Feb 15, 2002 ------------ - Introduction to numerical integration - complement of the numerical differentiation problem - Simple ideas from Monte Carlo integration - Picasso paintings - Basic integration rules - Trapezoidal rule - exact for linear - Rectangle rule - exact for constant - Composite integration rules - subdivide given region into smaller pieces - apply rules separately - add their results together - General format for integration rules - linear combination of function evaluations - so, function evaluations are key - Getting lower and upper bounds on integrals - Worked out Example 1 from your textbook - Can there be magic forthcoming? - Exact integration rules for linear, quadratic, and cubic polynomials - To evaluate Int[-1,1,f(x)], use f(-1/sqrt(3)) + f(1/sqrt(3)) - Exact integration rules for polynomials of degree <= 5 - use three function evaluations - at -sqrt(3/5), 0, and sqrt(3/5) - will explain how magic works later (in a future class) - Reminder: Test 1 next Friday Feb 18, 2002 ------------ - Recall - rectangle rules - trapezoidal rules - magic rules - Riemann-integrable functions - relationship to rectangle rules - Composite trapezoidal rule - rearranging summations to be more efficient about function evaluations - error in integration using trapezoids - Worked out example 1 from textbook Feb 20, 2002 ------------ - Started on Romberg quadrature - Building up the Romberg array - pattern of entries in first column - recursively calculating entries in second and other columns - Relationship between columns and error - Revise material for Exam 1 Feb 22, 2002 ------------ - Exam 1 Feb 25, 2002 ------------ - Return Exam 1 - statistics of scores - discuss questions and answers - Romberg quadrature - recurrence formula - improving order of convergence for composite trapezoid formulas - Basic intro to adaptive quadrature Feb 27, 2002 ------------ - Adaptive quadrature - Simpson's 3-point rule - where does this come from? - how it can be used to recursively construct quadrature rules for subintervals - accurate upto cubic polynomials - When to stop adaptive quadrature calculations - error estimation from successive integral evaluations Mar 1, 2002 ------------ - Recap quadrature formulas - nodes and weights - Creating our own quadrature formula - worked out example 1 on page 230 - Comparison of 3-point rules - Our formula: accurate upto quadratics - Simpson: accurate upto cubics - Gaussian quadrature: accurate upto quintics! - (magic rules) - How Gaussian quadrature works - 3-point rule - 2-point rule - How nodes are determined for Gaussian rules - roots of Legendre polynomials - Homework 5 will be assigned today - due after the break