up vote 11 down vote favorite
4

I predict (or at least would like to see) some questions will be raised about the new platform for general symbolic modelling. A lot of us use Mathematica for modelling simple or complex systems, and this new software might make it easier to develop such models. So, should questions related to SystemModeler be answered/encouraged here?

share
4  
Yikes! I had high hopes it would be part of MMA 9. Apparently, it's a separate product. Bummer. Good find and question. One problem: the user base will be quite small initially and many experts of MMA won't have access to it. I'd expect plenty of zero answer questions. – Sjoerd C. de Vries May 25 '12 at 9:07
I'd say: before making a final decision on this, let's wait until SystemModeler gains some popularity, and let's see if it will be bundled with (any) version of Mathematica 9 or not. For me the main question is: will a large fraction of Mathematica users learn how to use SystemModeler? Will they want/need to use it? If yes, it should be on-topic. If only a tiny fraction of Mma users will use SM (as with the original MathModelica), then maybe not. – Szabolcs May 25 '12 at 9:08
@Sjoerd Yes, I'm concerned about the same---will there be enough expertise? But what if it will come bundled with some "full" version of Mma 9 that many universities will purchase, and therefore lots of our regulars will have access to it? Let's wait and see! – Szabolcs May 25 '12 at 9:09
Any info on licensing? Does a standard Premier Service license include SystemModeler? What about Home use licenses? – Sjoerd C. de Vries May 25 '12 at 9:14
There was some discussion on this in chat yesterday... the general feeling was that they're off-topic unless it involves interfacing with the regular Mathematica in some way. I'll wait for those folks to post their views themselves – rm -rf♦ May 25 '12 at 9:25
@Sjoerd I looked at the announcement blog post again and the official comment replies (the ones with red frame) are rather vague on this. They carefully avoid saying whether this will come with Mathematica 9 or not. – Szabolcs May 25 '12 at 9:29
@R.M We already get questions about commercially available packages, like the Time Series question last week. Would you consider those to be off-topic? – Sjoerd C. de Vries May 25 '12 at 11:02
@SjoerdC.deVries I have no experience with any of these commercial packages and have no opinion on this issue. I wasn't part of that chat conversation either. I do understand though that if people ask questions about something and there's no one to answer, then it could give a bad impression – rm -rf♦ May 25 '12 at 14:22
1  
From what I've gotten from an (admittedly very short) look at the product page, I don't think it should be on-topic here (unless and until it gets part of a future Mathematica version). Just because it uses Mathematica (and is from the same company) IMHO shouldn't be sufficient to make it on-topic here. – celtschk May 25 '12 at 15:08
7  
@celtschk Since this may convince the folks that now crowd to Matlab/Simulink to discover Mathematica/SystemModeller it might be good for Mathematica and its community if we could offer this site as a platform for SystemModeller questions too. Where else should they turn to? This could make MMA much bigger and ensure job security for us all ;-) I furthermore suggest WRI provides copies to the top-25 people here and to a random drawing of 25 more to increase the expert base. – Sjoerd C. de Vries May 25 '12 at 15:35
@Sjoerd Why don't you make this an answer? – Leonid Shifrin May 25 '12 at 15:47
1  
@SjoerdC.deVries I got my copy... – Brett Champion May 25 '12 at 16:57
1  
@LeonidShifrin OK, see below. – Sjoerd C. de Vries May 26 '12 at 10:24
4  
@szabolcs This more or less confirms what I thought and what I stated in the 5th paragraph below. I'd say this favors restricting questions to interfacing with MMA only and not SM itself. – Sjoerd C. de Vries Jun 5 '12 at 11:53
2  
@rcollyer It's part of our site license. :-) – Brett Champion Jun 7 '12 at 2:45
show 9 more comments

5 Answers

up vote 10 down vote

Let me first state that I think SystemModeler looks like a very impressive product, tying the modeling language Modellica to Mathematica, providing a very nice graphical user interface to model development and simulation, the ability to generate standalone executables and a link to Mathematica to enable further analysis.

enter image description here

As far as marketing Mathematica is concerned this is a very strong move. The combination of Matlab and Simulink has always had an enormous appeal to the engineering world and I believe that the Mathematica/SystemModeler combination may finally break open this world and cause an influx of its user base. Mind you, this will take years, since curricula won't be changed in a few weeks, and there is a large amount of vested interest involved in Matlab/Simulink that will resist change.

