@MrWizard regarding the duplicate: I'm really not sure. On one hand, if we want to have here a new and complete site, we should accept duplicates so that our site is kind of self contained. On the other hand, SO is not out of the world which means that people who are looking here may look on SO too. To be honest, I was very confused at the beginning about the differences of SO and SE.
For instance, the toolbag question on SO contains pretty much all no-go's for a question but it's such a rich resource, that I would really like to have this information here on SE.
Do you think we should ask the SO mods to migrate that question (the PlotMarkers one) to this site? I am not suggesting we close questions on this site because one exists there. However I don't see that it is particularly helpful to post the same answers in both places, which is what I just did.
@MrWizard I think Anna Lear covers that here. It appears that it isn't uncommon for both sites to have similar questions, and after a point, the older one may be migrated.
@rcollyer I skimmed the answers there and I think I am asking/proposing something different. We are not in private beta, I do not want a mass migration. I am asking if it makes sense to migrate an old question when someone asks it again on the new site, then merge.
I would like to be able to browse this website from work (since I use Mathematica professionally), and due to IT requirements, this means using IE8, which is a little old. For some reason, code sections aren't indented properly when I use IE8 on this site, but they do display properly when browsi...
@MrWizard Perhaps might be a lot faster if you go the TL route to short circuit the flag queue...
Everyone: I think this question should be migrated here to complement this answer of Leonid's. He also references it and it might be nicer to have that one here
Alright, thanks :) If you need a migration done quickly (so that folks here don't waste time answering it again), you can ping an SO mod who's online or a comm team :)
Is there's a way to do a partial/pseudo NIntegrate? For example, Integrate[1+a x, {x, 0, 1}] gives 1 + a/2. What can I do to get 1. + 0.5 a other than by doing Integrate[...] // N
I ask because I have a complicated integral that can be solved numerically, but I want to retain one parameter as a symbol so that the result can be manipulated. NIntegrate complains that a is not numeric, and Integrate cannot handle it
for example the matlab code they use for their timing can probably be optimized (I am not a matlab expert though), but looks very similar to what I once wrote. that was faster than anything I could write in mma v7, and I could only get mma faster than that with compilation to C in v8
I'm currently writing an oscillator program, and it gives you the option of generating a Fourier saw and square waves, and you can choose the number of harmonics. I'm using the Wikipedia definition for the sawtooth wave
For a sine function, the maximum value sin(x) returns will always be 1 (and ...
But questions like this can maybe stay closed for 1-2 days as examples of what should NOT be asked? What do you think?
The problem with not deleting closed questions that will never be reopened anyway is that it's still possible to vote on them. People who asked them might continue to lose rep even after the question has been closed. Though this is not really a problem on this site.
@Heike It was more helpful than you might imagine, and I probably could not have done it even if I spend a lot more time (I did spend a fair amount on it). Also, you gave me a pre-packaged function which was very easy to use with my own plots or modify.
@MrWizard I am not hiding, but it is a really really busy time :-( even though my supervisor is skiing today, I have to show some results tomorrow morning
@MrWizard I'm afraid I have no idea. I compiled it from source on OS X; maybe if one has cygwin it will also compile on windows. But then you'd probably also need emacs or vim to edit things etc, so I am not sure how practical the whole thing would be.
@Szabolcs probably, I just compiled the interpreter and use emacs and textmate to write code
@acl Did you notice any obvious unix dependencies that'd be difficult to port? (like overreliance on unix tools like such as sh) I really hope they will try to make something general, and won't target unix only.
@Szabolcs i don't know. I ran make and it just compiled. I can take a look but I don't really know what windows has or hasn't, so I am not sure I can tell
well, if it catches on I suppose someone will port whatever needs to be ported. so far, after playing with it for only a few days and for nothing serious, it seems really really nice. I am not sure how nice it will look after I have to fight with all the little niggles that pop up when coding something big
@acl Well, it's not always that simple. It depends on the intentions of the developers. If they don't care about other platforms, and entangle it with lots of unix-specific things, then it'll end up like Sage and it'll require a lot of effort to port. I'm hoping they'll take an approach like python: let the core be as portable as possible.
i want to know what is the simplest possible way (slight error acceptable) for calculating the Error Function erf(x) or Error Function Complement erfc(x)
I have an exam where in the numericals, while calculating error probability over an OFDM channel, i have to use this function. And i am wond...
