Even better, it hasn't updated on the site page itself yet, but the Area 51 page visits/day numbers just clicked up again, to 1048

Traffic graph now looks like this:

hi @rcollyer

Hola, @acl.

Trying to figure out what answer @MrWizard and @Rojo were talking about above.

this one:

Just found it, myself. Thanks.

Damn, Andy Ross has over 200 points more than me, again. It makes it very difficult to catch him if he doesn't stand still.

Ah, I hadn't checked the user list for some time

I notice that I have crushed you both :)

(for now)

It's not like I'm working that hard at it.

Not like Mr.Wizard or Szabolcs.

@rcollyer yes, they're zooming away

Mr.W has bypassed Szabolcs. Maybe that whole collective thing is working.

I'm not answering many questions these days as I'm pushing to finish a number of things.

I understand.

But, even if I did, I doubt I could ever catch up with them, or give answers as elegant as most of Heike's

She does give beautiful answers, doesn't she.

and then I look at my actual code, the kind of code I write for my work. it's atrocious

@acl Having 3 non-mods above 10k is very good for us in the long run, though. More people with access to the moderator tools when we graduate.

@acl the just-get-the-damn-thing-running code is always atrocious

@rcollyer sure, and others will probably pass 10K before then

@yoda it's not until you have to go back and generalize it that it might be beaten into better shape.

@acl likely.

@yoda yes, and in my case it never gets fixed

How much testing (automated/manual) do you do?

@rcollyer My generalization attempts get grander and grander in the design and planning phase that I eventually give up on generalizing it :)

@rcollyer if you're asking me, very little

@yoda know that feeling.

@acl actually, yes I was.

(but keep in mind that I don't do large-scale stuff like you do)

@acl For one project, I've got acres of tests because i don't trust myself.

@acl Well, my biggest project isn't complete (at 4k lines of code), but I may work on it some tonight. However, most of my projects are no where near that large.

@rcollyer my stuff is much smaller usually. also, I don't usually do calculations where the method is well-known and I apply it on a large scale. so most of the complexity is in working out what to do, then doing it is not that hard (relatively)

@rcollyer I don't think I've written 4k lines ever... in total!

so I usually get lulled into not testing very quickly, with predictable consequences

If you're curious, here's some of it. I need to post the rest to actually make it work.

@yoda This was actually like remodelling a house. A lot of work.

@acl most of my projects are on the small side too, and 9 times out of ten, ad hoc.

Hmm.. hg huh. That's pretty much similar to git, right?

very.

well, I vaguely recognize STL stuff from some years ago but that's it... way over my head

That's neat that they allow private repos for free

Some of the generic programming is overblown, and needs to be scaled back.

@yoda I've marked that one as public and linked it to my careers profile.

I figured it may get me a job, or scare off the future employers ... either way.

@rcollyer that's quite well-structured code. if I showed you my C code you'd probably run away screaming.

I'm an instant no-hire for most programming jobs — illiterate in every other language :)

There are three pieces of code in there I'm sort of proud of: load_matrix (the whole thing), impurity_solver (solver_wo_analytic function), and chempot (chemical_potential function).

by the way, I played a bit more with Julia (the language :) ) today.

@acl Since it was going to be large, I knew I had to do it "right."

@acl how is it?

@rcollyer makes sense. wish I had some!

Some?

Not following.

(sense!)

sensei?

@rcollyer I reimplemented some quick and dirty stuff I wrote the last few days in mma (gutzwiller ansatz calculations, which amounts to solving a large number of coupled ODEs using eg RK or something as simple).

I just typed it out (with help from the docs), and it worked almost straight away. faster than mma, too (but, I was implementing RK by hand in mma, because I kept arguing with NDSolve and never won)

@acl something along these lines?

Julia is really friendly :)

And you didn't even have to get her liquored up.

@rcollyer a modification of that for bosons and with spatial and temporal dependence

@acl okay. nothing I've touched before, but have run across.

works exactly for hard-core bosons and for weakly-interacting bosons, seems OK between

Interesting.

Further down that page, there is a section about the band-structure of Nickel, and the difficulties DFT has with it. The code I linked to in principle should be able to deal resolve the issues, like Gutzwiller.

well as I said I use it for systems which are laughably simple compared to yours, so it's not hard to push the thing through

but: if your code can in principle solve this, why not shift gear and do it?

(well, after the thesis of course)

@acl it was supposed to be part of the thesis, but time ran out. I'll get back to it soon, though. But, I need a job soon.

