The University of Missouri at Columbia
Chemistry 433 - Computational Chemistry - Fall Semester 2002



Darwin is a Compaq cluster of 8 ES40 processors. Darwin is the main research computing facility on the MU Campus. The system was installed in 2001 with 4 processors and 4 more processors were added in 2002 with funding from RSC (2 processors) and NIH (2 processors).

Setup to Run "newzmat" in /dfs/tmp

Users who wish to run the newzmat utility need to include the following
two lines in their .login file:

   setenv g98root /usr/local
   source $g98root/g98/bsd/g98.login

That sets up the environment variables that are necessary to run

Darwin User Instruction

For PC:   
Use Secure CRT from the MU Internet Access Package
Connect to

For PC or Mac:
Logon to shiva, use your own username and password
Type:  ssh -l your_account {return}
Type:  your_password {return} 

Disconnecting: Ctrl + D

Creating Jobs: Make normal input file (.com file) as you do on shiva. You should specify the name of the chk, rwf, int and d2e files before the command line. The directory to which all of these files should be directed is /dfs/tmp. A sample code for this purpose is: %rwf=/dfs/tmp/bd1 %d2e=/dfs/tmp/bd1 %int=/dfs/tmp/bd1 %NoSave %chk=/dfs/tmp/bd1 #B3LYP/6-311G** opt=Z-matrix B3LYP/6-311G** opt of 1,4-diphenylbutadiene 0 1 C C, 1, cc1 C, 2, cc2, 1, ccc1 C, 3, cc3, 2, ccc2, 1, 0.0 ... After you create the .com file, you need to create a so-called PBS file. The file name has no extension, e.g. water, bd, benzene, etc. Use pico to edit that file. You need to type: run_g98 where is the .com file you prepared in step 1. Save the PBS file and exit pico. Now you are ready to run a job!

Submitting Job: Type: qsub -q g98 file in which file is the PBS file you created in the previous step. By default, the job will run for 8 hours of CPU time. If your job needs more time, instead of typing qsub -q g98 -l file {return} type: qsub -q g98-l cput=12:00:00 file {return} and in this case the job will run for 12 hours. *IMPORTANT: g98 is the only queue where you can submit your jobs. There can be only one job running on darwin at each time. If you know your job takes less than 8 hours on shiva, please use shiva. Darwin is NOT a replacement for shiva. It is supposed to be used for those large and huge jobs. And PLEASE DO NOT queue ten jobs at a time such that other people have to wait for a week go get onto the machine.

Other Commands: To check your job and get job ID: qstat To terminate a job: qdel job ID (the same as you do in shiva) For all other questions about the commands of PBS, type: man PBS To check Gaussian98 sample test jobs, type: cd /usr/sd/software/Gaussian98/g98/tests/com and then browse the directory with ls command. * You are supposed to be able to perform all of the default Gaussian 98 operations. If there is any problem, please contact Dr. Larry Sanders.

Retrieving: No FTP client works for Darwin because of system security reasons. To retrieve your files from Darwin, you have to log onto Darwin first. Then you need to Get into the folder which the file you are transferring is located at; Type: ftp {return} {Type in your user name and password for shiva} Type: bin {Get into the folder where you want to save the transferring file} Type: put filename.extension {return} The file should be transferred from Darwin to shiva. Now you can do whatever you want with the file in shiva.