I've created the following scene with a Chinese-style building surrounded by trees, and a horse and a rabbit grazing on the grass in Mathematica (don't ask me why there's a bust of Beethoven in there...). The Chinese-style roof was created using J. M.'s equations in this answer. The code to creat...
@EliLansey The reference I remember offhand deals with $\cot\,x=\kappa\,x$, but the techniques there should also be applicable to your transcendental equation...
Like I said, just a little thing. :) The other parts look fine. :)
(On that note, due to the convenience of two-argument arctangent, a lot of times I try checking if I can use it instead of other inverse trigonometric functions...)
@JM The full question I had written out (which is why the first paragraph says "code at the very end"... currently the "very" seems out of place) included a hawk on the tree eyeing the rabbit... the goal was to move the rabbit around to a position where it's shaded from the hawk's view. more than 10 s in its view and it's dinner.
I removed that part because I felt it might be too much work to expect anyone to do, especially when this is not really a problem that I face
Very likely... I'm sure that the current one by itself is a bit too complex to allow for real time flying on moderate cards...
Hmm... why is a function f considered to be "enclosing" g only when written as f[g[...]] and not as f@g[...] or g[...]//f, even though they're equivalent?
@JM I was hoping to avoid that :D... but a serious example: items Sowed are collected by the nearest "enclosing" Reap. Yet, the following give different results
R.M, not sure if it was a question or not, but the ones with Function don't reap anything because the argument is evaluated first, so that format invalidates the HoldFirst of Reap
There was a lot of talking on who was going to play Jobs on his biopic (done). Then there was the choice of writer (done). Now that all the small players are chosen, who is going to play S. Wolfram? If he appeared just for a couple of seconds, or if Mathematica was mentioned on the NEXT or the SIRI scene, that would be a huge commercial spot!
I've created the following scene with a Chinese-style building surrounded by trees, and a horse and a rabbit grazing on the grass in Mathematica (don't ask me why there's a bust of Beethoven in there...). The Chinese-style roof was created using J. M.'s equations in this answer. The code to creat...
@Szabolcs It's a great 3D scene. In my wish list (you know I have a huge wish list...), there's bump mapping and real-time raytracing (or at least, for the time being, a faster renderer with shadow casting).
Also, real world physics simulation associated to 2D and 3D objects would be nice... CFD built-in stuff was "promised" for M10...
@PFonseca Probably not real-time (i.e. synchronized to the clock), but MathModelica might have good performance simulation. I've never seen it. They bought the company making it, so maybe we'll see something in 9? Or it will stay as a separate product ...
@PFonseca Really? Where did you read it was 'promised' for 10?
@Szabolcs I no longer remember. But I think it was near the M8 launch. If I remember correctly, they talked about solids FEM kind of stuff probably for M9 (M8 already shows tetgen as a package), and CFD easy kind of stuff, probably just for M10 (they just added the CUDAFluidDynamics toy on M8).
@Szabolcs also, one of the creators of Imtek Mathematica Supplement has been working at WR for some years now.
@PFonseca In 8, ?**FEM*` returns many results. I don't know about 7 ... So at some point they were definitely thinking about FEM. I'm curious if 9 will have it. It'd be a big feature.
@PFonseca Does it have more than one creator? :-) He's pretty active on this site too.
@Szabolcs I would love to have this kind of functionality. Not that I'm expecting it to compete, in production environment, with tools like openFoam (or Ansys stuff), but at least, on my personal interest of learning a little bit more on the subject, this would be great!
@Szabolcs I don't know if there was more than an author, so I tried to stay politically correct...
But we never know... This type of computations are difficult to implement, more on the attach all together aspect, than exactly on the difficulty of it (whatever that means). So, being M a great integration package, probably it can be made into a pretty useful 3D open surface simulator (my subject of interest...)
Although, the pov can be estimated by examining the deviation from vertical for the sides of the jar. Similarly, the the angle above horizontal could be estimated by looking at the eccentricity of the top of the jar.
@JM yes, the tri-dactyl one does not appear to be around at the moment.
@rcollyer Well, I'd like to see dsp.se grow too... but I've grown distant with the site. We really lack some good image processing experts (we're doing good in the signal processing part)
Reminds me of the old joke: "This is an old picture of my grandfather. Yes, he is behind the cow he is milking. Could you maybe remove the cow, so I can find out what my grandfather looked like?"
@rcollyer I am not sure. if, for instance, we know that the label has a rectangle on it, perhaps this amounts to a) a routine to shift perspective (which I think mma includes), b) a routine to "unwrap" the label, which sounds like a point transformation, c) optimizing the inputs to those two in order to get a rectangle
so it is a mma question, if we view it like this, isn't it?
@Jagra Nice. I was using a not-too-crowded cluster, and I tried to let Mathematica manage how many kernels (i.e. how many processors) were in use at any given time. It used all processors if no one else had jobs in the queue and it shut down some to use only 1/3 or 1/4 of the processors if people wanted to run their own jobs
@Jagra It was quite hackish, and I would never have thought of doing this on a crowded cluster with long queues or many users. Fortunately I knew all the other people using it, and at that time it wasn't 100% used.
@Jagra In my program I was shutting down and starting up new kernels only after a full parallel operation (usually ParallelMap or ParallelTable) has finished running
@Jagra Thne question was about: what if I do this in the middle of a ParallelTable running? It seems that sometimes if I interrupt the whole thing (Alt-,), launch more kernels, then resume (Return[]), then it will use the new kernels. I'm not sure how safe this is.