@Szabolcs sage is a large complex of different packages, this isn't. anyway, this is about to descend into windows vs rest of world thing which isn't hugely productive
I like cross platform tools. If it doesn't run on Linux or if it doesn't run on Windows are both significant disadvantages to me. Macs are not something I can afford.
@MrWizard to be clear, funny in the sense that we have collectively (the answerers, upvoters etc) expended a lot more effort than it would have taken to just type "autocorrelation" in the search box of the help viewer
I think he's asking questions on his brother's behalf... he had previously mentioned that his brother is in finance, which explains the FinancialData question, this one and a few other rtfm questions
@500 I have myself observed that a number of your recent questions show a lack of understanding of the Help system in Mathematica. Asking a question on the main site that can be answered trivially by searching in Help is not really a useful thing. There is this option for moderators and those with close-vote privilege:
"too localized This question is unlikely to ever help any future visitors; it is only relevant to a small geographic area, a specific moment in time, or an extraordinarily narrow situation that is not generally applicable to the worldwide audience of the internet. "
I think that the case can be made that your latest question falls into this category. I have yet to see anyone tell you to "go away and stop wasting my time" and that is not my intention. However I don't think it is right to clutter the site with such questions. I suggest that first you enter this chat room and ask.
I also think that you need to put some effort into using the Help system yourself. I know that it is not the most intuitive design, but in the long run you need to learn to use it.
I am asking that you, in order: (1) Search the Help system when you need something. (2) Enter this chat room and ask for help using the Help system to find what you need. (3) Ask for help with the problem in this chat room if nothing is found in the documentation. (4) post a question to the main site if it is recommended
@Rojo. Sorry, I feel horrible asking wrong questions + what matters to me is getting my model working. So again sorry. I don`t want to argue and make all of us lose time.
I have some data with 19000 sublists such as :
{" 7.9080000e+01 1.9283193e+04"}
Where the first number is the value for variable A and the second for variable B.
All my attempts to transform this format have failed so far. I think my best guess was using ToExpression unsuccessfully.
How ...
@500 You wrote "So I use :" but I see nothing. Please try posting your code again. Also, are you saying that the file imports correctly, and you only have trouble with Export?
If I google for "adaptive sampling 2D", my question is the very first hit, even if I turn off personalized results ... and I'm still having trouble with that algorithm
The problem with that algorithm is that it tends to give very long triangles perpendicular to a transition region, like this:
What happened there is that I have a big triangle (3 actually) that has one vertex in the "dark" region and one in the "bright" region
1. start with a set of points where the function is already computed 2. compute the Delaunay triangulation 3. for each edge in the triangulation: insert a point in the middle of that edge if the function values at the two ends differ enough (and the points are far enough)
@Rojo Yes, but that only works if all sample points are calculated with that algorithm, and I needed something more flexible that works for any grid of starting points
I think I am simply going to set a limit on the resolution, then subdivide until I reach the limit and there's nothing to do any more, then reduce the limit and continue.
Ugly but should work.
So it should be something that can do a density plot while being able to map contours.
I know that this algorithm is very naive as it is and using just the gradient is not great, but it was good enough here and I really don't want too look at second derivatives. And this isn't the main issue anyway: it is how I subdivide. I think the solution would be avoiding complete retriangulation.
@Heike It does get divided, all large ones. But what happens then is that the triangulation algorithm changes its mind about where to insert the triangle edges and suddenly a new long edge appears between two points that have different values. A solution could be not to completely retriangulate after every subdivision, but to instead insert only as many new edges as the number of the new subdivision points. (...)
@Heike But if I do this, I'll also need to use this custom triangulation for plotting (and not the one that ListDensityPlot computes). The best solution would be to triangulate in a way that avoids long triangles if possible.
@Heike Maybe I can try this: instead of looping though each edge, loop through each triangle and if any two vertices differ insert three new sample points on the middle of each edge, even the one that has the same value on the two ends.
