« first day (78 days earlier)      last day (3123 days later) » 
00:00 - 19:0019:00 - 00:00

12:00 AM
phdcomics is very true in that sense
@yoda ah. well, of course if there's a conference and the girlfriend isn't around, I eat with the participants. so does everybody else
12:41 AM
@yoda would you add ?
@acl Have you tried increasing the number of MKLThreads via SetSystemOptions["MKLThreads" -> ...]? It may help if you have the cores available. Default on my machine is 2.
12:57 AM
@rcollyer no, I haven't. I doubt the bottleneck is that, but it's worth trying tomorrow (I have lots and lots of cores available... it's brain cells that are the problem)
yes it's two on my laptop too; so that's why it uses 2 cores. great!
@acl I just ran across it myself, today, and I figured any bit of help would be useful.
@rcollyer aye, that it is, also for other projects. halving the time something needs to finish is important even if it's from 1 minute to 30 secs, because 30 secs is semi-interactive--1min and I'm reading SE...
1min is not long enough to do something else, but too long to stare at a screen doing nothing
@acl 30 secs and I'm reading SE. My attention span has dropped.
30s of doing nothing I can (barely) tolerate
ah, I'm ahead of you there :)
If I wasn't behind the 8-ball here, I'd offer to see if I could port your code to c++ using VSIPL++. In most cases, I could do it in a week.
Possibly, 2.
It uses an LAPACK/BLAS implementation as a back-end, alongside an fft implementation, with MPI for parallel.
1:10 AM
@rcollyer not worth it, I could probably kill it with more computing power faster than that. and it needs manual intervention (I don't know the interesting parameter ranges yet). but thanks
Ah, still in the exploration phase.
Let me know if the thread count helps.
I was reading some papers from the 80s today during a talk; they displayed some plots. to check if I understood, I wrote the code and produced the talks, inside a manipulate, during the talk! all this on a macbook air
technology progresses
@acl That's awesome.
@rcollyer yes, we are adding the results of a new approach to a paper, and have to work out if it really does work or not
1:13 AM
@rcollyer you mean Compile to C, or linking to C code?
right, I see. everything that can be compiled is compiled (to C). I do that automatically
Compile to C parts of the code, per Leonid's answer I linked to, or even CompileEvaluate?
or, everything that I think can be compiled anyway
Never mind then.
I think I've compiled everything I could
Sounds like fun.
1:16 AM
well, it's OK. running out of resources is better than realizing the method you're using simply doesn't work.
Sometimes it takes more resources than you have to make that realization. The problem some of my code faces: 50+ GB RAM required if I don't rewrite to a more memory friendly/slower method.
@rcollyer well, I have 512GB per machine so that's not a problem...
or actually it is, but 512GB isn't that much more than 50GB for this sort of thing
@acl I wish. Although my dream machine consists of 4-16 core processors with 256GB, and I want 3 of them. :)
Maxed out, with a head node for storage, I figure approx. $25k.
@rcollyer my dream machine is pencil and paper and mathematica :)
1:22 AM
@rcollyer well, am I mean if I tell you we have a number of these easily accessible, in addition to the real heavy metal? :)
Well, I'm thinking how I would not have to sit in a queue waiting because the groups cluster is slowly dropping out one machine at a time.
and I have no use for them... but the others do
@acl nope.
Well, the design on my part is deliberate. The particular motherboard has 3 (possibly, 4) gigabit ports, so I could directly connect them to the other machines without having to run them through a router/switch.
Although, apparently the newer version is faulty ... back to the drawing board.
well I try to avoid relying on computers more than mma. They are good for a specific kind of problems, and I am interested mainly in a different kind of problem
yeah, well, my calculations require me to throw as much power at them as possible to get everything done. Of course, I will have performed about 1000 of them for the thesis.
Lots and lots of computer time, and I didn't write the code.
1:33 AM
@rcollyer this works if you have a well-defined but hard problem. but at that point I tend to lose interest, I'm more interested in the defining-the-problem phase, unfortunately
@rcollyer oh that sounds fun
@acl it is some of the worst code I've read, and it has put me off Fortran as a viable language completely. There are at least 5 versions of Simpson's rule integration spread throughout the code!
@rcollyer that can't be fortran's fault...
