@Szabolcs I saw you've been adding In[] to posts lately... any specific reason? I thought that we went through the whole trouble of discussing and wrapping output in comments and leaving away the In[] part so that we can easily copy the whole block and paste without having to sit and delete stuff
I followed the link in "migrated from math.stackexchange.com" at the bottom of the post you linked, which takes me to the original question on Math.SE. Don't you have that link too?
I am having hard time understanding Trott code in his books. Too advanced for me to understand. Too many nested calls. Too bad, because it does amazing things, but reading it is hard :(
@Mr.Wizard The real funny thing in First[Timing[prime*1*^100000000]] is that you feel the 2 seconds of calculation while Mathematica tells you it took nothing ;-) I call this distortions of perception
@NasserM.Abbasi Write it down and take it apart bit by bit.
@halirutan, I do break code down to understand. But it takes so much time. Mathematica experts code is hard for me to read due to all the nested calls. The other day I spend 1 hr just to understand one small function since it had 5 or 6 nested calls.
If I hadn't been running calculations in the background I would have noticed the time taken but I just chalked it up to general slowness. I think this Option will help.
@halirutan I have all his books, too many examples. But was reading his Graphics books. Amazing stuff he does. But hard to understand how he does it since code is too complex
@halirutan, it is not just one specific page or specific function, but general observation on the code in his books. ALso comments are spread between code, makes it hard. I am just saying code is too complex to just read and understand as I read it, like say when I read Java or Pascal or Matlab or other code, which I can sort of follow as I read. With Trott code, I can't just read it and just follow it. I have to stop at each line, break it apart to understand the logic.
But it is amazing code. I just wish I can understand it more easily, that is all :)
That I understand. I prefer electronic (preferably Notebook based) Mathematica books for that reason. It's just too difficult to parse a large chuck of code in my head. (Especially when it has too many brackets.)
I tend to use # too much myself, since I like brevity. It would be hard to read my code in printed text too I suppose, as I rely on Ctrl+. fairly often to resolve ambiguity.
@NasserM.Abbasi That's some rather poorly line-broken code IMHO. I would have expected better for printed publishing. I suspect that would be easier to read with a different layout. I think I'll try my hand at it, just as a matter of curiosity.
@Mr.Wizard, please do not do that for me, unless for yourself. I am not asking someone to explain it to me, it was a random page. I was just showing an example as I was to asked to show one.
@NasserM.Abbasi I didn't mean it that way. As I said: "just as a matter of curiosity." I'd like to see if what I come up with is any easier to read (for a third party) than the original.
@Mr.Wizard, sure thanks. I just did not want you to waste your time on my expense if you were going to do it just to help me understand it. For you, it will be easy to understand I am sure. For me, it is struggle.
No, actually it's not easy to read, and I haven't even tried to understand it yet. I feel that the formatting is seriously lacking for code that is intended for printing. This code in a Notebook should not be that hard to parse with the combined tools of syntax highlighting and Ctrl+. (or multi-clicking if you prefer). I suspect it was written for Notebook publishing (as I believe the Guide Books were) and printing as only an afterthought, which is a shame.
And the problematic thing in this snip is not only the many maps of pure functions but the little ( and ) which are easily to overlook.
Basically the code extracts points from a graphic and mangle them through different functions, I assume to implement some nice effect. The final points are then fed back to the Lines at the beginning.
@halirutan for me code readability is the most important thing. After all, we read code much more than write it. I think Mathematica code can be made much more readable by breaking logic into more steps instead of this obsession with deep nested calls. By making more steps, and each step is less complex rather than one big complex step, it makes the code easier to understand. This way, one can see the intermediate steps/calls being made, and understand the logic more.
@rm-rf In the reply by WalkingRandomly it's not clear which is the input and which is the output, so I thought it's important to clarify. Also, there's a single input line, so copying is not an issue.
Was there another instance where I added it? (I don't really think I did)
user.wolfram.com/portal --- the wri user portal is down for maintenance. v9 release imminent maybe? (I wanted to download the installer, and I wonder if I can still get v8 after v9 is released)
@All and @Szabolcs I have tweaked the image uploader a bit. It does now show a small icon of our site next to the main-buttons. Additionally, you can now differentiate whether you like to - mark an uploaded image for QA on the site. Then the url gets the style `` which can be directly used in a post - mark and image for usage in chat, where the pure url only is copied to clipboard The default button is always "Site" which means you can just hit enter and everything works as expected. Furthermore, I "Queued" the upload buttons so tha…
If anyone wants to try it, then he can download it (for now) from here:
@BrettChampion Heh, this is what happens when they try to heavily optimize for the average transit time... somebody always ends up getting a delayed package (or someone goofed up bad). I had a package ship from a place 50 miles away from me, fly to Kentucky (2000 miles away) and then back to me the next day :)
@rcollyer Could you please ask your dad what he uses to seal/polish/varnish the bowls he turns? (assuming they're meant to be used as food bowls). I'm fairly certain poly/tung oil is not a good idea...