I guess that the behaviour of the code I pasted is somewhat a "good use" of those cacheing bugs related to Update[] that Szabolcs pointed out to me a few days ago...
You can advance a certain number of steps by doing Nest[(Update[]; #) &, factorial[10], 3]
I want it to be useful, but so far I'm not sure, hehe
Yeah, it uses the caching "bug"... If in the condition you put /;(Update[]; guard) instead of /;guard it doesn't work... So probably it's not a good idea to get used to it
There was a nice big square (two triangles) which were all light inside. But then it retriangulated and connected the vertices of those triangles to some other (new) point which is dark.
I simply subdivide all edges that would be changed by the retriangulation
the problem is that if the thing has a high symmetry, edges tend to change a lot
so I first had to compute the convex hull, and slightly shuffle around all points that were not in it (i.e. those that were inside)---to break the symmetry
I must run now because the building is being closed (how tolerant they are depends on which porter is on duty), could we get back to it in about 45 minutes? Could you perhaps mail me what you have (even if it is messy)?
Consider a block (partitioned) matrix
matrix = ArrayFlatten[{{a, b}, {c, d}}]
where, a, b, c and d are each matrices themselves. Say, for example,
a = {{a11, a12}, {a21, a22}}
b = {{b11, b12}, {b21, b22}}
c = {{0, 0}, {0, 0}}
d = {{d11, d12}, {d21, d22}}
How can you find the block inverse o...
When I prepare a CDF file to be included into a web page, far too much blank vertical space is left over the pane. Is there a command to reduce that? This is illustrated by the files SimpleDemos.nb and SimpleDemos1.cdf at http://www.abstractmath.org/Mathematica/
@halirutan Any news about the syntax highlighter? Do we need to wait until it's integrated into prettify before we request the SE team to integrate it into this site?
@TimStone Do you know what the procedure is to get the syntax highlighter integrated into the site? Just a meta post with feature-request?
@Szabolcs No, unfortunately not. But as far as I know they only need to put the language extension file into their installation of prettify on the server and maybe set our default language to Mathematica. But I'm not really sure about the procedure.
Just link to the highlighter-only source and explain what's going on, and hopefully someone will look at it at some point soon. Actually, that reminds me to look further into that MathJax issue, which I think is still a problem right?
If you can get it pulled into Prettify directly beforehand, that's probably even better, but I don't think it's strictly necessary.
@Szabolcs Sorry about the post with Insert, but as I was thinking about it, you and Leonid came right into it with your answers and covered the basic approaches. This left me we something which was equally slow but I hoped without testing, that maybe Insert in combination with Fold is better than his. I recheck it later.
Possible Duplicate:
Clear complains that a variable is not a symbol or a string?
Clear["Global*"]` Works for most variables, however if i do something like
$ \delta_a = \alpha_a \Delta T d $
Then the Clear or ClearAll will not clear $ \delta_a $, short of Quit[] is there anything tha...
i want to know what is the simplest possible way (slight error acceptable) for calculating the Error Function erf(x) or Error Function Complement erfc(x)
I have an exam where in the numericals, while calculating error probability over an OFDM channel, i have to use this function. And i am wond...
Is there a built in function to compute the autocorrelation of a signal ?
Consider a block (partitioned) matrix
matrix = ArrayFlatten[{{a, b}, {c, d}}]
where, a, b, c and d are each matrices themselves. Say, for example,
a = {{a11, a12}, {a21, a22}}
b = {{b11, b12}, {b21, b22}}
c = {{0, 0}, {0, 0}}
d = {{d11, d12}, {d21, d22}}
How can you find the block inverse o...
When I prepare a CDF file to be included into a web page, far too much blank vertical space is left over the pane. Is there a command to reduce that? This is illustrated by the files SimpleDemos.nb and SimpleDemos1.cdf at http://www.abstractmath.org/Mathematica/
Possible Duplicate:
Clear complains that a variable is not a symbol or a string?
Clear["Global*"]` Works for most variables, however if i do something like
$ \delta_a = \alpha_a \Delta T d $
Then the Clear or ClearAll will not clear $ \delta_a $, short of Quit[] is there anything tha...