It isn't. I've seen some very nice Fortran code, but I hate debugging Fortran77 and earlier.
@rcollyer I hate debugging anything. it would be nice if it all worked like my pencil and paper calculations, with factors of 2 and pi reinstating themselves as necessary
and don't forget $i$ and $\hbar$.
1:39 AM
and if f[x got compiled to the right thing, like it does in my head (but it would probably not work in a computer as it lacks the big picture)
Things go seriously wrong if you forget $\hbar$.
@acl true
hbar ==1
oh no, $i$ is always included explicitly while $\hbar=1$ so no problem :)
@belisarius there you go, a fellow physicist! (hi!)
1:40 AM
@acl Hi!
This looks promising for a motherboard. Only dual core, though. :(
Every constant should be one
Like in "Imagine"
@belisarius ha yes
@acl it's always setting $i == 1$ that causes people so many headaches, and I can't understand why.
however, as a mathematician friend once put it when he saw my pages of calculations with nary a $\pi$ in sight, "let's hope you never try to put a man on the moon" :)
1:41 AM
The only useful application of dimensional analysis is to determine which constants you did forget in your calculation
@rcollyer but why would anyone do that? it breaks the structure completely
$\hbar$ just rescales
@acl sarcasm?
@belisarius "forget" is too strong a term... I prefer "omitted for brevity" :)
@acl Ahhh yes, "brevity" >)
@belisarius yeah, but it is that pesky conversion between the different cgs units that gets me every time.
1:44 AM
@rcollyer Thanks god I forgot all tah
cgs, what cgs. it's all =1
@belisarius I've never actually done it. But, there's this experimental paper I wish to compare my calculations to and they're in cgs. Basically, I'm screwed.
or $\ell_b$ if you have magnetic fields
@rcollyer ah yes, I tried this once. had to email them in the end
they thought I was an idiot, apparently (I later befriended the postdoc who did the experiments and he more or less told me this)
@rcollyer Just divide their result by yours and see if there is an integer multiple of hbar there
I get my stuff in $\mu_\text{B}$, you know reasonable units, and then I'm presented with this cgs crap!
(I hate backspace sometimes.)
1:48 AM
@belisarius always a good strategy
What smile and pretend I know what I'm doing?
@belisarius ah, the "go insane" option. :)
I ate it as an undergrad :)
@belisarius ha, is this what you used to do?
1:49 AM
@belisarius That explains more than a few things. :)
@rcollyer Thnxs!
@acl Yep. More or less. Solid state too.
I took GR, and it was interesting. But, it lost me somewhere, and every class I just got tensor.
@rcollyer The learning curve is steep. But it is VERY interesting
@belisarius GR and solid state? that sounds like a strange mixture
must have been fun
1:53 AM
@acl Landau is a hero. I guess nothing better than his first book (mechanics) has been written
@acl Both have non-orthogonal geometries. GR is curved, also, but besides that, I can see the overlap.
@rcollyer Those was two stages. Not simultaneous :)
@belisarius absolutely nothing. the first paragraph, introducing variational calculus, changed my life
sad but true
@belisarius One of my favorite HW assignments came from that class: determine the energy lost due to gravity of a shaken fist. All sorts of approximations came into play there because they didn't matter. It was fun.
but the style blew my mind
1:54 AM
@belisarius I understand. Just there is some non-trivial overlap.
@acl Yep. That explanation about why the Lagrangian could not be another thins is very difficult to forget
@rcollyer Well, in fact the renormalization technique used are pretty the same
@belisarius that's the second or third paragraph/section, I think :)
@acl Well it has been quite a few years ...
@belisarius yep.
@acl actually have not read it, and poorer for it.
@rcollyer What? You did not read Landau?
1:57 AM
@rcollyer I'm not sure it's useful for practical purposes
@belisarius "you're fired!" :)
@belisarius only the classical mechanics text
@rcollyer That the best
and even then, I was beyond it at that point.
so it didn't help.
problably the QM and statistics follows
1:58 AM
but the style is unbeatable
QM (vol 3) is messier, less elegant, but it's extremely useful
More elegant than Armani
I use it all the time
@acl Was it written by LD?
I can't reember
What do you guys use as a math methods text?
@belisarius yes
2:00 AM
What is "math methods"?
