Wolfram Mathematica

Adam Dreaver

12:43 AM

Does mathematica have traditional programming datatypes? Like an int that is 32 bits and it will obey overflow rules?

I'm trying to simulate a PRNG implemented in a programming like c++ which is using unsigned ints!

s0rce

2:18 AM

@Adam can't you just do Mod 2^32?

Adam Dreaver

Maybe, I'll try that out, ty you for the suggestion

kirma

6:52 AM

@Adam there are machine integers on compiled functions, but they're not really there to be relied for modulo 2^32 arithmetic. You have to know their limitations, though, if you try to use them as arbitrary-lenght integers. And sure, they can be occasionally used for improved performance: mathematica.stackexchange.com/a/13708/3056 (very un-Mathematica though.)

kirma

7:13 AM

@Adam and speaking of modelling a specific PRNG on Mma, using Mod 2^32 almost definitely the way to go unless you're aiming for speed (I suppose not).

Dominic Michaelis

huhu is someone here ?

Rojo

11:54 AM

Someone is

Oleksandr R.

@Rojo hi, someone

@Ajasja I'm going to update the Nelder-Mead answer at some point, BTW. I cleaned up the code a bit and fixed a bug with option handling. I also tried the so-called "adaptive" Nelder-Mead parameters of Gao and Han (here), which seem to be better in most cases for high-dimensional problems

Ajasja

That's Great! Although the current implementation is working very well for us, so no hurry.

@OleksandrR. I remember your comment a while ago about recommending Python for people that have no previous knowledge if either MMA or Python.

Rojo

12:10 PM

Somethree now

Ajasja

I was part of a workshop on scientific programming and have come to realize that the advice is spot on.

Python + numpy + scipy (+ simpy) is a pretty powerful open source toolbox used by a large community.

Oleksandr R.

@Ajasja I find that Python is pleasant to use and tolerant of a wide variety of programming styles without suffering horrible performance. I prefer to write functional code, but if people are used to a procedural or OO style, that's fine too.

NumPy's advanced indexing is also better than anything we have in Mma

rcollyer

I've never used python, but a friend of mine did all of his analysis and simulation work for his PhD in it.

Ajasja

@OleksandrR. Yup, so I have discovered:) Of course the licence is Pythons greatest advantage over MMA. Any scripts I run anybody can use as well.

And of course the availability of any number of external libraries.

CBenni

12:31 PM

So if I export a function to C, and compile, will I be able to execute the program on a PC that doesnt have mathematica installed?

I guess not

but thats why Im asking

Ajasja

@CBenni For simple compiled functions it's actually possible to do.

Don't know the exact license terms for the runtime though.

@rcollyer My thesis still awaits (eagerly). But I'll keep it MMA for now. Too much already invested time to switch. And besides I still like MMA better.

Well gotta go. (thesis calling:)

CBenni

@Ajasja mmhk... Apparently, my function is not easy enough...

CCodeGenerate::wmreq: The expression Runen requires Mathematica to be evaluated.   The function will be generated but can be expected to fail with a nonzero error code when executed. >>

Was to be expected

the function calls NMaxmimize :(

P. Fonseca

Hi everybody! I need some help, but it is probably not question worth... I have a couple of PDE (time and space/length) that are correctly managed by NDSolve, but to which I’m adding an ODE that “accumulates”, through time, a “parameter” coming from one of the boundaries of the PDE. The system is “linked” in the sense that the boundary varies with the accumulated value. I’ve tried a lot of stuff, but NDSolve keeps giving errors on the amount of parameters of the ODE, etc.

Turning it into a “PDE” didn’t help... I’m now thinking on managing the accumulation with the EvaluationmMonitoring, but it doesn’t sound right. Does someone has experience with this kind of scenario?

(I meant "not question worth" because I can't formulate it correctly...)

CBenni

12:48 PM

I would like to help, but I dont know either

sorry about that ^.^

P. Fonseca

@CBenni Thanks anyway :)

CBenni

YOu might just try to ask it on Mathematica.SE as a question

do your best to formulate it

add error messages

P. Fonseca

Did you understood my above description?

