@Rojo I'm stuck in fixing the es... I'm drawing a blank here. Can you point out why h_[x_[y__], x_[z__], a___] :> h[x[y, z], a] isn't a valid replacement rule at all levels?
Probably something very simple, but I'm not seeing it
@Heike I guess given the ability to Hold forms, one can basically delay evaluation, take the list, and reasemble it, which should eliminate the eneds of macros.
is it OK for WRI employees to post answers without disclosing their affiliation? (just to be clear, I am asking for opinions, not trying to lead the discussion to a "no").
The main thing that would worry me is that one can end with messy situations (see the talk page on the wikipedia mathematica page). It's unlikely the way things are now here, and I certainly would not want us to alienate anybody, hence the question here.
@acl IMO, yes, absolutely fine. If they want to remain unassociated with WRI as far as Mma.SE goes, or even post anonymously, I have no problem with it. It's not like most of the rest of us disclose our affiliations or even our real names so I don't see why WRI employees should be held to a different standard.
@IstvánZachar Unfortunately not any more ... I'm actually from Transylvania, and I just got back from there. It doesn't look like I'll get the chance to visit again until next summer. (I might go to some conferences to Budapest though---I'll let you know!)
@acl I think it's completely fine. Just imagine: what if one day WRI employs you? Your attitude won't suddenly change, and it would feel very bad if suddenly you couldn't post the same way ...
@user1311390 Unfotunately it doesn't, and I'm not sure they can be implemented cleanly and safely. Many of the things that are done with macros in other systems can be done in Mathematica.
@Szabolcs I think it's worth noting though that definitions can be made on arbitrary patterns; we don't have the limitation from other languages that functions and other types of values are different from each other in any way. So the requirement for macros is greatly reduced in the first place.
@OleksandrR Yes, thanks for stating that more explicitly!
I hope I didn't sound like I was in a fighting mood here, I was just trying to explain my concerns. I don't have strong feelings about the issue.
@BrettChampion After we graduate, these may be worth putting into temporary community promotion ads If someone from WRI is willing to put together the correctly sized PNG image and post it, I'd vote for these. They're of general interest.
@Szabolcs your answer seems quite moderate. Good points are made on both sides. Personally I'm ambivalent, but I haven't tried SystemModeler myself. It strikes me that these types of programs mainly solve systems of coupled differential equations, which can be done in Mma as well, if you have another method of setting up the calculation (be it manual or based on some intermediate definition). As such IMO the question which sparked this discussion is on-topic.
@Szabolcs A lot of work has gone into Mathematica's NDSolve already, and we know a lot more work still is being done for version 9. It makes sense that WRI would want to make use of this for SystemModeler, so Leonid's perspective on future developments looks very plausible. On the other hand, if Modelica and Mathematica-the-language are completely unrelated, I agree that we ought to think twice before allowing questions that may not relate to Mathematica-the-product at all.
@OleksandrR I did wonder how these modelling programs are different from ODE solvers. The first example in the SystemModeller tutorial just solves $y'(x) = -y(x)$. I put away SystemModeler for a while though and when I came back to it the trial was expired ... so I really only have very minimal experience.
@OleksandrR Maple has a competing product, MapleSim. The strong point of MapleSim compared to other Modelica systems is said to be that it can symbolically simplify the systems of ODEs before solving them, improving stability and speed (???). Or something like that ...
@Szabolcs yes, it makes sense. I was aware of MapleSim. MATLAB also obviously has Simulink. These two companies being WRI's biggest competitors, the appearance of SystemModeler is not too surprising.
I really have no experience in this field either. In fact the only application of these applications that I'm at all familiar with is in post-Keynesian economics (specifically, the work of Steve Keen, an Australian economist whose approach in using these methods is quite radically different to other economists I'm aware of).
Keen however generates his differential equations from a table that looks like standard double-entry bookkeeping as used in accountancy. And rather than Maple, MATLAB, or Mathematica and their simulation packages, he uses Mathcad and a custom solver.
@Szabolcs Ironically the physics department in my university uses Maple rather than Mathematica. Our site licence for Maple is much broader: all staff and students get to use Maple for free. I wish that were true for Mathematica! Then I wouldn't be the only person using it...