@rcollyer nowadays, mostly nothing, but usually either arfken & weber for simple things or morse & feshbach for heavier stuff (but it is heavy)
@belisarius sorry, that's probably an American colloquialism. mathematical methods of physics, or mathematical physics
and I learnt most of what (little) I know from those and whittaker & watson
@acl didn't use arfken, but what I have looked at wasn't to bad. more reference than anything else.
I'm currently enamored with Hassani.
2:02 AM
@rcollyer well, less than M&F or W&W!
@acl never even run across those two.
@rcollyer I have never used it
@rcollyer well W&W are probably not so much mathematical methods
@acl I got it while I was working in industry, and it changed how I thought.
And, there are parts I still don't understand. several chapters of them.
sure that is true for any book
W&W are more analysis than methods
M&F are M&F, no other way to put it. they casually do what we'd nowadays either do in mma or not do at all
@acl not all of them. But, some force you to change your approach completely. This one ties together so many disparate topics into a coherent whole.
@acl I hate those people. Although, at times I'm one of them.
2:06 AM
I wish I had time to read books instead of writing papers...
I don't envy you that.
@acl Referees often think the contrary :)
@rcollyer but those 2 volumes (of M&F) were written in a different era. there was no mma then. you either did it that way or gave up. nowadays, knowing this stuff is useless, if intimidating to your colleagues
@acl true.
@belisarius ah yes, referees.
the only good referee I ever met was myself :)
2:08 AM
yes, those little grey people
@belisarius a couple of days ago I got two referee reports on the same day
There is no good referee. Only one that does not cite himself. And he died young
as my collaborators put it, "arghhhhhhhhhhhh"
@acl There was this assignment in E&M that was just a brutal mishmash of trig identities. I learned I can hold about 15 sub-expressions in my head at once, while simplifying them, that day. Scared the hell out of a fellow student.
and that's my only claim to fame with regards to that.
2:11 AM
@rcollyer Elephants can do that and don't brag
@rcollyer yes this kind of skill becomes less useful with mma, maple and the like
@belisarius I'm not an elephant.
not sure
show me
I remember all sorts of integral identities involving bessel functions, for some reason. totally useless, but I do
@acl about 10 lines of mma code interspersed with text to do it. :(
2:12 AM
never used a single one of them, I think.
@belisarius how would he type if he was one?
@acl plenty of dexterity with the trunk.
That isn't a demonstration.
reading the screen, though, I think it would have to be a 50 inch plasma.
@rcollyer you're not helping your case here
Do you eat vegetables?
2:14 AM
@acl it's fun to play both sides. :)
@belisarius why yes I do.
is this a kind of Turing test?
Do you take long showers?
@belisarius and what of it?
you are an elephant
@rcollyer how do you feel about mud?
2:15 AM
@acl I'm kind of curious as to where he's taking this.
elephants eat vegetables and take showars
@acl not that great, actually.
@belisarius ah, but I like steak.
@rcollyer I am not quite sure if that disqualifies you
@belisarius see, he's changing tack now
@rcollyer what colour is your skin?
@acl He is afraid. Elephants usually are
2:16 AM
@acl geek white, yours?
Are you furry? Perhaps a mammouth ...
@belisarius sorry I don't discuss my body hair with any one but my wife.
Hi @Verbeia
You aren't going to believe what you walked into.
@belisarius oh, and Feynman could do that too, and he wasn't an elephant!
@rcollyer But he played bongos. That showed clearly he wasnt an elephant
@rcollyer wikipedia says that:
Since the early 1990s, young male elephants in Pilanesberg National Park and the Hluhluwe-Umfolozi Game Reserve in South Africa have been raping and killing rhinoceroses
2:20 AM
@rcollyer what do you have to say for yourself?
@acl only really awful puns are coming to mind, so I'll refrain.
raping poor rhinoceroces. You should be ashame, rcollyer!
@rcollyer indeed. I came in because I was wondering if anyone else has had problems printing graphics exported from Mathematica to PDF.
@Verbeia I think there was a Q on SO ...
@Verbeia fairly sure he's right.
2:23 AM
@rcollyer You still have to answer about poor rhinoceroses
@belisarius you still haven't proven I'm an elephant.