CBenni

the people there have been very nice always ;)

I think I got the point, yes

havent done much with PDEs/ODEs in Mathematica tho

until now, its been mainly Integration and Maximization

P. Fonseca

I'm more on the system simulation needs... so, some ODE and PDE, but still a little amateur :)

P. Fonseca

1:17 PM

Question posted...

Oleksandr R.

@P.Fonseca +1; I don't know the answer but it's an interesting question

P. Fonseca

1:34 PM

@OleksandrR. Thank you. It actually looks quite simple as a problem, and without it it seems difficult to manage simulation scenarios that go beyond the simple "one bar", "on pipe", etc. I'm almost certain that it is a question of correctly formulating the equations... I should also probably start working with Modelica...

Szabolcs

2:35 PM

Did anyone manage to get any graphical output when using the "EdgeLayout" suboption of GraphLayout?

The docs say,

> A graph layout is typically computed using several stages. With GraphLayout->{Subscript[s, 1]->Subscript[m, 1],[Ellipsis]}, stage Subscript[s, i] is handled by method Subscript[m, i].

So I did

Graph[{1 \[UndirectedEdge] 2, 2 \[UndirectedEdge] 3, 3 \[UndirectedEdge] 4}, GraphLayout -> {"VertexLayout" -> "CircularEmbedding", "EdgeLayout" -> "StraightLine"}]

And got

I simply can't seem to get any graphical output, regardless of the settings.

Vitaliy Kaurov

4:01 PM

New edition is out:

amazon.com/Programming-Mathematica-Introduction-Paul-Wellin/dp/…

Rolf Mertig

4:20 PM

Does anybody know why ReplaceRepeated is not called ReplaceAllRepeated? Since there really is no ReplaceRepeated,i.e. , repeated application of Replace. One has to resort to FixedPoint. The M-language should be rewritten.

P. Fonseca

4:45 PM

@RolfMertig Correction: "Wolfram Language", as defined a week or so ago...

István Zachar

6:09 PM

Does anyone now anything about this upcoming portal: Wolfram Community?
http://www.wolfram.com/community/?actkey=4807-1947-HH5UE9&version=9.0&os=Windows-x86-64&buildid=4055652&source=welcomescreen

Who could ask questions? Registered users? Who would answer? Only Mathematica personnel? Would this be a bit more open than e.g. bug reports to Technical Support?

Dominic Michaelis

huhu

Rolf Mertig

@IstvánZachar Registered users can ask and answer. Users should answer, but personnel might also do that. However, I lost interest. I like this site more.

István Zachar

@RolfMertig So we are not out of business here, good to know : )

Care to share some experiences? Or is it confidential yet?

Rolf Mertig

6:26 PM

@IstvánZachar Just try it out yourself: apply for the private beta and see. There are certainly very knowledgable people there and it does definitely have a use (e.g. suppose you don't want to let the world at large know what your problem is, or you don't want to follow the rules here.) Of course, WRI also has rules. Onle rule I like is that you have to sign up with your real name ... (while here one does not even know the mods names ...)

István Zachar

@RolfMertig Thanks for the information! I decided to wait for the public version.

rm -rf

@RolfMertig Oh please, "real names" can be easily faked... it doesn't take more than 5 mins to setup a pseudonym with a "nice" gmail account. Unless of course, WRI insists on seeing passport/driver's license verification (which can also be faked)

Jacob Akkerboom

6:58 PM

@RolfMertig yeah I agree that it is strange that there is no Replace<notAll>Repeated... I think code will usually faster if you only want to replace expressions at top level if you define a function that calls itself iteratively. See the speed comparison in my answer here (between ggg2 and ReplaceRepeated for some rule) mathematica.stackexchange.com/a/21801/4330 But I do think that ReplaceRepeated will often be faster than alternatives with FixedPoint (or even Nest in case of gg2)

cya all

Sjoerd C. de Vries

7:45 PM

@AndyRoss I was afraid something like that was going to happen when he was asking questions around here. Writing an in-depth review of Mathematica with just a few hours of practice simply won't work. His critique of TransformedDistribution is somewhat justified, though. It works for some basic combinations, but doesn't for many others.

Rolf Mertig

8:27 PM

@rm-rf You need a Wolfram ID. Usually you don't fake that one, do you?

Transcript for