I believe it will be important for many of us to learn this new stuff, perhaps from a job-security point of view. If this takes off, being a front-runner could be a definite advantage.

Opening up MMA.SE to SystemModeler and providing them a warm and cozy environment will contribute (for what it's worth) to the acceptance of SystemModeler and Mathematica and will, at the same time, improve our own understanding of the system.

enter image description here

From a Mathematica language perspective, SystemModeler is perhaps less of an paradigm shift. I may be mistaken here, but it looks like we can fire-up a model simulation from MMA, get back data and analyze that. They are not strongly integrated as a quick-scan by Szabolcs shows. Arguing from this perspective, one might say that SystemModeler questions should not be asked here except for questions related to interfacing MMA and SystemModeler.

The specific expertise on this area seems to be not very high either (as perhaps demonstrated by this control system question that is still open). So, asking questions here wouldn't be very useful to the asker, at least, not in the beginning. As I said in my comments above, I wouldn't mind if WRI would distribute some licenses to the avid users here to alleviate this specific problem.

Weighing the pros and the cons I'd say we pick a wait-and-see approach. People who would be annoyed by seeing SystemModeler questions appear here could easily add the tag to their "ignored" list. And people only interested in SystemModeler questions could gather under this tag like many of us gathered under the tag at StackOverflow.

share
2  
I agree with "wait-and-see approach", +1. – Szabolcs May 26 '12 at 14:23
3  
If the integration with Mathematica is comparable to JLink, I would not want SystemModeler questions here, just as I don't think Java questions would belong here. Questions related to the interface itself are fine, I think. – Jens Jun 6 '12 at 21:43
up vote 6 down vote

I strongly disagree with the opinion of @Szabolcs. While his arguments are formally in complete accordance with (and enforcing the) rules of SE, and Mathematica SE in particular, I think that banning Modeler-related questions will harm our community in many ways. Here are my arguments:

  • Since the Modeler is a result of WRI, being a re-branded and evolved version of MathModelica from MathCore (bought by WRI), it clearly shows the long-term intention of WRI to have the Modeler and Mathematica deeply intergated. If this integration is not so strong now, I am sure this is just because the work in that direction just started. In other words, I expect that in time, the percentage of questions involving the Modeler and not at all Mathematica will drop to a pretty low value.

  • One can argue that by the same token, we should also allow all WolframAlpha questions, because WA is another system integrated with (and written in) Mathematica. But there are several important differences:

    • WA has its own, and pretty powerful, deployment channel - through the web. SystemModeler does not have such a deployment channel. So, while formally it is true that it can be used without Mathematica, realistically one can expect it to be used with Mathematica most of the time.
    • While WA can be used from Mathematica, its purpose is, in some sense, the opposite of what Mathematica does. The Modeler, however, can be seen as a component extending and empowering Mathematica in a certain wide domain (modeling of Physical and Engineering systems), which is precise and quantitative, just as Mathematica itself.

  • Modeler is, compared to Mathematica, a more specialized application, and people who are interested in it currently have no place to go (except WRI tech.support). By banning the Modeler-related questions, we basically show the door to all these folks, while they could become a sub-community on Mathematica SE. There is no chance for Modeler to get a separate SE site in the near future. So what we do then is to destroy the future synergies we all may have with the emerging (or existing for prior MathModelica) Modeler community. This is harmful for everybody, since we don't give the emerging Modeler community a chance to use M SE infrastructure to get stronger, while we don't give ourselves a chance to learn about the Modeler and get an additional job security, mentioned by @Sjoerd.

Let us keep in mind also that Modeler is pretty much the first large-scale industrial application tied closely to Mathematica and produced by WRI. Crippling its emerging community, we also reduce its potential adoption in the industry, which, to an extent, reduces the potential adoption of Mathematica itself in the industry. Such an adoption for Mathematica is sorely needed, since IMO Mathematica is heavily underused in the industry currently.

What we risk is an increase in the number of unanswered questions. I don't know how much a problem this is for our community, and I also hope that once we get here some folks with some knowledge of the Modeler (and I would encourage WRI employees with such knowledge to take the lead), this will become less of a problem. Usually, a demand leads to a supply, so I hope that such people would naturally appear here after a while, if Modeler is any good (and I believe it is).