Maybe you can do some magic with Mathematica to convince them :-) Just let them watch while you're building something quick and dirty---they'll be amazed :)
The problem is, they're all casual users, most with no programming experience. I'm an experimentalist. The only others who use Mathematica here are theoreticians, and even then, only because one of the professors uses it. His students just use it for doing integrals AFAIK. People here use Origin or Excel for data analysis. I can hardly get over how awkward that is.
@belisarius haven't heard of that one. The only data mining program I was really aware of is the one produced by the company set up by the ex-CEO of Thinking Machines. But I forget her name and the name of the product, so "aware" might be overstating it. :)
@belisarius oh, I completely missed the fact it was free. That's worth suggesting to people then. Usually I suggest either Mathematica or (if they don't want a commercial product) Python + NumPy. For myself, I'm satisfied with Mathematica.
@belisarius The datasets I work with are 4-dimensional image data and are about ~70MB in size. I sum over slices and blocks and fit them using NonlinearModelFit with Daniel Lichtblau's (Daniel Glosemeyer's) trick for simultaneous fitting of multiple datasets.
:5298602, @Szabolcs I agree that the rest of us do not have to so it seems unfair. Note that I am absolutely not claiming that they have anybody has shown a bias or anything like that (which seems to be how Szabolcs interpreted my question). Now, I had a specific worry, which neither of you have addressed. Perhaps it is trivial and that's why but could you please discuss that? Saying "what you say is stupid" is fine if that's your opinion :)
What I'm trying to say is, there's a specific thing I'm worried about. What do you think about it?
It did happen before, and it's probably going to happen again. There were accusations on A51 before the site was launched and there was that reddit thread and the comments about the "planted question". But none of these people could explain what is wrong with being a WRI employee and posting answers (perhaps without explicitly pointing out their affiliation). I think it's going to be fine, no need to make a big deal about it. Just don't feed the trolls when they appear :-)
@belisarius did you see that wikipedia talk page I mentioned? it only takes a couple of combative people to make the whole thing a mess. I am, yet again, not accusing any employee of WRI of anything. Can the discussion continue without defending the WRI employees here?
That is a much more extreme case than WRI employees posting here, and people agreed that there was nothing wrong with it (his answers were helpful). He at least did has a personal interest to post those answers. That's not the case here.
Look. I have no objection to them participating. I am worried about what I will refer to as trolls taking the opportunity to mess things up. It hasn't yet happened, but more and more people are taking part.
The comment on Steven Jobs suggestion is one of the most stupid things you could add to this article. — Preceding comment added by (talk) 02:18, 10 November 2011 (UTC)
:Agreed. The placement is kind of insulting as well. It suggests to me, a Mathematica+Maple user, that the Wiki community thinks that Steve Jobs having suggested the name for the software is of first paragraph importance. I don't find that assertion to be notable. Maybe move it to a new section, we could call it "Things people who use maths software don't care about but which are of utmost importance to Steve Jobs fan...
That page has been a mess for many years, unfortunately.
The comment on Steven Jobs suggestion is one of the most stupid things you could add to this article. — Preceding comment added by (talk) 02:18, 10 November 2011 (UTC)
:Agreed. The placement is kind of insulting as well. It suggests to me, a Mathematica+Maple user, that the Wiki community thinks that Steve Jobs having suggested the name for the software is of first paragraph importance. I don't find that assertion to be notable. Maybe move it to a new section, we could call it "Things people who use maths software don't care about but which are of utmost importance to Steve Jobs fan...
There's that period in young men's lives when they can get excessively combative, and if they have nothing better to do this can turn into aggressively attacking anything non-open source online :)
@Szabolcs I don't see it being combative. And I don't get Leonid's argument at all. I understood it to be: allow questions on X because it may be integrated into mma in the future. I don't really understand this reasoning. Could be misreading it though.
@belisarius thanks anyway! For me the problem is not so much finding a functional form that fits the data, but rather extracting what I want from the larger dataset and then fitting it in a physically reasonable way. I'll definitely install Nutonian and see what it can do.
@belisarius Look. We are going in circles. I could not care less if you are Stephen Wolfram or anybody else. I am not saying X or Y are right or wrong in that discussion. I OFFER IT AS AN EXAMPLE OF WHAT I DO NOT WANT TO SEE HERE. Sorry for the caps.
