Fast solvers (MATH 2011)

Fall 2020
Mike O'Neil (oneil@cims.nyu.edu)
Lecture

Wednesdays 11:00am - 12:50pm
ONLINE

Office hours
TBD, or by appointment

Announcements
  • No class on Sep 9
  • Please join the course page on Campuswire. Email the professor for the passcode.
Description

The course will cover what are known as "fast solvers", algorithms for solving the linear systems arising from the discretization of elliptic PDEs.

Materials

The main textbook for the course is Martinsson, Fast Direct Solvers for Elliptic PDEs, 2019. The textbook is available for free (electronically) to NYU affiliates at siam.org.

Grading

The final grade in the course will be based on a course project and textbook presentations/discussions.

Schedule

Below is an updated list of discussion topics along with any documents that were distributed, notes, or relevant code.

Date Topics Materials
Sep 2 Introduction to the course Martinsson, Ch 1-2
Notes
Recording
Sep 9 NO CLASS
Sep 16 Low-rank approximation
Randomized linear algebra
Martinsson, Ch 3-4
Slides
Recording
Sep 23 Fast algorithms for rank-structured matrices Martinsson, Ch 5
Slides
Recording
Sep 30 HODLR complexity
The fast multipole method
Martinsson, Ch 6
HODLR slides
Notes
Recording
Oct 7 The fast multipole method Martinsson, Ch 7-8
Notes
Recording
Oct 14 Integral equations: Continuous theory Martinsson, Ch 10-11
Notes
Recording
Oct 21 Integral equations: discretization Martinsson, Ch 12
Notes
Comments
Recording
Oct 28 FDS for IE: Multilevel scheme Martinsson, Ch 13-14
Notes
Recording
Nov 4 FDS for IE: HBS error estimation, ID, skeletonization Martinsson, Ch 15-16
Nov 11 FDS for IE: Efficient compression Martinsson, Ch 17
Notes
Recording
Nov 18 FDS for IE: Scattering matrices Martinsson, Ch 18
Notes
Recording
Nov 25 FDS for PDE: Nested dissection Martinsson, Ch 19-20
Recording
Dec 2 FDS for PDE: Fast sparse solvers Martinsson, Ch 21-22
Ch 21 slides
Ch 22 slides
Recording
Dec 9 Project presentations Recording