What you need is BandpassFilter, which is new in version 9. Assuming your audio is sampled at 22400 Hz, you can do:
BandpassFilter[data, {60 π, 180 π}, SampleRate -> 22400]
to filter it to between 60-180 Hz.
@whuber Thanks for the flag; the post has been reopened (it was locked automatically as part of the migration rejection system). If possible, please edit the question to clarify it, since you know the subject (you're probably on it already) :)
how about we set ourselves a to produce a reader's digest version of the new features. Obviously some are more important than others. I would be good to get some guidance to what is really new and powerful and what is anecdotical?
So, to use units, I need to always use the Quantity[something] in the input form? That’s great, my outputs will become easier, but my inputs will become a mess to the eyes of the non mathematica user...
I have to confess that I'm disappointed with the frontend (truly). For the non expert user the front end is what makes Mathematica different from the others. Most novice users don't care about sophisticated stuff. They care about their spreadsheet environment... Does it at least have an undo that doesn’t force me to explain the unexplainable? (not kidding).
And also no “whole new level of interface” - internation Mathematica user converence - November 2009: “"Drag and drop" is a metaphor that’s pretty nice for some applications, and some interfaces. Flow charts or block diagrams are nice for some other interfaces. We’ve now for the first time in the Mathematica front end got the raw material to generalize all of this, and make a whole new level of interface. Well, we’re making very interesting progress on that. I think it’s going to be demonstrated elsewhere.”
@chris true, true, but currently we use it for random walk simulations, spectral imaging and for molecular dynamics post processing. Come to think of it Stephen always wanted to have a "universal" computing environment...
How am I going to explain what I’m trying to do with Quantity[9,"Meters"*”Meters”]/ Quantity[3,"Meters"]. It is unreadable! So I code in Mathematica, and I re-write it to make sense... Couldn’t we just do (9 m^2)/(3 m), with some sort of escape when placing m (like |unitMeters|)?!?
@P.Fonseca Yeah. I agree about Quantity being unreadable, but, there are usually ways around it. You could create some input alias and a nice tag box for the viewers if you need it
As to the front end, I'd wait until we can test it. It doesn't seem to say much in the docs
@Rojo I don't... But, if it is so simple to add a strategy to make it readable, why... And if there are no functions for spreadsheet or drag and drop, on an everything is an expression system documentation, then there's no such functionality.
@Ajasja Yes I was waiting for an answer to my question mathematica.stackexchange.com/questions/6169/… I am utterly disappointed that WRI doesn't feature my requests in their updates ;-)
@Rojo do you have any smokers around you? Could suggest it to them ;-)
@Ajasja to rephrase, I am not that impressed that mathematica can do now things other softwares have been doing for a while. Its useful, but not impressive.
may be finally plot legend will work in Mathematica? see reference.wolfram.com/mathematica/ref/PlotLegends.html this always been a problem in Mathematica to get nice plot legend. I hope finally it is solved. This is important.
@chris yes, but you'd need to buy extra toolboxes for all these new features with Matlab. They are all integrated in Mathematica. You can't believe how much pain it is in Matlab to deal with toolboxes
when I write something in Matlab and use toolbox A, I have to make sure the other side has the same toolbox A for code to work. Having Matlab only is not enough. It is like using 10 different products instead of one.
@NasserM.Abbasi yes, though this is a "business model" issue. Mathematica went through this phase of trying and sell add-ons... A real nuisance I concur.
@Rojo you should ask S. Wolfram, since he said “"Drag and drop" is a metaphor that’s pretty nice for some applications, and some interfaces". But I think that it all sums up to the point that we cannot do a system modeler kind of interface on Mathematica front end. So, they have been selling the argument that you can do all sort of softwares with Mathematica, but at the end, not even their version of system modeler interface is doable.
I have lots of applications that need links, snaps, etc, but that have nothing to do with system modeler. I know I can mimic it with locators, but I think we all know that it is time consuming, slow, and renders a thing can be called, at maximum, a toy...
It is sometimes the case that questions on Mathematica SE lead to requests for enhancement. So, I was wondering how best to get these enhancement requests into the hands of Wolfram developers?
I have a site license support contract with Wolfram and so I could, in theory, email them to say 'See ...
