User Guide

Chapter 12 Monte Carlo and sensitivity/worst-case analyses
284
Statistical analyses
Monte Carlo and sensitivity/worst-case are statistical
analyses. This section describes information common to
both types of analyses.
See Monte Carlo analysis
on page 12-289 for information
specific to Monte Carlo analyses, and see Worst-case
analysis on page 12-306 for information specific to
sensitivity/worst-case analyses.
Overview of statistical analyses
The Monte Carlo and worst-case analyses vary the lot or
device tolerances of devices between multiple runs of an
analysis (DC, AC, or transient). Before running the
analysis, you must set up the model and/or lot tolerances
of the model parameter to be investigated.
A Monte Carlo analysis performs a Monte Carlo
(statistical) analysis of the circuit. A worst-case analysis
performs a sensitivity and worst-case analysis of the
circuit.
Sensitivity/worst-case analyses are different from Monte
Carlo analyses in that they compute the parameters using
the sensitivity data rather than using random numbers.
You can run either a Monte Carlo or a worst-case analysis,
but you cannot run both at the same time. Multiple runs of
the selected analysis are done while parameters are
varied. You can select only one analysis type (AC, DC, or
transient) per run. The selected analysis is repeated in
subsequent passes of the analysis.
Generating statistical results
As the number of Monte Carlo or worst-case
runs increases, simulation takes longer and
the data file gets larger. Large data files
may be slow to open and slow to draw
traces.
One way to work around this is to set up an
overnight batch job to run the simulation
and execute commands. You can even set
up the batch job to produce a series of plots
on paper to be ready for you in the
morning.
Pspug.book Page 284 Wednesday, November 11, 1998 1:14 PM