Obviously, performance is a dominant issue. Algorithmic complexity analysis is extremely important.

Performance is also heavily influenced by how data is accessed and how the

**memory hierarchy**(disk, RAM, cache, registers) is used.**Mathematical software libraries**containing efficient, pre-compiled modules are widely used.-
Another important issue is accuracy. Real arithmetic is guaranteed to be "inexact" because it uses

**finite precision**to represent and operate on real numbers which may require infinite precision. The**stability**of algorithms (i.e., their sensitivity to the small changes caused by**roundoff error**) must be studied. - What algorithms lie under the surface of a numerical method?

CS1104 Main Page

Last Updated 01/05/2000

© L.Heath, 2000