Using Gauss

General Gauss References:


Useful commands for working with data are (see command reference for exact usage):

load - to load data
miss,missrv -- replace missing value with number and vice versa
packr -- remove all rows of matrix with missing values

A Reference Guide/Manual with commands can be found at: http://www.aptech.com/
(click on resource library - the command reference and training manual are both useful)

  • Student version** (UT only -- you must connect to this page through a UofT account)

  • Graduate Labs with Gauss at UofT : According to Pete Onge, IT Administrator, GAUSS is installed on all of the computers in both of the graduate labs, the existing one on the 14th floor of Robarts (accessible to MFE students as well as the senior PHDs); and the soon-to-be-available grad lab in the basement of Sidney Smith Hall. Information will follow shortly, but Pete Onge anticipates that the lab will be up by end of week (or whenever keys to the space become available through the Planning and Information Technology department).

  • Yale StatLab's Gauss Introduction

  • Felix Ritchie's "Gauss - A Beginner's Guide"

  • Mark Watson's GAUSS tutorial

  • Eric Zivot's GAUSS Resources

  • Gauss links from the University of Wisconsin
  • Maximo Alonso's time series programs
  • Free optimization routine ) and example file (by Eugene Choo)


    • A sample basic regression program created for last year's class


    (some links seemed to have changed. If you find the new link please let me know.)

    Using gauss on unix


    Use putty to access the account. To enter the gauss program type: tgauss at the command line.
    From here you can run commands at the prompt. This is good for learning the commands. For real work you will want to type your commands into a program. To do this type ed example.gss at the gauss prompt where example.gss is the the name of the file. This will bring you into the emacs editor. To quit and save type control-X control-C. Then at the gauss prompt type run example.gss to run it. Better yet you can run and edit it directly from the command line with out going into gauss as follows:

    To run the program divyield.prg type

    nice --adjustment=19 tgauss -b divyield.prg

    If the program only takes a minute or two to run you can just use

    tgauss -b divyield.prg

    To run it a second time just use the up arrow to get the same command back again.

    To edit a divyield.prg just use emacs divyield.prg.

    You probably want two versions of putty open at the same time, one to run gauss. The other to edit the file.

    To run gauss without having to keep a window open, you can type following command where here I assume fx.prg is the name of the program you are running:

    nice --adjustment=19 nohup tgauss -b fx.prg &

    Since this is to long for me to remember. I usual copy this into a file in the same directory where fx.prg sits using emacs. Then I save the file as say "r" or any other file name. Then on the command line I type

    chmod u+x r

    which allows the file to be executed. All this only needs to be done once. Then each time you want to run the program you just type

    ./r

    from the command line in the directory where fx.prg is in.

    To see if the program is still running you would type

    top

    Here is a link with more and explanation and details on job control

    Miscellaneous Gauss Commands

    To set a seed to create reproducible simulation results: use:

    rndseed 2006; @where 2006 could be any number@

    at the top of the program. The rndseed command is documented on p. rndnseed is documented on p. 28-668 of the Gauss 7.0 manual.) Then just use the usual

    rndn(n,k);

    command to create normal random variables.

  • Documentation on viewing Gauss tkf graphics files. readmefile