No video available
Please refer to the materials section for this topic.
Floating-Point Arithmetic and Numerical Error
Learning Objectives
- •IEEE-754 representation, machine epsilon, ulp, and overflow
- •Roundoff error, cancellation, and catastrophic cancellation
- •Forward error, backward error, and conditioning
Weekly Outcome
By the end of this module you will be able to construct vectorized numerical experiments, quantify floating-point and discretization error, and produce reproducible diagnostic plots with dimensionally consistent physical quantities.