Some tiny improvement ideas for the new academic license delivery

Hi @afattahi 

Glad to react here.

The new academic licensing scheme is part of a larger effort to also introduce a pretty effortless scheme to request a free community edition of AIMMS we’ll be introducing later this year. Both of these license types are aimed at providing a true interactive developer/learning/research experience with specific limitations. And while it is fine to call an AIMMS model from e.g. a python script you’re developing to combine optimization with ML/AI work you’re doing there, these AIMMS licenses are never intended at being able to embed AIMMS models in an automated ‘production’-like of way into a larger application. If that’s what you want, then our commercial development licenses or deployment platform come into play.

The ‘annoying’ dialog is to explicitly remind you of the license limitations we impose on these licenses every time you run AIMMS, and also to make production- or automation-type mis-use of these licenses harder. It is the small price you have to pay for the value you get for free (but admittedly remembering your previous ‘accept’ choice could be a tiny removal of burden ).

That being said, if you’re developing or experimenting with a model in AIMMS, the need to restart AIMMS a lot is not really there. You can interactively develop on your model, where the data you were already working with will remain in memory where ever possible, or load a completely new data set into an existing session, or run a lot of scenarios with different data in a loop in a single session. In that sense AIMMS is different than regular programming languages where you have to stop a running session and start and compile before you can run with changed source code. With AIMMS you can easily get away with running a session for hours, while changing your model completely and running multiple experiments. 

As far as the burden to start the launcher every time and having to explicitly choose an AIMMS version: I think you didn’t realize that the launcher will register itself with Windows to actually allow you to double click on an .aimms file directly. In doing so, the launcher will detect the closest matching AIMMS version, and start that. If you specified an academic license URL in the launcher, that license URL will be used as well. I guess that will address the last issue you made. 

Hope this helps you understand the choices we made. 

Dear @MarcelRoelofs,

Thank you so much for your prompt reply. 

There might be some misuses of the academic license. However, adding some 'burden' to accept the license agreement would not prevent misusages, rather, it just adds unnecessary complexity to academic users. There are other applications which support free academic licenses but they tend to facilitate the process as much as they can, for instance, the Gurobi solver suite. I do not think that explicitly "reminding” your academic users about the license's limitations every time is the smartest idea to attract more users. Moreover, I do not think that AIMMS believes academic users should pay "small prices" for getting the value for free. From your words, I get the impression that providing a free academic license is a burden for AIMMS, rather than an opportunity. This directly contradicts the impression I got from @Gertjan, who told me AIMMS fully supports the free academic license program in its long-term strategy. 

As an academic developer, I have to restart the AIMMS each time after a model run, since AIMMS keeps the tree of parameters in the RAM and it can cause some numerical issues in the next consecutive runs. We have already discussed this issue in another thread. Unfortunately, AIMMS cannot start fresh even after emptying all parameters through clean and empty functions. Therefore, to avoid unwanting numerical issues, I restart AIMMS after each run. 

All this being said, I appreciate you and your colleagues' efforts at improving the software every day and providing a free fully-functional license for academia. I was in contact with your colleagues to resolve some issues, and they fully supported me as an academic license holder. Currently, I advertise the AIMMS software in academia as a state of the art tool which fully supports free academic licenses. Recently, I and my colleagues published a scientific paper in the prestigious Applied Energy journal, in which we stated the use of AIMMS and it's support for free academic licenses. 

I hope I've explained myself well. 

Hi @afattahi,

Sorry if I gave you the impression that the academic license is a burden for us. Not at all, like @Gertjan I think it is quite strategic, just like the community license we’ll be introducing shortly.

It is good to learn from users like you how these licenses are actually used, so that we can validate whether what we think is reasonable will actually work for the use cases you present, and how we can adapt our approach to facilitate those use cases.

I’ve seen some other use cases that necessitate me to go back to the drawing board to make some small adaptations to make things work more smoothly.

Thank you very much for making the case. It provides us with good learnings.

Hi @afattahi 

You might want to check out whether the changes we made in AIMMS 4.78 will address your concerns. You can download and install from within the AIMMS Launcher as usual. 

In short: the Acceptance dialog is shown only once a day, and you can simply accept by pressing the ‘I Accept’ button (so no double action required any longer). 

The latest version of the AIMMS Launcher (https://download.aimms.com/aimms/download/data/AIMMSLauncher/AIMMSLauncher-1.1.0.17.exe) will also pass along any additional arguments you want to call your project with, giving you some more freedom to call it from the commandline (or via a shortcut on your desktop). 

Hi @MarcelRoelofs 

I have just downloaded the new launcher and AIMMS 4.78. 

I would like to thank you for your swift response and for considering your users' demands. 

I these days, I realized another new limitation. Before, I was able to run two instances of AIMMS at the same time. However, now I can only run one instance at a time. Here are my two main uses for parallel instances of AIMMS:

1. I am dealing with a quite heavy energy system model that runs around 2 to 20 hours (depending on the level of details). Sometimes I need to run several scenarios for a specific sensitivity analysis. It would be very useful for me if I can run the model on several computers at the same time.
2. Sometimes I want to continue coding in AIMMS while it is running a problem. 

Would it be possible to run two or more instances of the model at the same time for academia? 

@afattahi , 

Not sure if you noticed, but we have extended the use to 2 instances at the same time when using the Academic license. Enjoy!

@Gertjan Yes! this helps a lot, thank you and @MarcelRoelofs for the swift response.