Computing Notes

A somewhat random collection of computing notes related to the Unix and PC platforms, and various software packages. If you have material to contribute please email Stat Web Master .

Unix Platform
(including software on Unix)		 PC Platform
(including software on PC)

Software
(general information)

Unix

Resources

  • The Unix lab in Hutcheson 408 has 19 computers on which you can run jobs:

    • anise mint coriander tarragon paprika nutmeg ginger sage thyme onion dill
      rosemary marjoram cinnamon curry cilantro allspice clove garlic

  • Also, there are two multiprocessor computers available:

    • oven - 2 processors

    • broiler - 4 processor

Documentation and help

Assessing the Unix machines

The Unix computers are accessible remotely only via Secure Shell (ssh). See the systems administrator to obtain a copy of Secure Shell for PC.

Etiquette

  • While these Unix computers will support multiple jobs competing for resources, the best performance will be gained when only one compute-intensive job run. So, once connected to the computer use the "top" command to determine if another job is running.

  • Give your jobs low priority via "renice 19".

  • For multiprocessor computers, use only half of the available processors, i.e., run no more than 2 jobs on broiler.

Running batch jobs in Unix

  • UnixPrompt% nohup R CMD BATCH filename & | | | +-- run job in background +-- allow job to run after logout

    The process number (PID) will be displayed.

  • Give your jobs low priority:

    UnixPrompt% renice 19 PID

Example: Running R batch jobs in Unix

  • Log into (perhaps remotely) to the Unix machine of choice.

  • In the directory where your R source code file is saved as 'filename':

    Unix Prompt% nohup R CMD BATCH filename &  

  • Output will be saved in this same directory (unless specified in source code). In addition a .Rout log-file will be created. Alternately, the command can be executed from any of your directories and output will be created in that same directory. Just make sure the filename includes the path to your source code file.

Editors

XEMACS: http://www.gnu.org/software/emacs/emacs.html is an editor on the Unix machines. If you use emacs in conjunction with Emacs Speaks Statistics, you can run R from within emacs. Two quick start guides (1 page PDF) are available: Emacs R & ESS.

Documentation and help

File management tool

R

jpeg2eps

A Unix tool for converting JPEG (JPG) images to encapsulated postscript (EPS). The syntax is

jpeg2eps [-v] [-c] [-d] [file [file [file …] ] ]

PC

Resources

The PC lab in Hutcheson 401 has 10 computers available for your use.

LaTeX

  • MikTeX: http://www.miktex.org/ is one version of Latex.

  • Instructions for installing packages in MikTeX and changing the package repository in MikTeX are found in the PDF files:

    Note that these instructions are for PCs and may change with version.

  • Packages can be found online (Google search) or you can configure MikTeX to automatically install them if they're needed. The best way (IMHO) to learn LaTeX is to borrow a template from a more experienced user and play!

Editors

WINEDT: http://www.winedt.com/ this is a Windows based text editor. This makes typesetting in latex much easier, particularly for a beginner as there is a GUI math symbol menu. It's fairly cheap for a lifetime license and pretty easy to learn. R also has a package called RWinEdt that provides an interface between R and WinEdt. This will allow you to edit R code in WinEdt and send it directly to R (without copy-paste). This has the advantage of your code being formatted which makes editing easier.

Software

LaTeX

R

Basics on R can be obtained in this PDF document. Also see the editors section in either Unix or PC platforms and the documents at http://www.stat.vt.edu/software/R.doc/.