2.096J / 6.336J / 16.910J Introduction to Numerical Simulation
Fall 2005
Professors: Luca Daniel, LIM Kian Meng, Jacob K White, Collin M Stultz
TAs: Kin Cheong Sou, Bradley Neil Bond, Ngoc Son Nguyen
Lecture: MW EVE ( 8PM - 9.30PM ) (1-390)
Information:
Introduction to Numerical simulation is an introduction to computational techniques for the simulation of a large variety of engineering and physical systems. Applications are drawn from aerospace, mechanical, electrical, and chemical engineering, biology, and materials science. Topics include mathematical formulations; network problems; sparse direct and iterative matrix solution techniques; Newton methods for nonlinear problems; discretization methods for ordinary, time-periodic and partial differential equations, fast methods for partial differential and integral equations, techniques for dynamical system model reduction and approaches for molecular dynamics.
Announcements
CDO Seminar
MIT Distinguished Speaker Series in COMPUTATION for DESIGN and OPTIMIZATION(CDO)TITLE: “Solving Very Large-Scale Convex Optimization Problems: A Return to Subgradient Type Methods”
SPEAKER: Aharon Ben-Tal, MINERVA Optimization Center, Faculty of Industrial Engineering and Management (currently visiting MIT Sloan School of Management)
DATE: Wednesday, December 14
TIME: 4-5pm
PLACE: MIT room 1-390
ABSTRACT: The need to solve extremely large-scale (104-106 variables) convex optimization problems, such as those arising in medical imaging, shape design of mechanical structures and more, enforces one to reconsider simple gradient-type algorithms as the only methods of choice, since more sophisticated algorithms (such as interior point) require in a single iteration a number of arithmetic operations that are prohibitively large. We propose in this talk a new subgradient-type method, the “Non-Euclidean Restricted Memory Level (NERML) method, ” for minimizing large-scale nonsmooth convex over “simple” domains. The characteristic features of NERML are: (a) The possibility to adjust the scheme to the geometry of the feasible set, thus allowing essentially dimension-independent rate of convergence, which are nearly optimal in the information complexity sense; (b) flexible handling of accumulated information, allowing for trade-off between the level of utilization of this information and the iterations’ complexity.
Announced on 22 November 2005 11:58 a.m. by Luca Daniel
Watching lecture videos from outside campus
If you are attempting to watch the videos from a location outside of the MIT network, you need to use the MIT VPN to do it. The software you need is available on the following sites:For Windows:
https://wserv.mit.edu/fcgi-bin/softcount?product=mit/win/mit-vpn-4.6.exe
For Macintosh: href="https://wserv.mit.edu/fcgi-bin/softcount?product=mit/macos/mit-vpn-4.7.dmg
More information is at: http://web.mit.edu/ist/services/network/vpn.html
Announced on 13 October 2005 3:07 p.m. by Luca Daniel
Computation for Design and Optimization Distinguished Speakers Seminar
Click here for the seminar scheduleAnnounced on 07 October 2005 2:28 p.m. by Luca Daniel