So I didn't really do it much (since I had to make sure my program works reliably)
The bigger question is: what if I kill kernels instead of starting up new ones?
I didn't figure this out completely
there's a setting somewhere controlling what should happen, and it is different for different situations (cluster, grid, local multicore machine, etc.)
let me see ...
sometimes it re-launches new kernels and tries to use them
For my application I had to prevent re-launching new kernels, and somehow try to force it to redistribute the jobs among the running ones. I didn't manage to do it reliably, and I wouldn't feel safe about doing it...
@Szabolcs -- I know, while Wolfram, may have worked all of this out, it seems like we don't have all the documentation we need to both understand and control these kinds of things.
Oh, and the cluster kernel launching documentation and package (ClusterIntegration) was not very pleasant to work with ... I ended up reading the code a lot, and changing some things in the package ... (i.e. just hacking a solution together)
@Szabolcs -- I'd think that Wolfram would have built in support for losing kernels and re-aquiring them during calculation. Even the earliest parallel computing infrastructures supported this. If so, killing a kernel, shouldn't present a problem.
@Jagra Did you see the option Preferences -> Parallel -> Try to relaunch failed kernels?
@Jagra What I can say is that in a very basic test, when I had something like ParallelTable[Pause[1]; $KernelID, {20}, Method -> "FinestGrained"], I could interrupt, launch/close kernels, and everything seemed to work fine. But when I tried some more involved things, such as calling a shared function (shared functions always evaluate on the main kernel) from a subkernel that might close/launch kernels, bad things started to happen.
@Szabolcs -- Sounds like something to revisit. One of the nuisance problems I have is when I try to abort an evaluation in the front end because I've done something wonky in the code and it won't stop running or it just keep generating errors. I frequently need to restart my servers. Do you know a way to specifically abort processing on the remote kernels?
@Jagra I don't remember exactly when went wrong but I do remember that I didn't feel safe closing/launching in the middle of a parallel evaluation. So instead I just split up the parallel evaluations and closed/launched inbetween.
@Jagra Why do you need to restart? Can't you just kill them? I remember that when things went wrong during an abort, sometimes they went very wrong, and some subkernels generated huge expressions that filled up memory, and it was necessary to restart the machine (they didn't respond because of swapping)
So I made sure that the computation would abort completely on any message generated
this can be annoying
but at least the machines won't lock up ... (it was a problem for me because I didn't hav the key to the server room)
Mathematica is a bit unusual as a programming language because it never stops on messages, regardless of whether the message is an error, a warning or just an informative message. It simply prints the message and continues with evaluation. This can cause an avalanche of messages and unpredictabl...
@Szabolcs -- In my experience, in situations where I wanted to start over by killing the remote kernels, I would some pretty serious degradation of performance after restarting the remote kernels, as if some buffer didn't clear. Also, sometimes an abort didn't seem to reach through to the remote kernels.
@Szabolcs --You gave me a couple of ideas to pursue. Thanks. Gota run now. Best, J
I posted an answer on StackOverflow that gave an index of all the "New in version X" pages for Mathematica, but now I cannot find it. Does anyone recall what question this was in reply to?
@JM I think it is to open-ended. There isn't a correct answer. And while there are 1000+1 ways of doing things in MMA, they're all "correct" in a sense that doesn't apply to that question.
@MrWizard I think that might've been closed just after you posted that answer, and I might've voted to delete it too...
It essentially was a what's the list of pros/cons/changes between versions or soemthing like that.. (if I remember correctly)
Nothing of value is lost, really... it was a link to a website then and you might as well link to the same website now instead of linking to your answer which links to... you get the drift
@MrWizard well, you could just ask an SO mod to do a search for "user:618728 [mathematica] deleted:1 is:answer +new +version" (without the quotes) and see if it's there..
My biggest issue is how broad in scope the question is, and is hence not answerable. A narrower scope may be able to bring it within range of what we can answer within the scope of Mathematica.
Forgive me for posting this, but a little physics humor from reading [this](http://arstechnica.com/science/2012/05/the-state-of-particle-physics-a-report-from-pheno-2012/) :
> SUSY pairs each particle of the SM with a supersymmetric partner. These have unfortunately comic names: quarks are paired with squarks, W bosons are partnered with winos (pronounced "WEEnos"), and so forth.
@all: any thoughts on reopening this? It is a reasonable question and not very different from most other questions asked (re: implementing some functionality) and it seems like the close votes were piled on after F'x's comment (which wasn't very off the mark either — Alan does have a habit of issuing "calls for implementation")
@belisarius I'm not disagreeing... just there's a valid question, since quantile functions always give the smallest x for the given p and intervals might be useful (I've had a use for them at some point). I'll cut through the fluff and trim it down to just that
@JM I agree, this NKS stuff is completely independent of mathematica
if I asked "how can I use cellular automata to simulate hydrodynamics", would that be on topic here? because that's a more focussed question (yet still way too broad), and also has nothing to do with mathematica even though it may in fact be answered using mathematica
anyway, I agree with the question being closed (should anyone care)
@acl Heh, neat. I didn't mean I will ask it now; just that if I had to ask that question, I'd choose scicomp.SE, not this site. And yes, the NKS thing is independent of Mathematica.
How do I get (using Mma) this:
from this:
Unknown POV, unknown cylinder radius, unknown jar content.