@Rojo also, spreadsheet interface thing is of extremely importance. Everybody knows Excel. Very few people know programming. So, when they look at Mathematica they see: too complex, not what they are looking for. If they saw an interface where they could mix text with computations, program their functions in between, use them in a spreadsheet type of super cell placed after the text, and plot something just after, they would see it as a much productive tool than what they can find in Excel...
@Rojo I have been waiting/talking about this for a long time, and Wolfram has been point to "a whole new level of interface" for 3 years now. So... disappointment. I'll have to wait 2 years more... since honestly, the constructability of a serious user interfaces was not pending on ListPicker.
I never understand why CDF's have not got more popular than they are. Looking around, there is really no other technology like CDF's (what else would one use to make computation run inside a browser as easily and with such advanced features?) other than Java applets, which no one wants these days due to all the security problems, there is really nothing like CDF's out there in the merket now. Yet, I do not see them used much out in the world. WRI needs to market them better I think
@NasserM.Abbasi they work fine as demonstration toys. As soon as your interface becomes a little complex, it starts to become slow. If you want to import export, it starts to become expensive. So, not easy to implement on a production environment.
Concerning the front end, my last judgment will go to the 3D graphics mouse rotating functionality. Is it now possible to dynamically change the pivot point (as some kind of snappy draggable point that I can mouse displace it to another image referenced location)? Or is it still the completely outdated center of graphic static kind of thing that renders rotation completely useless after doing the slightest zoom on the graphic?
@P.Fonseca Everybody knows Excel, yes. It's easy to learn because it hardly does anything, and what it does do it does mostly incorrectly. :) I never understood why people still use Excel, to be honest. I haven't touched it since I first got Mathematica and I never liked it even before that.
@rcollyer so now we know who was in the beta program, eh?
@P.Fonseca well, that I agree with. But if people want to make things difficult for themselves by using Excel, that's their stupid decision. :)
Wrt. M9, the new general linear algebra and the NDSolve enhancements look very nice. It seems though that a lot of what we thought would be new in 9 is actually not going to be ready until 10... looking forward to the Premier Service download because then I'll be able to see how the undocumented functionality has moved on...
@P.Fonseca a lot of people use Origin as well. I suppose they must just be masochists. For crying out loud, people... just download R or Python, would you!
@OleksandrR. but since I generaly work with people, their stupid decision is my problem. A common interface means more users. More users means more budget for new functionalities. Will have to wait until 2014.
@P.Fonseca ah, that's too bad. I once knew someone who decided to write an SVM classifier for fluorescence LIDAR data in Excel. I was stunned. The result was quite good but the process must have been incredibly painful for him.
So many examples of people using Excel in completly non profitable ways, and still they use it. What's wrong in this picture? I believe people are not wrong, they are just a fact. It's the Mathematica strategy to bring this people on the wagon that is non existing. And these are millions of people...
@P.Fonseca the SVM guy didn't use Excel because of the spreadsheet interface. In fact he used that part only as a sort of GUI and for charting the results. I think people just tend to stick with what they have, even if it's crap, rather than learn a new tool.
In the beginning people don't have those prejudices ingrained already. But what they get exposed to first is usually Excel, so it sticks...
@Ajasja sure, right tool for the right job is a good argument. If you're an accountant who doesn't care that Excel's numerics is a complete botch job, it's perfect. ;)
@P.Fonseca if all you cared about is the interface then you could buy Excel Link; it doesn't cost that much. I think people are just naturally resistant to new approaches.
If they like Excel then I think that's fine (as long as I don't have to be involved). But I dread to think of the hours of productivity lost through using such tools.
@OleksandrR. if I tell excel users to try Mathematica because they can still use excel with the link plugin, what's the point? If I tell them that there's is a thing called Mathematica, where they can work as they did before, but add a huge amount of other possibilities to it...
@P.Fonseca well, the fact is Mathematica doesn't work like Excel now and hopefully never will; a tabular interface just won't go far enough in reconciling the differences to make Excel users comfortable. But if you give people two dissimilar tools and let them choose between them (after informing them of the relative benefits), they're much more likely to choose the right tool for a job than if they just stick with the one thing they know. If they can use them both together, that's even better.
People are just used to GUI-driven programs these days. This never used to be the case: people used to get their start with a BASIC interpreter (or FORTH if they were lucky). It's easy to graduate from BASIC to Pascal because they're fundamentally similar in concept. Mathematica and Excel aren't.
So, how can we captivate all the Excel users that would profit a lot more with Mathematica? (I state Excel instead of any other tool because if has a huge user group). Keep in mind that those users are no longer in an academic environment, and any hour of their work needs to be justified as production.