We, the Greenpeace task force in MMa SE are waiting
@rcollyer It is a fact. Facts don-t need profs
@belisarius I feel like there was too, but not finding it yet. They are refusing to print unless I "Print as Image". This would cause far too many glitches for our production process.
2:25 AM
@belisarius true, facts don't need professors, but verification helps.
@rcollyer OK. Let's discuss that at the zoo
@Verbeia Let me see
@belisarius if I'm an elephant transportation to the zoo will require a lot of coordination and money. Are you footing the bill? Or, are you just going to duck any responsibility?
@Verbeia does printing not work correctly if you rasterize? in my case I get glitches if I print without rasterizing, not if I do rasterize (same for PDFs)
@rcollyer Pehaps your decorated cage disguise its real nature
@belisarius hmm, it would be tough using the toilet in the house if that we're true.
2:28 AM
@rcollyer I'll no discuse that in public
@belisarius then we are at impasse.
@rcollyer OK. We will decide about the elephant or mammoth issue later
@belisarius neither. hairless ape.
@rcollyer Ok. But move your trunk slowly
2:32 AM
@belisarius and the ears
@belisarius That's part of the issue, but I am finding that they are just refusing to print at all. I haven't posed it as a question on the site because (a) I'm using a custom package that changes almost every default in the graphic and (b) it could be the printer driver, and therefore "Too Localised".
@acl Sorry my ears and nose don't extend that far.
@Verbeia post it. If it's the driver, we might close it. But, you may find what you need.
@Verbeia What if you try to export without the custom package? Does it work?
I mean, a standard plot
@rcollyer Do they have webcams at the zoo facility?
just to check ...
@belisarius you're failing as a scientist, that is a complex question as it assumes that I live at a zoo, and would be on exhibit there.
@rcollyer I am playing the Konrad Lorenz part here
2:41 AM
@belisarius No, and the default graph doesn't work on a different printer either. Nothing to do with Arial as the font, either. A bog-standard DateListPlot fails.
@Verbeia So I guess it is a platform problem
@belisarius ah. still don't know him, though.
@rcollyer He is (or was) brilliant
Cool. Still not an elephant. :P
Anyway. Good night.
2:48 AM
@belisarius I give up. You are the master!
@acl Poor rhinos!
3:07 AM
man.. leave 3 physicists in a room and they start talking about gravitational pull of elephants
and raping rhinos
@rcollyer ElephantQ[rcollyer]
@yoda If you're buying the beer, you can call me whatever you want.
mma doesn't lie.
Sure it does.
lol, if I'm getting drunk, I'll sure as hell call you whatever I want ;)
My dad's bird says: "get me a beer."
3:13 AM
Is it a norwegian blue?
For @CHM 's, nascent interest in bio:
@rcollyer nooo, don't kindle that fire! :P
@yoda It's already burning, and we didn't start it.
That article is actually interesting.
well I understand @CHM to some extend; some people are thick as a brick
@yoda Of course, if we're speaking about drinking there's this one about red wine and superconductivity. Yes, you read that right.
3:20 AM
(ok that was sort of mean to some questioners on the biology site, but I couldn't resist)
right, I'm off
night all
@rcollyer I drank mostly gin while studying the BCS
@acl night
@belisarius I opted for whiskey
now it's very clear and simple
anyway, night
@acl Yeah electrons go straight away from the spinning phonons.. I can see them
3:22 AM
@belisarius I understand. apparently, though, someone was drinking while experimenting and saw a jump in superconducting temp when they spilled the wine on their sample.
Anyway, thought I'd share.
Night all.
night rc
3 hours later…
6:42 AM
@@F'x I can't upload to anywhere from work, but I will try later when I get home. I've got the files. So does @MikeHoneychurch, since I have his email address.
2 hours later…
8:17 AM
@Verbeia okay, thanks; I'm not so good at Mathematica, but I've become kind of a PDF nerd over time :)
@rcollyer duh, Wannier code; we're working in the same field, I suppose
@rcollyer "computational materials physics" — OK, my field is nanoporous materials, close enough :)
@rcollyer and regarding Fortran, yeah, some old forms can be a pain to debug
but modern Fortran can also be a nightmare; for example, I don't know if you've ever looked at the source code of cp2k.org
9:21 AM
OK, maybe reading the backlog wasn't such a good idea.
