CS450 Calendar
Note: Some calendar entries are clickable and link to entries below.
Lecture 28
(May 1, 2019, 1 p.m. - May 1, 2019, 2:15 p.m.)
Topics:
- Boundary value problems
- Shooting method: implementation
- Non-uniqueness for nonlinear BVPs
- Finite difference methods
- Collocation method
- Weighted residual/Galerkin/Finite element methods, weak derivatives
- Solving large, sparse linear systems
- Coordinate format, CSR format
- Stationary iterative methods for linear systems
- Conjugate Gradient Method
Lecture 27
(April 26, 2019, 1 p.m. - April 26, 2019, 2:15 p.m.)
Topics:
- Predictor/Corrector Methods: Heun's method
- Runge-Kutta methods
- Butcher tableaus
- Implicit/explicit/diagonally explicit
- High-order vs low order, efficiency
- Dissipation properties, relationship to stability region
- Multi-step methods
- Stability regions for RK methods
- Cost metrics, trade-offs, more advanced methods (IMEX, Multi-Rate, Runge-Kutta-Chebyshev)
- Boundary Value Problems: Intro, General formulation
- Shooting method
Lecture 26
(April 24, 2019, 1 p.m. - April 24, 2019, 2:15 p.m.)
Topics:
- Forward and Backward Euler methods
- Global and local error, order of accuracy for IVP solvers
- Stability of IVP solvers
- Stability regions
- Forward and backward Euler: stability properties
- Stability for systems and nonlinear ODEs
- Unconditional stability
- Stiff ODEs
- Examples, definitions, and computational treatment
- Accuracy and Stability concerns
Lecture 25
(April 19, 2019, 1 p.m. - April 19, 2019, 2:15 p.m.)
Topics:
- Initial value problems
- Introduction, implicit and explicit methods
- Examples of IVPs
- Properties: linear, homogeneous, constant coefficient
- Existence, uniqueness, sensitivity
- Transformation to first-order form
- Transfomration to autonomous form
- Stability and asymptotic stability
- Forward Euler and experiments on its stability
Lecture 24
(April 17, 2019, 1 p.m. - April 17, 2019, 2:15 p.m.)
Topics:
- Composite quadrature: Setup, error estimate
- Numerical differentiation
- Conditioning, noise sensitivity, cancellation
- Deriving finite difference formulas via Taylor
- Deriving finite difference formulas via interpolation
- Scaling and shifting FD formulas
- Error estimates for numerical differentiation
- Richardson extrapolation
Lecture 23
(April 12, 2019, 1 p.m. - April 12, 2019, 2:15 p.m.)
Topics:
- Quadrature: Introduction, Existence/Uniqueness
- Interpolatory quadrature, weights, nodes
- Clenshaw-Curtis, Newton-Cotes
- Method of undetermined coefficients
- Midpoint, Trapezoidal, Simpson's rules
- Degree of exactness of Newton-Cotes quadature rules, odd-order exactness increase
- Order of accuracy estimate based on interpolation
- Stability of Quadrature
- Gaussian quadrature: Introduction
Lecture 22
(April 10, 2019, 1 p.m. - April 10, 2019, 2:15 p.m.)
Topics:
- Interpolation Error Identity
- Connection between error result and Chebyshev best-approximation
- Asymptotic error estimates for interpolation
- Piecewise interpolation and splines
Lecture 21
(April 5, 2019, 1 p.m. - April 5, 2019, 2:15 p.m.)
Topics:
- Interpolation: Setup, problem statement, purpose, error model
- Interpolation: Existence, Uniqueness, Sensitivity, Lebesgue Constant
- Runge and Gibbs phenomena, advantages of edge-clustered nodal sets
- Monomial interpolation and its conditioning issue
- Lagrange and Newton interpolation, complexity of interpolation
- Orthogonal Polynomials: Legendre, Chebyshev
- Chebyshev interpolation: basis evaluation through recurrence, nodes, Vandermonde matrix
Lecture 20
(April 3, 2019, 1 p.m. - April 3, 2019, 2:15 p.m.)
Topics:
- Levenberg-Marquardt as a least squares problem
- Equality-constrained optimization
- Lagrange Multipliers
- Inequality-constrained optimization: Lagrangian, KKT conditions, computational approach
Lecture 19
(March 29, 2019, 1 p.m. - March 29, 2019, 2:15 p.m.)
Topics:
- Newton review
- Quasi-Newton methods, BFGS
- Nonlinear least squares
- Gauss-Newton method, Levenberg-Marquardt
- Equality-constrained optimization, Necessary conditions
Lecture 18
(March 27, 2019, 1 p.m. - March 27, 2019, 2:15 p.m.)
Topics:
- Sensitivity of optimization in $n$ dimensions
- Unimodality, Golden Section Search
- Newton's method in one dimension
- Steepest Descent in $n$ dimensions, convergence
- Extrapolation-based methods, momentum, heavy ball method, relationship to CG
- Newton's method in $n$ dimensions
Lecture 17
(March 15, 2019, 1 p.m. - March 15, 2019, 2:15 p.m.)
Topics:
- Fixed point iteration in $n$ dimensions, convergence criteria
- Newton's method in $n$ dimensions
- Broyden's method
- Optimization: Problem statement, definitions
- Global/local minima, coercivity, existence
- Convexity, uniqueness of minima
- Necessary/sufficient conditions for a minimizer, Hessians
- Sensitivity of optimization (1D)
Lecture 16
(March 13, 2019, 1 p.m. - March 13, 2019, 2:15 p.m.)
