Talk:Finite Arithmetic

This page is just a skelleton. I will add content as time permits. The main purpose of putting this up is to define what goes where in the track Numerical Algorithms.

Goals

edit

After finishing this course, students should:

  • Know how floating-point numbers are represented, specifically as defined by IEEE 754.
  • Know what eps, realmin, realmax and denormalized numbers are.
  • Know the common problems in using finite-precision arithmetic and know how to avoid them when possible.

Take-Home Message

edit
  • Numbers have a finite relative precision.
  • Floating-Point numbers may cause avoidable errors.
  • Special care has to be taken to avoid such errors.
Return to "Finite Arithmetic" page.