3 hours later…
12:14 PM
@Fx no, I haven't looked at cp2k; I haven't had the need to use it. I think any language can be written poorly, and made more difficult to debug. For instance, have you used any of the template functionality in c++? If you haven't the error messages can literally be pages long per message.
@rcollyer oh yes, I unfortunately have
@Fx there's a saying: with c, it is easy to shoot yourself in the foot, but while in c++ it is harder to do that, when you do, you blow your leg off.
12:58 PM
in this analogy, I suppose Mathematica would be like a rocket dragster: 99% of the time you amaze friends with your masterpieces, but the 1% remaining involves you spectacularly exploding
1:52 PM
@Fx And the goal is to either whittle away at that 1% so it explodes less, or wrap your self in enough protective gear, that when it does explode (and it will) your well padded.
@rcollyer But the more protection you use, the more risk you're willing to take because you feel safer.
Are we still talking about MMA?
@Heike yesterday I crashed mma by attempting to have it display something a little to big for it to handle.
@EliLansey probably. but, the innuendo is there if you want to grab it.
2:17 PM
@rcollyer when innuendo is involved, just don't grab anything!
@rcollyer Mathematica tends to do that. When mma crashes on me it's usually when I'm doing something with Manipulate.
2:36 PM
@Fx :)
2:52 PM
does anyone know of a command line or visual diff file viewer (for '.nb' files) for MMA? From some google searches, it appears that the '.m' files are amenable (I guess they're plain text?) while '.nb' files would probably need MMA itself.
@tkott Notebooks are ultimately just text files. However, they do store meta-data in a file outline cache which would not yield anything good with a diff file viewer. That said, if you turned it off, it should work. As an alternative, there's workbench which may include some sort of diff mechanism as it is built on top of Eclipse.
posted on April 05, 2012 by Paul-Jean Letourneau

In Stephen Wolfram’s recent blog post about personal analytics, he showed a number of plots generated by analyzing his archive of personal data. One of the most common pieces of feedback we received was that people wanted to know how they could perform the same kind of analysis on their own data. So in this [...]

@rcollyer Hmm, i forgot about workbench. I think I have access to that too. I'll check it out. I tried doing a diff using a text parser, but with figures and such, there was a LOT of noise. That's why I was thinking MMA itself might be better...
3:08 PM
@tkott probably. There's also this question which may help.
@rcollyer thanks for the link! I guess my google skills need work :)
@tkott this wasn't google, per se, but remembering that the question was asked on SO, and then looking for it.
@rcollyer but, but, google knows all!
They try to, at least. :)
4:06 PM
This discussion is starting to annoy me: mathematica.stackexchange.com/a/3907/46
4:24 PM
@Heike I sympathize.
@rcollyer thanks for the link.
@CHM you're welcome. it's an interesting article.
@acl I've expressed my thoughts pretty clearly in some comments, and was apparently not alone to think as such. There's no way their site can get better before there is an active mod team, which seems dormant, to say the least.
I'll want to read his original paper someday, when I finally decide to do it.
Me too. Need to brush up on information theory ...
I think he's pretty known, actually. But that's probably univeristy bias.
4:32 PM
@rcollyer His paper
Ooh. Thanks!
Just reading your post on bio.meta now.
I feel a bit bad for being rude.
But it is what it is.
oh well. they'll either get over it, or not. but the criticism was needed.
2 hours later…
6:39 PM
@CHM shannon?
he's pretty known yes :)
@Heike just ignore it if it annoys you (I was told in a different discussion under that question that it is not true that Re[f[z]] is not differentiable; discuss)
6:55 PM
@RM Hi
@SjoerdCdeVries While LeafCount is a good measure of complexity in most cases, don't you think it can be misleading if used blindly?
The OP wants to deal with one part of the piecewise only
My take on the question was that the OP wanted to just extract whatever equation is there that involves his variable x so that he can, say, plot it or something
You need a criterion
Usually all parts may involve x
as my first example shows
I'd say the question is a bit underspecified
6:59 PM
It's the "business part' that has meaning to him
00:00 - 19:0019:00 - 00:00

« first day (78 days earlier)      last day (3123 days later) »