article

Matthew Gillespie posted Nov 8, '21 Jordan Johnson commented Apr 3, '23

FlexSim 2022 Beta available

FlexSim 2022 Beta is now available. (Updated November 23)

To get the beta, log in to your account at https://account.flexsim.com, then go to the Downloads section, and click on More Versions. It will be at the top of the list.

The More Versions button does not appear when logged in as a guest account. Subsequently the beta is available only to licensed accounts and accounts that have a license shared with them. Learn more about downloading the best version of FlexSim for your license here.

If you have bug reports or other feedback on the software, please email [email protected] or create a new idea in the Development space.

FlexSim 22.0.0 Release Notes

Added several new features to the Experimenter including:
- Experimenter can define and run multiple different jobs.
- Experimenter results are stored in a database file.
- Additional replications of a scenario can be run without needing to rerun previous replications.
- Improved visuals for status of replications.
Added a Reinforcement Learning tool.
Added Label charts.
Added Tracked Variable charts.
Standardized charts created by Pin buttons.
Added sort options to table views.
Updated MTBF/MTTR time fields to show unit conversion options.
Added InObjectsInternal and OutObjectsInternal properties to the Visual Tool class.

Backwards Compatibility Note:

The following changes may slightly change the way updated models behave.

Deprecated the optquest() command. It no longer executes any logic when called. The associated OPT_* macros have been removed. FlexSim now interacts with OptQuest through the Optimization Job object.
Changed the default query handlers that deal with the Experimenter in MAIN:/project/exec/globals/serverinterface/queryhandlers. Custom applications that create web queries may need to be updated.

Process Flow

Added Process Flow Activity Statistics charts.

· 24

5 |100000

Attachments: Up to 12 attachments (including images) can be used with a maximum of 23.8 MiB each and 47.7 MiB total.

Clair A commented · Nov 09, 2021 at 01:03 PM

We can't install it.

I think there is an issue with the installer. On the website it says that the installer size is 629 MB:

But once the download is complete, the file size is 434 MB:

1 ·

1636462787686.png (32.9 KiB)

1636462803482.png (15.9 KiB)

1636462851319.png (17.0 KiB)

1636462943999.png (4.7 KiB)

Phil BoBo ♦♦ Clair A commented · Nov 09, 2021 at 02:46 PM

@Ben Wilson

0 ·

Ben Wilson ♦♦ Clair A commented · Nov 09, 2021 at 02:51 PM

@Clair A,

I have re-uploaded the installer.

I just tested a new download and installation with this updated upload, and everything worked well for me.

Please try again and let us know of any problems.

0 ·

Clair A Ben Wilson ♦♦ commented · Nov 09, 2021 at 03:02 PM

Thanks @Ben Wilson , with the updated installer it works well, I have completed the installation.

0 ·

Maryam H2 commented · Nov 10, 2021 at 01:25 PM

@Matthew Gillespie I see nothing happens when I hit that help button for "Tracked Variable", "Label Chart", and "Process flow activities". Is there any documentation for those charts that I can use?

0 ·

Matthew Gillespie ♦♦ Maryam H2 commented · Nov 10, 2021 at 08:17 PM

@Maryam H2 Some of the documentation for the newest features is still under construction. It should be up soon though.

0 ·

Ralf Gruber ♦ commented · Nov 11, 2021 at 10:03 AM

It seems a 22.1 OptQuest license is not recognized by the beta, while a 22.2 works just fine.

Is that a bug?

0 ·

Phil BoBo ♦♦ Ralf Gruber ♦ commented · Nov 11, 2021 at 04:26 PM

What does "is not recognized by the beta" mean?

Seems to be working fine to me:

0 ·

1636648275806.png (298.8 KiB)

Ralf Gruber ♦ Phil BoBo ♦♦ commented · Nov 16, 2021 at 06:40 PM

Phil,

In the "Jobs" tab I can add an Experiment only, but not an Optimization. If I load an old OptQuest model, I can duplicate the Optimization, but still not add one. Menu just shows "Experiment".

However, it now does not work with the updated 22.2 license either.

Originally I tried 22.1 based upon @Felix Möhlmann not being able to use the Optimizer as well. When I upgraded to 22.2, it worked, but now it does not anymore...or do I do something fundamentally wrong? Should I not even in a blank model be able to add an Optimization to the Experimenter?

Thx

0 ·

optquest.png (78.7 KiB)

Show more comments

Lars Jacobsen commented · Nov 17, 2021 at 08:36 AM

I am not able to install stable-baseline3. It says it depends on torch. What should I do? I have attempted to use Conda, as it is better at resolving package dependencies, but I am not able to activate Conda from with Visual Studio.

0 ·

1637138137161.png (17.4 KiB)

Jordan Johnson ♦♦ Lars Jacobsen commented · Nov 17, 2021 at 04:49 PM

I had this problem when I was using python 3.10; I couldn't get torch to build with that version. When I switched to 3.8 or 3.9, it worked fine. I also read online that it can happen if you are using the 32-bit version of python, when I was trying to figure out what was wrong.

I also used Visual Studio Code, rather than Visual Studio. The documentation has been updated for that case. If you want to use Conda, I'd recommend that option. I was able to use Conda within VS Code (which is not part of the documentation); as long as you specify a python version when you create a Conda virtual environment, VS Code will let you pick that environment as your default interpreter. I learned that on this page, which may have other helpful information for you:
https://code.visualstudio.com/docs/python/environments#_create-a-conda-environment

1 ·

Clair A Jordan Johnson ♦♦ commented · Apr 03, 2023 at 03:54 PM

Hi @Jordan Johnson, some of our users reported that they were blocked when following the Reinforcement Learning tutorial. The tutorial says:

Download and install VS Code, its Python extension, and Python 3

(https://docs.flexsim.com/en/23.0/ModelLogic/ReinforcementLearning/WorkingWithGym/GettingStarted/GettingStarted.html#configuringenvironment)

But it is not precised which version of Python 3 needs to be installed. If you install for example Python 3.11.2, you will be blocked later in the tutorial at the step where stable_baselines3 needs to be installed (https://docs.flexsim.com/en/23.0/ModelLogic/ReinforcementLearning/WorkingWithGym/Training/Training.html#trainingalgorithm), because there is no compatible version of stable_baselines3 with Python 3.11.2.

Could you please add a warning note in the tutorial here:

To precise that this tutorial was initially designed to work with Python 3.8. And maybe add the download link if needed:

https://www.python.org/downloads/release/python-3810/

+ 1 comment that if the user installs Python 3.10 or later, he will not be able to complete this tutorial, because some provided packages in the tutorial are not compatible with Python versions that were released after this tutorial was created.

0 ·

1680536586015.png (9.0 KiB)

Show more comments

David Seo commented · Nov 17, 2021 at 11:14 AM

I can't install the new Nov 15 version beta. The program looks like to have a problem.

The manual installation program is same also.

0 ·

Phil BoBo ♦♦ David Seo commented · Nov 17, 2021 at 04:42 PM

@Ben Wilson

0 ·

Ben Wilson ♦♦ David Seo commented · Nov 17, 2021 at 05:39 PM

@David Seo I tested a new download of the .exe installer. It was able to successfully upgrade my previous 22.0 beta installation. I then used it to uninstall the beta, and reinstall. It was successful in all 3 instances.

I then tested a new download of the .msi installer. Here I found problems. I have since re-uploaded the .msi. Since then I have used a new download from my account to uninstall and install, and both are working.

Let us know if you encounter any issues.

0 ·