plots the error incurred by using machine arithmetic to evaluate the expression f in a small neighborhood of a.


plots the error from a-n ulps to a+n ulps.

Details and Options

  • To use MicroscopicErrorPlot, you first need to load the Computer Arithmetic Package using Needs["ComputerArithmetic`"].
  • The scale on the vertical axis represents the error in ulps.
  • The following option can be given:
  • JoinedTruewhether to join points
  • Valid settings for Joined are True, False, and Real.
  • With the setting Joined->Real, the line joining the plot points represents the mapping from the set of real numbers to the error in the machine-number output.