so, in case anybody can feel my pain, after 22 hours of computation, using 20 cores, and consuming 320GB of RAM, all I got is: "Write failed: Broken pipe"
@acl simulations I was running on the cluster: 32 cores, failed to produce anything as the save routine wasn't getting the correct directory to save it in, so it refused to overwrite. I feel your pain.
@acl I honestly wasn't trying to engage in one-upmanship here, only trying to highlight shared pain. Unfortunately, I can't figure where the <censored> thing is screwing up.
@rcollyer that's the problem, neither can I. which pipe is broken and where? and how am I supposed to work it out? terrible. I should have become a doctor!
@acl Among my friends (whom are physicists), we actually treat the M.D.'s as if they're the ones that aren't "real" doctors.
@acl The worst part is that it is my own code that is going wrong, not the third party application here. It is somehow worse that I don't get to blame them.
Good luck finding the error. I now have to fix code that produced some of the plots in my preprint, likely affecting their interpretation. I love bugs.
Over on Computational Science, I was able to suggest using mma for capturing ones notes on a project. Maybe they'll take the hint? Oh, and I also brought up the whole mma v. matlab speed thing just for good measure! :)
@rcollyer You might also want to link to this answer and point to the PDF in the answer... pretty stunning how neatly (and quickly) you can take down notes in real time if you are deft with mma shortcuts and typesetting capabilities
@yoda you're correct in that assessment. Yet, I view each and every launched site that we bypass as further evidence that we'll make it, most likely sooner, rather than later. Especially considering that they're a lot older than we are. Currently, I view Skeptics as the one to match. I would be comfortable with 5k views/day, but we'll beat them in terms of questions long before we hit that point which would put us at what Physics is at at the moment.
@Mr.Wizard I posted a new question about code reviews because in the previous thread there was no discussion about code review, only about code golf, and also because I think it's better to keep the two discussions (code golf and code review) separate.
@Heike yes, he got a little bit of a tough time over his original question, though it had a fair amount of information in it, he probably felt there wasn't much interest. He had quite a short deadline - but it if had remained perhaps he would have found some interesting ideas.
@image_doctor I didn't fully understand the question, but it did seem to have some interesting aspects. I think the main problem was the unwieldiness of the data file.
@heike I had downloaded the data and used the posted code to unpack it, it wasn't too big less than 20MB packed. I'd started building some generative models from the data, as whuber suggested I wasn't too sure a Gaussian based approach was optimal.
@whuber, I am actually gonna a go with you I don`t think this simulation makes sense anymore. I have to implement the effect of the Center of Gravity. Thank You very much again.
I went to my fifth driving class today, and I feel like I barely made any progress. I wouldn't dare go out on the road without an instructor. I just cannot understand why the U.S. driving school I contacted only includes 6 (six!) hours of behind the wheel practice...
@Szabolcs They've been experimenting with self-driving buses in the place where I live, but these buses have their own dedicated bus lane and there is still a bus driver present.
@Szabolcs Spending some time in Edinburgh should cure you of that, most of the inhabitants regularly step out in front of cars. Presumably under the assumption that they will stop ... as Edinburgh's population isn't in terminal decline, I guess they are correct :)
@Heike Is the answer to this one to just build the Manipulate controls with Dynamic and Slider directly? I can do it with Dynamic, but I don't use Manipualte that often ...
@Szabolcs That's how I would do it. To be honest, Manipulate is still black magic to me sometimes. Every time I think I understand Manipulate it crashes horribly on me taking the whole front end with it.
@Szabolcs Yes indeed! You wouldn't want to drive much in London on a normal day, never mind during the Olympics! I have tickets for the kayaking final, but that's outside of London and even that you can't get any nearer than 5 or 6 miles and then have to take public transport.
@Heike Just plain Ctrl-K-ing. There was a recent question about extracting the viewpoint from a 3D graphics. I want to make a palette that does that easily and automatically. I was looking through the AbsoluteOptions[NotebookSelection[]], and there is a Graphics3DBoxOptions there.
@Heike Neither for me after a kernel restart. Do this: Graphics3D[Cuboid[]] then Button["press me", opt = AbsoluteOptions[NotebookSelection[]]], select the cuboid (not the cell, but the graphics box), and press the button. Then try auto-complete again.
@Szabolcs I'll try that later. I have to go to a reunion now. I just hope the weather forecast is wrong about the heavy thunderstorms they're predicting for tonight.
@Heike Have fun! Something possibly related is under Texture[] doc page, Applications, Sky Box. There's a Method -> {"RotationControl" -> "Globe"} options for Graphics3D
@Szabolcs So, it appears to be created when you select something. Note, for me it shows up as Cell$$8028`Graphics3DNavigationFunctions as I'm running that nb under CellGroup contexts.
