Ubuntu
arpack package

multiple invocations of arpack give different results?

Asked by nic-stange

Hi,

I'm trying to debug a problem with scipy and while doing so I discovered some very strange arpack-behaviour: I've taken the snsimp.f from EXAMPLES/SIMPLE and modified it to carry out the computation twice. The problem: The results are slightly different.

The test program ftp://ftp.zmaw.de/outgoing/stange/snsimp.f creates two files: snsimp_1st.out, snsimp_2nd.out with the output of the two runs.
I've compiled with gfortran -o snsimp snsimp.f -larpack on lucid lynx.

Do you've got any hints? I'm completely stuck.

Thank you very much!

Nicolai

Question information

Language:
English Edit question
Status:
Solved
For:
Ubuntu arpack Edit question
Assignee:
No assignee Edit question
Solved by:
nic-stange
Solved:
Last query:
Last reply:
Revision history for this message
nic-stange (nic-stange) said : 		#1

I've found the answer by myself: If you don't give ARPACK's dnaupd a starting vector resid (indicated by setting info to 1), it'll create a random one.