Running Statistical Software on Statlab's Linux Machines

Statlab UNIX machine: NOTE The old Unix server known as biscu.its.yale.edu has been replaced by a new Linux server.

The new Linux server called statapp.yale.edu has SAS, SPlus and Stata. To apply for an account on the new Statlab Linux machine, send an email to stathelp@yale.edu or call 432-3277.

Connecting to statapp.yale.edu:
Once you have an account, you can logon via SSH or kerborized telnet. If you are going to run a X session then you have to connect using SSH.

SSH

Use SSH software on campus networked computers or on your home computer with a Yale PPP connection.
SSH gives you a plain character based interface.
Note there are options for running the software in batch mode.

X Terminal

An X terminal session provides an interface similar to Windows.
All of the Windows machines in the StatLab (main Lab and Annex) have software loaded so that you can use them as X terminals. Please read the X session information.

X compatible software:

SAS	Start with: sas -fsdevice X11
S-Plus	Start with: Splus Once in S-Plus open the graphic device with X11()
STATA	Enter: setenv STATAPD "b /usr/local/stata/pd.X" Start with: stata

UNIX	Task	DOS
ls	short file listing	dir /w
ls -l	long file listing	dir
cat	display contents of file	type
more	display contents of a file one page at a time	type | more
cp	copy a file	copy
mv	move or rename a file	move or ren
rm	delete a file	del
mkdir	create a directory	md
cd	change directory	cd
pwd	display which directory you are in	cd
lp	print file to default printer, which is Statlab's printer EVEN	print
man	help on UNIX commands	help
du -s -k	display used space in Kb of current directory including all subdirectories, use it to check if you are close to your quota limit	N/A
tcsh	a C-shell, which allows recalling and editing commands (and more!)	doskey

Obviously most commands above have some parameters, like the filename.

ATTENTION: Unix is case sensitive!

Handling Batch Jobs One of the advantages of using UNIX as a platform for statistical analysis is, that you can run a job for a long time without tying up a terminal. To do this, you have to run the job in background . Then you get your %-prompt back and you can log out while your job keeps running.
You can run any UNIX command in background by appending an &.

Running Statistic Software in Batch

SAS	sas myprog.sas &
Stata	stata -b myprog &

In order to run software in batch you need to create a file containing the commands, i.e. myprog.sas for a SAS program. You can use your favorite editor on the PC or Mac and upload the file using SSH . If you want to use an editor on the SUN, you can use pico or emacs.

To check on a program you are running in background

ps -f -u NETID	shows all your processes. Replace NETID with your NETID,
kill -9 PID	kills your program. You get the PID from the ps listing,
jobs	lists of the program(s) you have started in the current shell
fg %x	returns a job to foreground. Replace x with the number you got from jobs

 

To check on the load of the machine

You are not the only person using statapp. Before you launch another computing intensive job, check how busy the machine already is.

uptime	shows the userload. If the load is 3, you can't expect more than 30% of the CPU time for your process (25% if you launch an additional job).

---

Certifying Authority: Ann Green, Director Social Science Statistical Laboratory

lm: February 28, 2002