I can also find it using AbsoluteOptions[$FrontEnd]. Warning!! This locked up my front end a few minutes ago, and even if it doesn't, it shows some errors that might indicate something is going awry
AbsoluteOptions[EvaluationNotebook[]] looks safer. I'll try to modify the option for a single notebook now.
`SetOptions[EvaluationNotebook[], Graphics3DNavigationFunctions -> Print]` does not print anything while rotating. Neither does `SetOptions[EvaluationNotebook[], Graphics3DNavigationFunctions -> {Print}]`. (The option name is plural.)
Has anyone used a game controller or some sort of 3D navigator with Mma?
@rcollyer This one. No, it's just something I found interesting looking. It's very simple, just a few random walks, superimposed, with a slight transparency. I was looking at how transparent lines overlap themselves.
@Szabolcs Very good question! It is supposed to be a supported card. I have a working CUDA installation. But CUDAQ stubbornly returns False. I even tried asking in a Wolfram Seminar on GPU computing. But it was never resolved
@image_doctor A friend who's also using Linux said that the problem with his NVidia-based card was that while the graphics card is using NVidia's chips, it was not built by NVidia, but some other vendor (maybe the same one who made his laptop?). So not all cards using the same NVidia chips are compatible with each other. He said that the vendor changed some details that NVidia's Linux drivers didn't support. I'm not sure about the exact details.
A couple of months later NVidia updated their drivers and now he can use CUDA on Linux with his card.
@Szabolcs I have an ASUS card, it probably one of the 3 most common cards of that type. I can't recall if the exact version was on the supported hardware list.
@Szabolcs I really appreciate your offer :) But I fear you'd both just be wasting your time, I spent about a week trying to nail it, with all the howto guides, input from Wolfram. All to no avail. It probably needs a wave of the magic wand from Wolfram CUDA guru :)
Does this
Position[#, First@Nearest[Flatten[#, 1], {0.32, 0.65, .8}]] &@
ImageData[tree]
(*
{{162, 74}}
*)
do what you want?
OK, try this:
tree = ExampleData[{"TestImage", "Tree"}];
dat = Reverse[ImageData[tree]\[Transpose], {2}];
dim = Dimensions[dat][[{1, 2}]];
nearfunc = Nearest[J...
pretty amazing how easily you can play with things in mathematica
of course, for larger-scale stuff, I am currently fighting some sort of memory leak (probably the symbols leaking from Modules that were discussed here some time ago)
I need to create a Manipulate with two control parameters which are linked by some mathematical relationship. So the user can decide to use either control and when that control is changed, the other will change too.
The example below works as required, using If statements to determine if one of ...
@acl On the other subject, I see "large-scale stuff" more like a symptom than an actual defect. When well managed, it can pass almost unnoticed (Wolfram|Alpha huge "stuff" is, apparently, a good example of well managed Mathematica programming implementation).
@PFonseca to be honest, I do not have much experience with large-scale programming in mathematica. but in this case I am seeing a linearly (with time) increasing memory consumption. my last attempt ended up using almost 800GB of RAM before it finished, and it simply does not keep anything that large in memory.
(assuming you were referring to my recent troubles with mma, of which I had complained here...)
So I guess the question is: was the memory usage a bug from the integration implementation, or from the garbage collector? We definitely need a modernized garbage collector...
Lately I have a growing tendency to think like in Julia project: I'm greedy. Why can't we have it all? A modern garbage collector with all memory leak difficult to manage scenarios treated; a groundbreaking JIT compiler as it happened for JS; a fast user rule dispatch; multicore everywhere where it's possible; a more natural implementation of parallel processing;
two times more Dynamic interface objects, working 10 times faster; a better integration with engineering tools, and more notoriety in the industrial world; well implemented built-in object paradigm function; and, most important, a better first word suggestion when doing spell checking...
@Heike I actually went to see the amount of memory I had, to make sure I was not mistaking letters... It's kind of late here...
Cell groups generally have a little triangle next to the outer cells that you can click to fold/unfold the child cells... Mine are missing all of a sudden:
Does anyone know what they're called so that I can look for it?
Oh well. Never mind... I just found out Cmd-' does the same and I needn't lift my hands off the keyboard
Does this
Position[#, First@Nearest[Flatten[#, 1], {0.32, 0.65, .8}]] &@
ImageData[tree]
(*
{{162, 74}}
*)
do what you want?
OK, try this:
tree = ExampleData[{"TestImage", "Tree"}];
dat = Reverse[ImageData[tree]\[Transpose], {2}];
dim = Dimensions[dat][[{1, 2}]];
nearfunc = Nearest[J...