What we also risk (and this is IMO more important) is to lose, to some extent, the sharp focus on Mathematica, and somwehat dilute our pure Mathematica discussions with those about another system which has a large non-Mathematica component following different logic. How big of a problem is this? I think, this depends on how we would like to view Mathematica SE, and also on the degree to which WRI will be able to integrate Modeler and Mathematica into a single smooth workflow in the future. My own opinion is that restricting Mathematica SE to only strictly Mathematica questions is a rather limited way of developing our community. StackOverflow, for example, benefits a lot from having tags for various programming languages co-exist in one place - so why can't we also? In some sense, Mathematica SE can become a hub for Mathematica-related technologies (living in different tags) just as StackOverflow became a hub for various programming languages. And those who don't want to see any Modeler-related questions, can just set up a simple tag filter.

So, banning Modeler-related questions may look like a right decision tactically, but I think it will be a wrong decision strategically for our community, and for general Mathematica adoption as a platform.

share
I agree with this (and Sjoerd's answer), but I would also proceed with a bit of caution as Sjoerd mentioned. Not because I think System Modeler is inherently bad for us — I don't, and we can always house it under a tag and move on as if nothing happened — but because this is the only (so far) avenue for such questions. As such, if the adopters of System Modeler are neither mma users nor SE users, then we might have a problem. If the question influx is not of the type that we would like, we might have to tighten our rules surrounding it, but until then, I would welcome such questions. – rm -rf♦ Jul 10 '12 at 0:44
@R.M I agree, but I would go even further and say that we may want to have a decent amount of patience and not judge by first impressions. We have the power to control all that, because, until a number of Modeler users (who will not be also power Mathematica users) earn enough rep, we can pretty much dictate the rules we want. So, we will be in a position to impose such stricter rules even much later. With the basic / homework questions, the situation is different, since the user base corresponding to them is orders of magnitude larger, and they will have a large cumulative voting power. – Leonid Shifrin Jul 10 '12 at 0:53
1  
@R.M. In fact, this controlling mechanism is also nothing new, and is one really important and IMO often overlooked implication of all languages co-existing on SO: if something really wrong is happening in some of the tags, high-rep users from other tags can interfere and help bring order. We were suffering from this on SO, since other folks just didn't get Mathematica, but more often than not I think it works fine for pure programming tags. We can have a similar mechanism here - this is basically what I described in my previous comment. – Leonid Shifrin Jul 10 '12 at 1:02
I don't have a strong optinion on this, I merely think it was time to come up with a decision, and I'm glad you wrote this answer. But after reading it, I do have to ask: did you try SystemModeler? My opinion on the matter changed after I tried the trial version (before I also thought it's a bad idea to make it off-topic). The gap between it and Mathematica may be larger than you realize. I wonder if it is really the case that people who use SM do most of the time you it in conjunction with Mma ... it is true though that WRI said they want to increase the integration. – Szabolcs Jul 10 '12 at 8:50
@Szabolcs No I did not try it. And I am pretty sure that currently it is far removed from Mathematica. The thing is, I am not put off by the gap, however large it is currently. It may be that right now, many people use the Modeler stand-alone. But I think we should be looking at the future trends, and they seem to indicate a much tighter integration in the future. – Leonid Shifrin Jul 10 '12 at 9:32
@R.M and Leonid: The comment I started writing got too long, so I posted as a separate answer. I think the advice of someone who has experience with both technologies (Mathematica and Modelica-the-language) is sorely needed here... – Szabolcs Jul 10 '12 at 10:45
up vote 5 down vote

I completely agree with Leonid. Besides we already have a resident expert in WSM : Malte Lenz, who appears to be ready and willing to help newcomers sort out problems with WSM. Please also cosider that WRI has probably heavily invested in WSM and if it flops, this might have negative ripercussions even on MMA develpment. I have tested WSM and will likely buy it soon. With WSM you can easily model a system, but if you want to control it, you basically need MMA. So no serious user of WSM will go a long way without MMA. Thus these WSM users will end up here probably with questions relating to the control system sub-language in MMA. I therefore strongly advise to welcome ALL kinds of questions on WSM.

share
up vote 3 down vote

I have (very) briefly tried the trial version of SystemModeler. In its current state, it seems to be a program completely separate from Mathematica, which has a Mathematica interface. It does not require Mathematica to functions. Using the interface, simulations can be run from Mathematica and the values of all variables (and their time dependence) can be retrieved.

