Closed AmyMcCall closed 1 year ago
Hi Amy,
The createODB function uses the default eigenvalue solver which has been problematic in the updated OpenSees. For now, you can change the solver to fullGenLapack to make it work.
@u-anurag Please provide a MWE showing how the default eigen solver is problematic now but was not problematic in a previous version of OpenSees. This is the first I've heard of the default eigen solver presenting new problems.
@mhscott my comment was purely based on the discussion I saw on the Facebook group. I'll run my models with the updated OpenSees to see if I get the same behavior. @AmyMcCall did it work with the fullGenLapack solver?
@u-anurag I don't recall any discussions about issues with the eigen command due to an updated version of OpenSeesPy. If you find something though, let me know.
@mhscott I was already running the fullGenLapack when I directly called the eigen function (see attached image. Also note, I could not get the eigen function to run for all modes without the fullGenLapack in the older version of Openseespy), but I did not include it when I called the createODB function (which later calls the eigen function, see my code in the attached image). The difference now is that I can't run the createODB with all modes included, where I could in the previous version of Openseespy. Should I include fullGenLapack in the createODB function like this: vfo.createODB(FileLoc, EigenLCName, numModes, '-fullGenLapack')?
@AmyMcCall I am not familiar with how vfo
interacts with the core OpenSees, so I will leave that to @u-anurag. Also, I am not aware of any new issues with the eigen
command in OpenSees.
@AmyMcCall Would you mind sharing the script so I can see what is happening?
@u-anurag, the script is pretty complex with several sub-routines and sub-folders. Is there a way I can get you the folder and I can step you through which file to open and run?
@AmyMcCall You can share via Google Drive or Dropbox. We can schedule a Zoom/G-meet call if that suits you.
I am happy to schedule a meeting. What works for you?
I am available at 12:30 pm (pacific time) tomorrow. Let's take this conversation to email. Send me an invite on iitg.anurag@gmail.com.
u-anurag,
I updated my program (which I was running in Openseespy v. 3.3.0.1) to use the vfo function instead of the postprocessing.get_Rendering commands. I am using vfo.createODB but was getting the error shown in the attached figure (createODBError). One of the variables read in were the number of modes, in this case there are 8. I played around with this number and was finally able to get vfo.createODB to work by using 3 modes instead of 8. Why doesn't this vfo command work with the actual number of modes as it used to with the postprocessing.get_Rendering command in the older version? This may be an issue with the openSees commands since it is getting hung up on ops.eigen(Nmodes+1).
Amy