In light of this question and its pursuant comments: Functions tag -- best usage?
Please consider questions of this type:
Functions vs. patterns
How can I create a function with optional arguments and options?
Passing down arguments
I believe that these are distinct from the greater body of ...
@Szabolcs I don't like the second one (with SetDelayed rather than Set). It's not immediately clear (at least to me) on what symbol the pattern is being defined, because of special handling of List.
It seems like a convenient solution when a function will return mutliple results as a list and these need to be assigned to a function
I use the {a,b,c} = {1,2,3} unpacking syntax all the time (in particular when functions return mutliple arguments), but to see it used with DownValue definitions, that was very surprising. I have never used that.
It is, it just somehow never occurred to me that it can be used with DownValue definitions as well
@JM What I'd sometimes like to have is this: {a, b, ___} = valueList. This of course doesn't work but what I mean is: ignore everything above element 2. Or more complex: {a, _, b} = triplet --- get the first and third, ignore the middle one. In these cases I use valueList[[{1,2}]] or valueList[[{1,3}]]. Any better solutions?
@Szabolcs IMO, it should be {a1[x_], a2[x_], a3[x_]} := Evaluate@CoefficientList[(x + a)^2, a]. At any rate, it's convenient, but an exception to the general rule. For instance, q[a1[x_], a2[x_], a3[x_]] := Evaluate@CoefficientList[(x + a)^2, a] is an assignment on q, but the first one isn't an assignment on List (thankfully).
@OleksandrR In what way is {a1[x_], a2[x_], a3[x_]} := Evaluate@CoefficientList[(x + a)^2, a] different from {a1[x_], a2[x_], a3[x_]} = CoefficientList[(x + a)^2, a]?
@Heike oops, sorry, yes, my session got mixed up a bit.
As long as x has no value when the definition is made it should be OK.
But I'd still like to know how it is that the downvalues are indistinguishable between the Set and SetDelayed cases, yet Information still knows the definition was made with Set.
I've seen it before. :) Actually, I wrote my own version years ago to see what was going on while I was crunching the numbers, but never got around to publishing it...
(I'm pretty sure every single one of those could be done in Mathematica, with judicious use of FixedPoint[] and DensityPlot[]. And possibly Compile[].)
@acl Not Newton's, but this, and for a non-polynomial function. Now look what I get when I calculate the same thing but I set RuntimeOptions -> "Speed" in compile ...
One problem is that many people from MathGroup still don't know that there's a new site, and when they see an URL with "stack" in it, they think "Oh, it's just SO, I've seen that and it wasn't so great" (Someone actually told me something similar in email)
That's a shame, but frankly we are doing all we can to leave sign posts to the new place, and if they cannot be bothered to read them I question the quality and persistence of their potential contribution.
@JM and @all Can someone please check this for me? Do you have any hints what's going wrong when I set RuntimeOptions -> "Speed"? (This was a quick and dirty experiment to plot some fractals using that root finding method) ge.tt/8lrV5lD?c
@Szabolcs stackoverflow.com/q/9319056/1078614 here for instance. The trick with sequence is really nice. Here you had gotten at least 10 upvotes. Not to mention, that I got an "Thanks, I like the approach" but no upvote..
I wrote lately a C++ Wolfram Library binding for a Dynamic newton fractal viewer. I think I posted it already anywhere on SO
@halirutan I learned that delete-duplicates trick from an old version of the docs. Did you write the core code in C++ for that? You should post it somewhere!
@MrWizard This is exactly the reason why I hate it when people add information in comments... Comments are not versioned, not un-deletable, not searchable, ephemeral and its existence is always at the mercy of the commenter.
Also they are searchable via Google, etc. But I see your point.
@halirutan speaking of low vote totals (warning: here comes some blatant rep whoring) I think this answer of mine deserves a higher count if it works robustly, and if it doesn't someone really should have the courtesy to tell me.
@JM I did, I even upvoted it :) I've always thought of doing that, since jet has a simple piecewise linear structure... but as you must have realized by now, my "I want to do it" thoughts don't always translate to action :P Thanks for doing it :D
@MrWizard The behavior of questions with so many answers is sometimes surprising. It seems people don't go through all answers carefully enough and what is last is the list, stay behind, since it is seen too late.
While I'm waiting for Szabolcs's plots to render, I have a quick question for people who still remember old Mathematica: there was once a nice graphic displaying the colors in Graphics`Colors` in a neat grid. I can't seem to find the code for that now; anybody remember?
I'm not sure how to draw a triangle if all I care about is the length of the sides. (I'm happy to place one of the vertices at the origin and place one of the sides on the nonnegative side of the $x$-axis, but that doesn't really matter.) Is there a straightforward way?
Given a triangle with...
This for sure :) Both the other answers are not complete!
I haven't voted on that one but it's on my "to review" list (what I've been using Favorites for on the new site, in addition to some actual favorite questions).
@JM well, math.se definitely has more volume... and I got the impression that folks generally compartmentalize themselves and restrict to their subject area
I don't think the downvote was appropriate here. I am happy that this is a much friendlier community than SO and I want to keep it that way. This had to be closed, not downvoted.
I don't like to downvote when the intention is good
@MrWizard What Szabolcs said. If you just type @@, it should give you a pop-up. Choose the second option (network user) and enter his user ID. You can't use the first because I assume he hasn't visited chat and created an account yet
@David Because many people don't have access to antialiasing. You need a good graphics card with good drivers to have it. (And you know the situation on Linux)
@David I have Intel integrated graphics, which means no antialiasing on Windows, and I'm happy if it doesn't crash on Linux (even though the Intel drivers are actually opensource---but something's wrong with then in Ubuntu 11.10)
Which reminds me --- I must get some work done before my supervisor returns from his skiing trip ...
@David I'm the one guilty in making his confetti so slow ... just remove the antialias part---your graphics card does support it natively anyway (and the minor quality loss is negligible for animations)
ImageResolution is the correct option to use here. It still has the problem that tick marks in 3D won't scale properly. I think this is a bug.
I'll also make fonts blurry, so it's not really ideal. But I did try to implement a simple subpixel scaling to ameliorate the problem. I don't have the code here now though ... and MrW complained that on his rotated screen there was a lot of colour fringing (of course, since it wasn't horizontal RGB)
The subpixel scaling was based on this: code.google.com/p/windjview-subpix (if you read the algorithm, it's obviously inaccurate and thickens lines slightly, but in practice it worked well for me)
And even if it's just a remainder that what you do can be represented by something pretty every once in a while.
>Bachelors, Masters and PhD's available in your field! Eliminates classrooms and travelling.They are fully verifiable! Get a chance to obtain a diploma of your dreams! With our help absolutely any of your dreams may become real and you don’t have to waste your time on that! WISH TO DOUBLE YOUR INCOMES? Feel interested already? Want to know more? Then contact us right now!
@Szabolcs on darn, forgot to wear the moderator hat. Okay it's probably not the most fitting question, but darn, it's going to come up sooner or later. Now I've put my foot in it.
J.M. I make my answer a CW so that I am not getting points and so that if anyone has something they simply must add they can. Does this seem reasonable?
oh hmm. I'm used to the v8 one now, and didn't realize it was different from v7. I wouldn't be surprised though, because Image processing functions got massively improved (and added) in v8
My point is that this was not a question. It was put in the form of a question, but the motivation of the asker came from annoyment, and not wanting to solve a problem. There's no good answer that can be accepted and there's no good answer that can be useful to people (other than "yes, others are annoyed too, and others are complaining too, but there's still no undo")
Regarding whether or not this kind of question can be answered I think it reasonably can be, but it requires either an official answer or something carefully reasoned. I have myself asked question about the design of Mathematica and I hope that others will be able to do the same without getting hammered. This one has been argued so much that I think the only answer can be to direct someone to past arguments.
speaking of that... is it possible for you to say, copy a particular function (built-in) from v8 and run it on v7 (assuming you have all other functions it depends on)?
Here is a question of mine that I would not like to see closed, for example.
@yoda I suppose if the code is available in a .m package or by unsetting ReadProtected, and the code does not depend on anything else that is unavailable by these means (or easily implemented) then yes.
@MrWizard This is simply my personal opinion, and others will have different opinions, but I very much do not want to see such questions here. One can go to the mathgroup for that.
I don't think the real question (the motivation behind the question) was "Why did they decide to do this? What's the advantage?" I think it was "Do you agree with me that this is so annoying?". Every other word processor has multi-level undo, at this point it's quite clear to anyone that there's no good reason not to implement this except maybe technical difficulties (the design of the front end may make it difficult to implement)
@Szabolcs Okay, let us grant that. What about a question like mine?
@Heike that one I assume is because NestWhile allows one to write certain common loops in a more Mathematica way; "FoldWhile" doesn't match an existing paradigm that I am aware of, but I believe it can be fairly easily implemented with Throw/Catch or Return.
@MrWizard Not at all, I am referring to the undo thing. Yours is thought-provoking (although again it is not "I am trying to solve this problem") and the answer you accepted was insightful. "why no undo" is not thought provoking and I literally cannot imagine what a helpful answer would look like (apart from "I work at WRI, am responsible for all frontend development, and I won't put undo because X", which won't happen and anyway is not the kind of answer this site is for, in my opinion)
@Szabolcs How does it make it difficult? If I ignore the state of the kernel, then it becomes matter of multi-level undo in a fancy word processor. If I don't ignore the state of the kernel then it is a different story.
What @acl said. @MrWizard Your question looks like the motivation behind it was to learn/understand the design of the language (and perhaps become a better programmer). To be honest I don't think a lot of interesting things would come out of your question, but I would not have though of closing it because it looks like an honest question that stems from thinking about the design decisions behind the language (and not a complaint)
@acl If they haven't given a concrete answer on MathGroup after participating all these years, they probably are never going to do that here either... so better to close than invite rants and guesses
@yoda Yes, I simply ran out of space to add that. Also perhaps someone (a moderator, or one of the closers) could explain to the person having asked the closed undo question that it was not closed because "undo is not possible" but because X Y Z. Otherwise it looks like an arbitrary closure.
@MrWizard your question is also relevant when you're thinking about implementing a package and the kinds of functions that it should export. It has relevance.
First the background. I'm trying to solve for the roots of a rather messy complex equation. This is not the exact equation, but it's a decent (simpler) stand in:
Tan[x - I a] + I == I (x - I a)^(-1/2)
I can use FindRoot to solve for a particular root, i.e.
FindRoot[Tan[z] + I == (I) (z)^(-1/2...
I think your suggestion might work (it works for the original sample I suggested), but I'm not sure how it might be generalized for the more complicated example.
One should really try to do as much analytical effort (root asymptotics, etc.) as can be mustered if one is going in the business of solving transcendental equations repeatedly.
@JM OK, but within ten minutes of it being posted, two people think that the OP's comment "Interesting that this was closed even though links to five satisfactory, concrete answers were provided. I'm glad the closers waited long enough for those to get posted. If the answer to a question is "that's not possible," does it cease to be a question?" is a great comment.
Well I was thinking as long as in fact none of your functions is an InterpolatingFunction[] or something, there might still be hope for analytical attack...
In light of this question and its pursuant comments: Functions tag -- best usage?
Please consider questions of this type:
Functions vs. patterns
How can I create a function with optional arguments and options?
Passing down arguments
I believe that these are distinct from the greater body of ...
I'm not sure how to draw a triangle if all I care about is the length of the sides. (I'm happy to place one of the vertices at the origin and place one of the sides on the nonnegative side of the $x$-axis, but that doesn't really matter.) Is there a straightforward way?
Given a triangle with...
First the background. I'm trying to solve for the roots of a rather messy complex equation. This is not the exact equation, but it's a decent (simpler) stand in:
Tan[x - I a] + I == I (x - I a)^(-1/2)
I can use FindRoot to solve for a particular root, i.e.
FindRoot[Tan[z] + I == (I) (z)^(-1/2...