Because SystemModeler is separate and independent from Mathematica, and because it is based on a separate and unrelated programming language, my opinion is that SystemModeler should not be any more on topic than any other system that also has a Mathematica interface. Questions about the Mathematica integration and interface should be on topic, but questions about SystemModeler that don't require Mathematica to solve should not. (Just like Java questions are off-topic but JLink questions are on-topic.)

Notes:

  • My experience with SystemModeler's trial version was very very brief, so I may be wrong about some points.

  • Here's the first SystemModeler question that triggered this answer. The question seems to be about SystemModeler only (not Mathematica), so I'd consider it off topic.

    How to do System dynamics simulations / diagrams in Mathematica?

share
1  
As the one asking the questions, my main interest is being able to do system dynamics modeling with Mathematica, and not so much with SystemModeler so I belive the spirit of the question is on topic: should I delete any reference to SystemModeler from the question to avoid deletion? – Luxspes Jul 9 '12 at 19:35
I have the opposite opinion on this, which I expressed as a separate answer since I could not squeeze it in a comment or two. Don't take it personally, I can see your points as well and this is not at all a clear-cut case. – Leonid Shifrin Jul 9 '12 at 23:06
@Luxspes I'm not suggesting to remove all traces of SystemModeler from the question, just to re-focus it on Mathematica rather than Modelica in general. I must mention though that this is my personal opinion and not site policy (that's why I didn't vote to close the question). I am simply trying to get the community to come to a decision regarding allowing purely SystemModeler-related questions here. Your question is the first of this kind. – Szabolcs Jul 10 '12 at 10:11
up vote 3 down vote

For the context: I was planning to post this as a comment to Leonid's answer, but it got too long, so I'm posting as an additional answer.

To rehearse first:

If I understand you correctly, your argument for allowing SystemModeler-questions is that 1. SM is a technology closely tied to Mathematica and 2. that people are likely to use them together. For 1., from what I have seen during the few hours I played with SM is that currently it is not really tied closely to Mathematica. Mathematica simply has an interface to it, just like it has an interface to TetGen or to Java (which doesn't make TetGen or Java questions unrelated to Mma on-topic). You are correct that the WRI blog posts on SM suggest that a much tighter integration is planned in the future---this integration is not here yet though. 2. Is it true that SM users typically use the Mma interface to it? I really don't know, the fact that SM is also a WRI product now makes it possible. And questions about the Mma-SM interface can of course be on topic. Is this a good reason to make SM questions not related to Mma on-topic?

Another thing to consider is that SM is an implementation of Modelica. From what I read about Modelica, one of the motivations behind its development (as a standard language) was to make different modelling systems interoperable. If we do allow non-Mathematica-related SystemModeler questions, shouldn't we allow all Modelica-questions, regardless of which system the asker is using? Especially is these systems are interoperable or the code can be transferred between them with no or minimal modification ... Several of the questions would be valid not only for SystemModeler, but for other Modelica implementations as well. Note that the library mentioned in this question is not for SystemModeler specifically---it's a generic library that can be used with any Modelica system.

My updated suggestion: Let's allow any question which is clearly related to Mathematica---e.g. requires using Mathematica or writing Mathematica code. But let's not allow SystemModeler questions which are only about SM, which can be solved without making use of Mathematica, which are mainly about Modelica, or could be valid for other Modelica implementations too. (Without exaggeration, I think this is not that different from not allowing purely Excel questions just because WRI sells an Excel link.)

If SystemModeler and Mathematica will really get a tight integration in the future, that will automatically and naturally make more SystemModeler questions on-topic here. Until then, let's not open up the site to Modelica-the-language, and restrict SystemModeler questions to the interface between it and Mathematica.

share
1  
Thanks for sharing your points once again, but I don't see how the core of your suggestion here differs from what you suggested previously. As to the subject matter, I think it is far easier to distinguish questions on general Modelica (and ban them) from those on SystemModeler, than distinguish, within SystemModeler, questions which are and are not on topic (from the viewpoint of being consistent with our rules). My suggestion was to be generous and allow questions on a system (Modeler) which currently has a large non-Mathematica component, for the future benefits of all of us. – Leonid Shifrin Jul 10 '12 at 11:03

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged