High Performance Linear Programming (P02376)

? Credit Points : 10 ? SCQF Level : 11 ? Acronym : MAT-P-HPLP

Linear programming (LP) is the fundamental model used in optimal decision making. In many industries practical LP problems can have very large numbers of decision variables and constraints, so efficient solution methods are essential. It is also common for practical problems to have inherent structure, the most common being the Block-Structured Linear Programming (BDLP) model. It is valuable to develop specialist solution methods to exploit this and other related structures. This course will study such structured LP problems in the context of application areas where they arise, and methods for exploiting their structure. Computational techniques for implementing these methods, in particular those suitable for high performance computing, will also be studied in detail.

Entry Requirements

none

Subject Areas

Home subject area

Operational Research, (School of Mathematics, Schedule P)

Delivery Information

? Normal year taken : Postgraduate

? Delivery Period : Semester 2 (Blocks 3-4)

? Contact Teaching Time : 10 hour(s) per week for 2 weeks

All of the following classes

Type	Day	Start	End	Area
Lecture	Sunday	09:00	09:50	KB

Summary of Intended Learning Outcomes

1. Understanding how structured LP problems occur in practice.
2. Understanding of common LP structures.
3. Understanding of solution methods that exploit LP structure
4. Understanding of computational techniques for implementing methods for structured LP problems, in particular techniques for high performance computing.