(and I mean every hour, discretized on an hour basis)
How many Excel users do you think actually need Mathematica? How many people who use tools like Excel actually know or care what a CAS is? I guess it's less than 1%. Most people are scared of math, anyway. Other tools, like R, or Python, seem a better choice for mainstream users, especially considering their cost in relation to Mathematica's.
If they need something better than those, then they might be interested in Mathematica. But Excel to Mathematica is a big leap.
I think that a lot more than one can imagine. I used excel for 10 years, and droped it after learning Mathematica (much more productive). Since in my company there are hundreds of people doing more or less what I do... But my transition was forced in someway, which I cannot do with others...
What about the Visio interface? Graphs are completly integrated within Mathematica, but still there's no way of easily tracing them on a GUI (with drag and drop, snaps and everything)
@OleksandrR. You'd be surprised by the things that people use Excel for... some of my friends who design complex exotic commodities (the kind of crap that gets us in mess) in investment banking firms almost exclusively use Excel. They have huge and complex models and simulations all coded up in it and these firms even have intensive Excel training for new hires
Lets have the best tool, but still miss the main users. Who cares if they don't use it! Well, everybody that needs to work with them should care, and every user of Mathematica that thinks that more budget would mean more functionalities should also care!
I think one of the main attractions (for them) is the dynamic updating of data and charts and the complex ways in which you can set up dependencies. It can be done with Mathematica, but it's a pain to figure out the dynamic stuff once you get into large and complex uses
@rm-rf that's the kind of thing that requires a dataflow language to do well. If that's really the main attraction for them, I'd consider it one of the only good reasons to use Excel.
@asim good question. Do we know how many users of each there actually are? WRI claims millions of users but it seems doubtful to me (millions of copies sold is a different matter). For MATLAB I can imagine perhaps millions of users.
I think that Mathematica has a positioning (symbolic stuff) that does not attract a large user base. Even though it is very good in numerics, that is not sufficiently emphasized in Mathrmatica's marketing message.
Well, the bottom line is that MATLAB makes linear algebra extremely simple and most engineering problems can be boiled down to manipulating a bunch of matrices
For someone who has had some exposure to MATLAB or python, switching to Mathematica can be a pain because simple matrix operations can sometimes require 30 characters or more to type out
@asim although I also consider the MATLAB question important, Excel question shouldn't be forgotten, since it probably represents more than 100 time the MATLAB + Mathematica + Maple + etc, user base :-)
Yes, that is true, but Mathematica could have emphasized its compilation to c capabilities somewhat more. This is a definite advantage, if widely known
2
@fonesca yes, Excel is wildly popular. I guess people like menu driven stuff and an ability to see data that they are manipulating.
@asim Yeah, and perhaps more documentation on how to compile effectively would be useful too... I see no reason to hide the list of compilable functions or not have examples on using the compiler report to see if there are calls to the main evaluator and work around them, etc.
I think WRI's focus is on broadening the base and making mma more of a black box than it should be. Folks like us will mostly stick with it despite it not having our pet feature.
Mathematica's popularity could possibly increase if there is an easy kill switch to switch off the symbolic stuff. One need not then worry about packed arrays, symbolic evaluation etc..
I mean, if people can still stick with it after 9 versions without a sensible undo, they surely must be either infatuated with the product or extremely masochistic.
@asim we can compile to C but at the end we need a full Mathematica license to access it as a runtime, while others give the runtime for free :-( . So if people like to see the data they are manipulating, should we change people or add that functionality?
@asim That's actually a good point... that's how it is in MATLAB. You need to explicitly declare certain variables as symbolic (although I find it painful). I like how mma is setup, but a kill switch would certainly be useful. Perhaps maybe designate a particular kernel as a numerics only kernel...
@fonseca changing people is difficult. changing products is easier. Simplicity is highly valued too, hence the popularity of Excel
I think the Rlink feature in version 9 might be useful, if performance is not a major issue. But again, R users will have no reason to suddenly use Mathematica because of this. This just helps the existing mathematica user base
@asim just write 1.0 instead of 1 and then Mathematica computation will be faster, this is the switch you want :)
@rm-rf I have a web page on using Matlab and Mathematica for many common engineering problems. I find that actually Mathematica is simpler to solve many problems, even for matrix operations. It is just that Mathematica requires more skills. But once one learns it, it is simpler and easier to do many things compared to Matlab. Please see 12000.org/my_notes/mma_matlab_control/index.htm
@nasser yes, that is partially true, but would not work in optimization etc.. sometimes symbolic differentiation is much slower than algorithmic differentiation..
for example, filling up a sparse matrix, say 3D, is very confusing in Matlab and many students in my class struggled with it. I found in Mathematica it is much simpler to do. i.e. the API was more intutive.