@acl Yes, you're right. It should definitely be avoided. I've seen that talk page many years ago, and I lost all desire to ever click it again (or to edit the main page)... I think something similar has happened twice. Once someone was arguing on A51 that SE, a free platform, must not "support" a proprietary product, and of course he was pointing out that several of the committers are WRI employees. Then there was that reddit thread where someone (who later apologized) accused
@acl the poster of "planting" a question, implying they may be working for WRI and advertising Mma. Both were ugly, but both were resolved, and most traces are gone. Fortunately neither happened right here on this site (even if they were about this site). Based on this experience, and the staleness of the arguments brought up there, I'm not really worried. Also, most do disclose their affiliation. So personally I don't think there will be problems---but it's good to bring this topic up,
we should make sure if there's trouble, it's handled well, and doesn't grow out of control like on Wikipedia.
@acl I'm sorry, I completely missed that part of your question. My perspective on this is, Mma.SE will graduate soon, and then we're autonomous enough that we don't have to care about this possibility any more. If trolls appear, we can just moderate them away again. None of those who fought so valiantly for the beta and not that many (as a proportion) of the active users currently are WRI-affiliated, so it will be hard to cry foul in that sense.
@Szabolcs SE is not a free platform. The software is proprietary and they will want to monetize the network eventually (when and how they will do that I don't know). I get the impression they chose CC-BY-SA to protect their own interests as much as anyone else's. So I think the scope even to push the tired old "free-as-in-freedom vs. proprietary" argument will diminish in future.
@acl I think your worries are sensible. But ultimately this site and the SE model is not based on egalitarian democracy like Reddit or Wikipedia. Instead it's a technocracy with some democratic elements. Thus there is a lot less scope for small elements and unpopular viewpoints to cause trouble here than elsewhere.
I missed most of this and couldn't figure it out from the transcript — was there something that triggered this? Most people here have so much pent up frustration from former well-intentioned but annoying SO users and other not-so-well-intentioned trolls on area51 (this was a very nasty episode) that any new trolls will be put down pretty quickly
@OleksandrR I don't particularly want to get into it, but the reason Steve Keen's methods are unusual is that he came late to economics, as a mature age student. I can't remember what he did before but it influenced his methodological choices. There is a lot to dislike about mainstream economics and its methodologies but he has chosen to start again from a blank sheet of paper. No building on the shoulders of giants for him.
@Verbeia he cites Schumpeter, Marx, Minsky, and Keynes as his influences. An eclectic mix, to be sure, but I think there's a lot to like about his approach.
I am trying to run NonlinearModelFit in mathematica on some data for a project. It is a fairly complicated model having 9 coefficients, and 5 independent variables. I created the data the way that the model needs, and I think I called the function correctly.
Here is a sample of the data:
{{16.58...
@RM I was tempted to use your Composition@@ to build linked lists, but I bumped into a bug I think. You think it's a bug I should report? MMA8.0.4, W7x64, (Composition@@Range[80000])[EndOfString] makes my session die. Up to 50000 for example works great
So we can't buy and download Mathematica anymore ;) I'm told that "This item is not currently available for purchase on the web. Please contact Wolfram Research Customer Service for assistance in ordering this product."
@acl @szabolcs @belisarius I don't think it's a problem as long as the PR department is left out of the loop. Personally, I would prefer full disclosure but I can imagine there are reasons for some not to do that. The list of users with links to WRI is probably longer than you imagine. I'm aware of 34 users, some who I think are interns trying to learn something here.
@user1311390 @SjoerdCdeVries Your account is associated with Stack Overflow, where you don't have a name set. If you change your chat association to Mathematica, then it'll automatically refresh in 30mins-1hr, or Sjoerd can force it manually
so I'm learning about Listable functions: let me check if this is right: Listable is NOT something the evaluator/kernel evaluator automatically does. It's something that the Funcions do. I.e. they provide a Function[x_] option and a Function[ list of crap ] option .... ?
i.e., for functions wehre Attribute[Fn] does NOT include Listable, the kernel does not autoamtically listable-it
@SjoerdCdeVries I wasn't worried about them but about trolls (for lack of a better word) using that to create a mess. however I now think it won't be a problem (see comments by @szabolcs and @oleks
Clojure is basically Scheme for people who wants to get shit done, and are willing to give up continuations, tail cail recursion, and embrace the infinite libraries of the JVM
I do find Mathematica, in some sense, more terse than Clojure, perhaps due to the ideas lifted from APL.