Altair RISE

A program to recognize and reward our most engaged community members

Nominate Yourself Now!

EDEM video tutorial issues

I want to use the quality in the statistical domain of EDEM as a response in Hyperstudy, and then optimize the particle parameters. The official staff gave me a video tutorial, and I watched it. It was indeed what I wanted, but I followed along and couldn't generate a CSV file to generate the response. This is the video link: https://youtu.be/TXeqwkpr82o?si=IxlsexZOdlF9fIsQ ；Between the tenth and eleventh minutes, he generated the CSV after testing, but I followed suit and did not generate it;
Figure 1: After completing the modeling in Hyperstudy within the video, a CSV file is generated after running the test;

So based on the above purpose and tutorial, I found a function in EDEM that generates Bin attributes during the simulation calculation process, as shown in Figure 3. I tested and found that after the simulation was completed, this function could generate CSV files. However, even after updating the EDEM settings in my Hyperstudy, I still couldn't generate the CSV format file I wanted, so my Hyperstudy optimization couldn't continue.
Figure 2: Use of CSV function generated during simulation calculation process

Adding this setting to Hyperstudy still does not generate CSV

In addition, when testing this feature, I found that there is an issue with the quality export. The data exported for total and total over time are the same, displaying the quality in the Bin, which is completely different from the total quality in the Bin that I exported for the total time. Is there a problem with this feature?
Figure 3: The two exported data are the same

Figure 4: The results of using Total over time in Bin are different

Find more posts tagged with

EDEM

HyperStudy

Accepted answers

Stephen Cole

Hi,

Regarding point 1 did you add the two lines to the EDEMpy script mentioned here https://community.altair.com/discussion/comment/195011

import matplotlib
matplotlib.use('TkAgg')

YOu don't need to neccesseraly change the name of the script, so long as the name of your script matches the one in the hyperstudy connector settings. However you will likely have to edit the script to include the above two lines.

for point 2 i've raised this as a bug with the product team. The 'total over time' is giving the same results as 'total' if using the Simulator Queries. You could go to the EDEM ANalyst > File > Export Results data and setup the same query, if you set it up there it will give you the correct 'total over time' value.

The differnce is the simulator queries get written out live as the simulation runs. The analyst queries get written out on completion of the simulation.

Regards

Stephen

All comments

Jin1

So, how can I generate responses for the Bin attributes of EDEM in Hyperstudy?

Stephen Cole

Hi,

Regarding point 1 did you add the two lines to the EDEMpy script mentioned here https://community.altair.com/discussion/comment/195011

import matplotlib
matplotlib.use('TkAgg')

The differnce is the simulator queries get written out live as the simulation runs. The analyst queries get written out on completion of the simulation.

Regards

Stephen

Jin1

Thank you again. Following your method, I added these two lines of code to the script and deleted the EDEMPY folder. The entire process is now normal, and the data in BIN was also generated into a CSV file after the testing model was completed. I have disturbed you many times. Thank you again

Jin1

Hello, sorry to bother you again. I have another question about EDEM and Hyperstudy. I have asked it on the forum before, but still haven't received a good response. I think you are an expert in using EDEM, so I would like to learn from you. When optimizing EDEM in Hyperstudy, I chose the GRSM method, which may require more than 20 calculations. I have two EDEM GPU licenses and am solving two tasks in parallel. So, should I choose different GPUs for each subsequent task? Assuming that I use GPU1 for task 1 and want to use GPU2 for task 2, when task 1 and task 2 are completed, task 3 will continue to use GPU1, task 4 will continue to use GPU2, and so on, can this be set in Hyperstudy?

Stephen Cole

Hi, I don't believe there is a solution for this currently, it's set as a feature request.

If you setup EDEM to use 2 GPU's it will always run each job on 2 GPU's regardless of the HyperStudy settings. So if you used multi-execution = 2 it would run both jobs simultaneously, but both jobs would use 2 GPUs.

If you selected 1 GPU for EDEM and multi-execution = 2 it will run each job on 1 GPU, and this will always be the same GPU. There isn't any functionality currently, within the HyperStudy connector, to submit the jobs to the least loaded GPU.

You could get HyperStudy to 'write input files' but de-select the 'execute analysis and extract output responses' options. Then you could use a batch script to run half the jobs on 1 GPU and half on another (the -D flag sets which GPU is used. Running a simulation with -D 0 runs on the first GPU in the list, -D 1 runs on the second). The results could then be imported back into HyperStudy on completion. It works for a pre-defined DoE type run but not if there is any optimisation of parameters.

https://help.altair.com/edem/topics/edem_simulator/running_edem_in_batch_mode_t.htm

Jin1

Thank you, I understand the method you mentioned. Thank you for your answer