Could someone who has V9 allready please try this integral on it: Integrate[ x/(1 + x^6*Sin[x]^2), {x, 0, Infinity}] , the claim in the stackexchange math group is that it converges, yet Mathematica 8 says it does not converge.
@belisarius No I haven't. But the problem is not the need for a link (although it could help). If people don't know what finding a root is all about (they use “goal seek” in excel), it is not because they have a link that they learn new things in Mathematica. We need to get people out of the excel interface, but presenting Mathematica to an Excel user is like presenting hell.
If they could see in it something similar, it would be easier. And then, using the tool on the everyday basis, would mean learning new things (and there’s no software better to learn new things than Mathematica).
@P.Fonseca I mean, I agree that it has limitations for really serioius appls, can get slow, and the front end may crash, but people seem to get stuck before that point. At the point of not knowing how to use it properly, how to figure out where to put dynamic when the app isn't too small. How to make code modular and not end up with a kilometric dynamic when the app isn't so simple, etc
@Rojo without wanting to brag myself, I think I'm not so far away to the limited of its capacities, and they are not huge. Again, how can I make a simple drag and drop with snappy corners, without it becoming a horrendous thing... How can I create a matrix of input “cells” next to each others, with multiple selection, copy, etc... Most technical software needs for some sort of tabular input...
Again, I ask WR to check how much of the system modeler interface cannot be recreated in a professional way with Mathematica.
@P.Fonseca The lack of more boxes, or events of event handler, or keyboard key checks of CurrentValue, etc, will be improved I guess, but I don't think that's a problem with the whole idea of dynamic
I am not saying that it can ever be really professional, I was giving a general impression I get, clearly not from you, that people get stuck way earlier than necessary with dynamic
I more or less like that at least that is mathematicaish, with so many new objectish stuff introduced that I don't like much
@rm-rf If it is misunderstood, perhaps for being misdocumented, it won't be found easy to use
@Rojo when Mathematica introduced the dynamic stuff, it was pretty unique compared to other high level tools. Since then, Sage and now more recently, R has introduced similar functionality (See Shiny).
I try to do numerical/statistical stuff with Mathematica, but it can be sometimes frustrating. For example, the statistical machinery introduced in version 8 is great, but, then not so great for ordinary numerical stuff. Most of the distributions cannot be compiled, and are not listable/vectorized.
It could be.. though, performance may be an issue, with data being exchanged from one side to another. Also, from what I looked, R models (formulas) probably cannot be exchanged easily, so time will tell whether this is useful.
It has gotten a lot better though... about 2/3 of his first 15 questions were heavily downvoted, closed and deleted. He was suspended a few times and he has learned (hopefully) from it... no deletions since
@rm-rf we still have arbitrary code blocks, arbitrary tags, barely comprehensible wording... IMO, nobody should be vomiting out questions like this onto the site.
@OleksandrR. Yes, but the community hasn't closed or commented (negatively) on any of his questions post suspension until this one. From what I can see, he has certainly improved (a string of about 15 OK questions) but if this becomes a pattern again, I'll bin him — possibly for a longer sentence.
@rm-rf okay. I just wanted to make the point that improvement from unbelievably dreadful to merely awful doesn't really change the fact that these are still far below what anyone would consider an acceptable contribution. I guess we'll see how it goes and what the rest of the community thinks about it.
I agree, but... not cringeworthy enough to bin him. Bad tags can be fixed easily by retagging. I should note that a LOT of folks use terrible tags in their post... often it goes unnoticed by most (as it should be) because JM is very quick at retagging.
@halirutan The newest user that registered a few minutes ago has an ID of 4803. So somewhere around that number (not exact, because of account deletions, merges, etc.)
Of course, registered users will be smaller than that
@SjoerdC.deVries Well, 30 mins can make a hell lot of difference... The guy asked this question once and I sent it over to Signal Processing, because mma doesn't have decent DSP support and he didn't know the algo part.
He deleted that and reasked and by that time v9 docs were out and I saw that support for DSP had been added and they had a function that he could use =)
What you need is BandpassFilter, which is new in version 9. Assuming your audio is sampled at 22400 Hz, you can do:
BandpassFilter[data, {60 π, 180 π}, SampleRate -> 22400]
to filter it to between 60-180 Hz.
It is sometimes the case that questions on Mathematica SE lead to requests for enhancement. So, I was wondering how best to get these enhancement requests into the hands of Wolfram developers?
I have a site license support contract with Wolfram and so I could, in theory, email them to say 'See ...
