebook. Make good use of headings, text, comments, and descriptive function/variables names. Good code is code that is easily understandable by others!
Your report and presentation should both include:
- Background information for your algorithm / physical application.
- A concise mathematical description of the algorithm you're using.
- Performance and error analysis for your implementation.
- Why do we need fast PDE solvers? Modeling, simulation, design, and inference.
- Solving PDEs on the computer: discretize, solve, and analyze.
- Finite differences: epsilon accuracy, at what cost?
- Symmetry and structure in a simple boundary value problem.
- What structure appears in the discretization and how do we leverage it?
- Locality and sparsity, translation invariance and Toeplitz matrices.
- Fourier modes and diagonalization.
Further reading: Finite difference approximation review, chapters 1 and 2 in LeVeque.
- The Discrete Fourier Transform (DFT) matrix.
- The Fast Fourier Transform (FFT) algorithm.
- Radix-2 FFT: polynomial evaluation, even-odd decomposition, and roots of unity.
Further reading: Johnson & Frigo "Implementing FFTs in Practice".
- Fast Fourier Transform (powers other than 2).
- Discrete Cosine and Sine Transforms (Type I).
- Fast FFT-based Poisson solvers (1D).
- Dirichlet boundary conditions and odd extensions.
- Neumann boundary conditions and even extensions.
- Circulant embedding for fast Toeplitz mat-vecs.
Further reading: Strang "The Discrete Cosine Transform".
- Poisson's equation in a 2D separable domain.
- Separable domains, vectorization, and Kronecker products.
- Fast transforms and solvers on separable domains.
- Complications: variable coefficients, time dependence, and nonlinearity.
- The role of fast BVP solvers in time-steppers and nonlinear iterative solvers.
- Iterative methods and preconditioning for BVPs.
Further Reading: Saad "Five Key Concepts That Shaped Iterative Solution Methods for Linear Systems".
- Domain decomposition techniques.
- Patching solutions for 1D and 2D Poisson problems.
- Block structure and Schur complements.
Further Reading: Martinsson "The Hierarchical Poincare-Steklov Solver for Elliptic PDEs".
- Hierarchical merging in domain decomposition.
- Mapped elements and meshes.
- Intro to spectral methods.
- The Fourier spectral method in 1D.
- Test space, trial space, and truncation errors.
- Convergence rates for smooth functions.
- Spectral methods with sine (cosine) basis.
- Nonperiodic problems with definite parity.
- 2D Fourier spectral methods.
- Chebyshev polynomials and series.
- Fast Chebyshev transforms.
- Chebyshev differentiation matrices.
- Polynomial interpolation.
- Approximation properties of interpolants.
- "Good" and "bad" grids: Chebyshev vs. equally spaced points.
- Spectral collocation matrices.
- Boundary bordering and rectangular projection.
- Nonlinear problems: Newton's iteration and Frechet derivatives.
Further Reading: Driscoll and Hale "Rectangular Spectral Collocation" and Aurentz and Trefethen "Block Operators and Spectral Discretizations".
- Chebyshev "T" and "U" bases.
- Sparse Chebyshev "T"->"U" differentiation matrix.
- Fast solves with "almost-banded" discretizations.
- Ultraspherical bases.
Further Reading: Olver and Townsend "A Fast and Well-Conditioned Spectral Method".
- Multiplication matrices in ultraspherical bases.
- When are multiplication matrices "banded"?
- Pseudospectral methods for nonlinear IVPs.
- Multidimensional sparse spectral methods.
- Block structure for coupled ODE.
- Block structure for PDE discretizations.
Further Reading: Vasil et. al. "Tensor Calculus in Polar Coordinates Using Jacobi Polynomials".
- The challenge of multiple nonperiodic variables.
- Fast spectral methods for Poisson equation in separable domains.
- ADI and factored ADI for Sylvester equations.
Further Reading: Fortunato and Townsend "Fast Poisson Solvers for Spectral Methods" and Townsend and Wilber "On the Singular Values of Matrices with High Displacement Rank".
- Low-rank matrices and functions.
- Storage and computation with low-rank functions.
- Integral operators with low-rank kernels.
Further Reading: Townsend and Trefethen "Continuous Analogues of Matrix Factorizations".
