asdeasoft.net

Dear STKO Team:

I'm a new bird for using STKO.

When I learned the lessons from your official website, I found the example can't show the total model on the postprocessing page.

I wonder why this appeared.


Thanks!

It seems you're using parallel analysis with a partitioned model.
If STKO encounters an issue with one of the MPCO databases (of your partitioned model) it will write something in the STKO's terminal (error messages in red).
Can you copy-paste here what it says?

Thanks for your reply!

I got the warning info below:
-----------------------------------------------------------------------------------------------------
Warning : partitioned model, "Numberer" (in AnalysesCommand) different from "Parallel Reverse Cuthill-McKee Numberer" attribute, there may be some problems
Warning : partitioned model, "system" (in AnalysesCommand) different from "Mumps" or "ParallelProfileSPD" attribute, there may be some problems
Warning : partitioned model, "Numberer" (in AnalysesCommand) different from "Parallel Reverse Cuthill-McKee Numberer" attribute, there may be some problems
Warning : partitioned model, "system" (in AnalysesCommand) different from "Mumps" or "ParallelProfileSPD" attribute, there may be some problems
-----------------------------------------------------------------------------------------------------

However, after I made relative changes the results didn't converge. I also post the info below:

---------------------------------------------------------------------------------------
ERROR: the analysis did not converge
while executing
"error "ERROR: the analysis did not converge""
("for" body line 30)
invoked from within
"for {set STKO_VAR_increment 1} {$STKO_VAR_increment <= $initial_num_incr} {incr STKO_VAR_increment} {

# before analyze
STKO_CALL_OnBeforeAnalyze
..."
(file "analysis_steps.tcl" line 222)
invoked from within
"source analysis_steps.tcl"
(file ".\main.tcl" line 81)
Process Terminating 0
------------------------------------------------------------------------------------------

Lastly, I wanna know if there is any method to output complete results with the TCL recorder command when I use parallel computing.

So it seems you were running a parallel analysis with OpenSeesMP without using the proper settings.
Please use the ParallelRCM numberer and the MUMPS system in all your Analysis commands

Oops!
I forgot to change the first analysis setting. And now It works.

By the way, because of the inconvenience of extracting chart data in postprocessor, I decide to output data by tcl commands. However, If I run the parallel model, the data files only contain a part of the data rather than the complete content. Is there any way to record the complete data in one file under the parallel model by tcl commands?

Thanks!

By the way, because of the inconvenience of extracting chart data in postprocessor, I decide to output data by tcl commands. However, If I run the parallel model, the data files only contain a part of the data rather than the complete content. Is there any way to record the complete data in one file under the parallel model by tcl commands?

No, that's one of the drawback of plain recorders.
However you can use STKO Python API to automate the extraction of data from partitioned databases.

Otherwise, if you want to use plain recorders, you must create them one per process. Something like this:
recorder Node -file "ANodeRecorder_[getPID].txt" ....
where the [getPID] function will return the ID of the current processor.
So, in this way you will have N files, one for each process.
Then you can do some scripting to merge them

Thanks!

Could you please tell me how to find and use the Python API?

I tried to add a customCommand which you mentioned "recorder Node -file "ANodeRecorder_[getPID].txt" ".
The results showed something was wrong, and I hope you can help me explain it.

I appreciate your reply!

The error is saying that you forgot to:

• Add the node or list of nodes
• Add the dof
• Add the result

Please refer to the recorder documentation:
https://opensees.berkeley.edu/wiki/inde ... e_Recorder

Thanks!

I tried as you pointed out with a command " recorder Node -file "ANodeRecorder_[getPID].txt" -time -node 1 2 -dof 2 disp; ".
It did show two files with the name "ANodeRecorder_0.txt" and "ANodeRecorder_1" because I divided the model into two parts.

However, the two files can't make up to a complete one. One contains three columns and another only has one.

What's wrong with it?

I appreciate your reply!

It's correct:
You asked for time + Uy of nodes 1 and 2.
The first column is the time (in both files).
The other 2 columns are only in the second files because those nodes (1 and 2) are only on the second partition