Nonlinear Programming

Fall 2007

MATP6600 / DSES6780

• Course outline.
• Material on reserve in the library.
• Grades through the final exam, sorted by total score. You can collect your final exam from me.
• Homeworks:
  • Homework 1 (pdf file).
  • The solutions to homework 1 are now available.
  • Homework 2 (pdf file).
  • The solutions to homework 2 are now available.
  • Homework 3.
  • The solutions to homework 3 are now available.
  • Homework 4. Information about AMPL and a link to the NEOS server.
  • The solutions to homework 4, questions 1-4, are now available. Here are the model file, data file, run file, and results file for questions 5 and 6.
  • Homework 5.
  • The solutions to homework 5 are now available.
  • Homework 6.
  • Solutions to homework 6: A model file, data file, and output file for question 1. Solutions to questions 2 and 3.
• Exams:
  • The Final Exam will be in class on Friday, December 7. It will cover the whole course. You will be able to bring one sheet of handwritten notes.
  • The solutions to the final exam from 2004 are available. The mean on this exam was 48%, with a high score of 79% and a low of 25%.
  • The final exam from 1999 is available. The mean on this exam was 73.5%. Here is the solution to a corrected question 4.
  • The solutions to the midterm are now available.
  • Midterm exam.
• Notes: These are notes for the algorithmic portion of the course from the Fall of 2004.
  • Introduction to algorithms, including algorithmic maps.
  • Algorithms for unconstrained problems.
  • Algorithms for constrained problems.
• Handouts:
  • Linear algebra.
  • Subspaces, affine spaces, convex sets, and cones.
  • Dimension, polyhedra, and faces.
  • A simplex iteration.
• Papers and resources:
  • An introduction to conic programming, by my former graduate student Kartik Sivaramakrishnan.
  • The NEOS Guide to Optimization Software contains a brief descriptions of the SQP package SNOPT by Philip Gill et al.
  • Another SQP method is filterSQP by Sven Leyffer and Roger Fletcher, who have also written an introduction to filter methods.
  • Comparing four good nonlinear programming solvers: Assessing the Potential of Interior Methods for Nonlinear Optimization, by Jorge Nocedal et al., the team behind KNITRO.
  • Papers on LOQO by Robert Vanderbei et al include this and this.
  • A painless introduction to conjugate gradients, by JR Shewchuk.
  • Convex Optimization, a downloadable text by Stephen Boyd and Lieven Vandenberghe.