Topics:
- Fixed point iteration recap
- Newton's method, convergence properties
- Secant method, convergence properties
- Muller's method, inverse quadratic interpolation
- Trust region methods, hybrid methods
Lecture 15
(March 8, 2019, 1 p.m. - March 8, 2019, 2:15 p.m.)
Topics:
- Rates of convergence recap
- Stopping criteria
- Bisection Method
- Convergence Analysis for fixed point iteration
Lecture 14
(March 6, 2019, 1 p.m. - March 6, 2019, 2:15 p.m.)
Topics:
- Arnoldi and Lanczos Iteration: Recap and Computation
- Computation of the SVD
- Nonlinear equations: Problem statement, existence, sensitivity
- Rates of Convergence
Lecture 13
(March 1, 2019, 1 p.m. - March 1, 2019, 2:15 p.m.)
Topics:
- Computational Expense in QR iteration, Householder similarity transform
- Krylov space methods
- Orthogonalization in Krylov space methods
- Incremental computation in Krylov space methods
- Arnoldi Iteration
Lecture 12
(Feb. 27, 2019, 1 p.m. - Feb. 27, 2019, 2:15 p.m.)
Topics:
- Power iteration: convergence
- Rayleigh quotient
- Normalized Power Iteration, Inverse iteration, Rayleigh quotient iteration
- Cost of power iteration variants
- Orthogonal iteration
- QR iteration, without and with shift
Lecture 11
(Feb. 22, 2019, 1 p.m. - Feb. 22, 2019, 2:15 p.m.)
Topics:
- Eigenvalues: geometric and algebraic multiplicity
- Diagonalizability and counterexamples
- Similarity transforms and eigenvalue preservation
- Eigenvalue transformations: shift, inversion, power, polynomial, similarity
- Bauer-Fike eigenvalue sensitivity bound
- Power Iteration: Idea, Convergence, Failure Modes
Lecture 10
(Feb. 20, 2019, 1 p.m. - Feb. 20, 2019, 2:15 p.m.)
Topics:
- SVD: Introduction
- Computing 2-norms, 2-norm condition numbers
- Rank and numerical rank
- Low-rank approximation, Eckart-Young-Mirsky theorem
- Moore-Penrose pseudoinverse, SVD solution to total least squares
- Relative cost of LU, QR, SVD
- Eigenvalues: Introduction, spectrum, spectral radius
- Eigenvalues: Computability
Lecture 9
(Feb. 15, 2019, 1 p.m. - Feb. 15, 2019, 2:15 p.m.)
Topics:
- Householder QR
- Givens QR
- Total Least Squares (Problem Statement)
- SVD: Definition
- SVD: Properties (rough overview, to be repeated)
Lecture 8
(Feb. 13, 2019, 1 p.m. - Feb. 13, 2019, 2:15 p.m.)
Topics:
- QR for Least Squares
- Gram-Schmidt
- Existence of QR, economical and full
- Modified Gram-Schmidt
- Intro to Householder
Lecture 7
(Feb. 8, 2019, 1 p.m. - Feb. 8, 2019, 2:15 p.m.)
Topics:
- Least squares
- Polynomial data fitting
- Normal equations
- Existence and Uniqueness for LSQ
- LSQ: Geometric interpretation, connection to orthogonal projection
- Pseudoinverse and condition number of non-square full-rank matrices
- Sensitivity and Conditioning of LSQ
Lecture 6
(Feb. 6, 2019, 1 p.m. - Feb. 6, 2019, 2:15 p.m.)
Topics:
- LU with Pivoting
- Computational cost, finding inverses, solving with multiple right-hand sides
- BLAS/LAPACK
- LU Existence for non-invertible, non-square
- Reduced LU
- Schur complements
- LU and Roundoff error
- Sherman-Morrison
Lecture 5
(Feb. 1, 2019, 1 p.m. - Feb. 1, 2019, 2:15 p.m.)
Topics:
- Residual and error in linear systems
- Matrix change bounds in linear systems
- Fw/Bw substitution
- Elimination matrices
- LU with Pivoting (intro)
Lecture 4
(Jan. 25, 2019, 1 p.m. - Jan. 25, 2019, 2:15 p.m.)
Topics:
- Implementing FP arithmetic
- Catastrophic Cancellation
- Matrix norms
- Matrix condition numbers
Examlet 0
(Jan. 24, 2019 - Jan. 26, 2019)
As you may have seen in our class policies, our "examlets" will take place in a computer-based testing facility ("CBTF") in Grainger Library.
You must schedule your test appointment with the Computer-Based Testing Facility at this link. This examlet is now available for scheduling.
Find out more about the testing facility, such as:
- where it is
- when to show up
- what to bring (and not to bring)
The exam will be 110 minutes in length, and it will cover the material of the first two lectures, plus linear algebra basics and Python coding basics.
Lecture 3
(Jan. 23, 2019, 1 p.m. - Jan. 23, 2019, 2:15 p.m.)
Topics:
- Fixed point
- Floating point
- Subnormals, IEEE FP
- Rounding rules
- UFL, OFL, machine epsilon, rounding error
Lecture 2
(Jan. 18, 2019, 1 p.m. - Jan. 18, 2019, 2:15 p.m.)
Topics:
- Forward/backward error
- Conditioning
- Example: Conditioning of function evaluation
- Accuracy and stability
Lecture 1
(Jan. 16, 2019, 1 p.m. - Jan. 16, 2019, 2:15 p.m.)
- Introduction
- Class Policies
- Truncation and Rounding
- Well-posedness
- Norms