@rcollyer right. any progress on the job front?

@acl Honestly, i've been focused on getting things done, so the job front has been more like window shopping. Also, I don't know who would hire me as a post-doc, I've only got a conference proceedings and an arxiv paper.

so, someone who knows you

I mean, you have specific expertise which is hard to acquire. surely there are people who need it?

@acl At the moment, that's limited. So, I'll have to start working the network and see if that can lead me to anything. Although, I have met a few individuals I wouldn't mind working with, but only the once.

The one guy I was speaking of would have use for what I've done. So, that's a possibility.

of course there's the question of whether you really want to do a postdoc in the current climate.

True, very true. I'd like to stay in academia as I'd like to teach. However, I'm not a fan of publish or perish.

Although, I am making progress on adapting ErrorListPlot to more general tuples, and I may get to use it to analyze some data in the next couple of days.

So, WRI is always an option ... Just have to finish the damned code samples.

or a programming job? more money, for sure. and one is probably less dependent on contacts (but maybe not, I don't know)

Eli has passed my resume along to his future employers, so we'll see. It's programming, but physics programming. So, definitely nice.

It's in NY city, so not so nice.

@rcollyer I'll admit here that I've never been to NY

which apparently makes a minority of one...

@acl Most of the state, is semi-rural with pockets of moderately sized urban areas (250k, at the most), and then there is the city. The citizens of NY state divide themselves into downstaters and upstaters, and they don't like each other.

where I am from, there is a village that is split into "upper" and "lower". the two have separated elected councils. they have been in a legal dispute over land (actually some churchyard, I think) for 35 years now. they refuse to speak to each other and a couple of years ago there were fistfights in the street over it

total pop: 3000

so...

human nature it seems

@rcollyer what specifically do you mean by code samples? Samples of your code that you wrote for other stuff or samples for specific problems that they asked you to solve?

@yoda on their website they ask for code samples if your applying for a programming job. I've put together a notebook that incorporates some snippets from SE, stuff I wrote special for the notebook, and some of my other work.

@acl yep, you're right there. Here, it's over who supports who. The perception in up-state NY, is that all of our taxes go to support NYC. True, or not, that is the perception of things.

regarding the job hunt. you could always go for this:

guardian.co.uk/world/2012/mar/31/…

why not?

they posted an advert.

@acl That's hysterical. I don't think they'd take me: I'm an unrepentant heretic.

DataListPlot[
  With[{mn = Min@#[[All, 2]], dat = #},
   {#1/4 // N, (#2 - mn), #3, #4} & @@@ dat
   ],
  PlotRange -> All, Frame -> True, Axes -> False, PlotRange -> All,
  Joined -> True,
  PlotMarkerFunction -> {(
     With[{x = #1, y = #2, err = #3, conv = #4},
       {If[conv,
         ## &[Darker@Blue,
          Disk[{x, y}, Offset[{$PointSize, $PointSize}] ]], ## &[
          Darker@Red,
          Rectangle[Offset[{-$PointSize, -$PointSize}/2, {x, y}] ,
           Offset[{$PointSize, $PointSize}/2, {x, y}] ]]],

ah yes, one needs to be a communicant Anglican and a Commonwealth citizen, apparently. you're neither, apparently.

tough luck

nope, the citizenry part was settled long before I was born.

@rcollyer well, all your taxes come from NYC :P

@yoda most likely. But, it is ultimately the rural vs. urban tension that continues to divide this country.

of course... and there are plenty of politicians and media folks who keep kindling the us-vs-them mentality

With the above code, if that's all it did, it wouldn't be something to write home about. The key feature is I've built a mechanism to allow comparisons across all data sets in determining how to display a point, e.g. comparing energies to correctly label which one is the ground state.

@yoda you've noticed that, haven't you.

you get sick after a point. Can't get away from it either

any way, big mac overhaul planned — full reinstall

Nope. And this election is particularly horrifying. Some of the candidates are just plain bat-shit crazy.

@yoda best of luck.

@rcollyer that's putting it mildly

@yoda King of understatement here.

@yoda thoughts on the code?

@rcollyer yeah, gonna try and get debian/fedora on it too. The problem is that if I mess up and screw anything up, I don't have a working cd drive to do any repairs --- yikes!

@yoda that is a little horrifying.

I've done upgrades on the cluster, remotely. But, the most significant ones, I was only a floor away. So, when something went wrong, and it usually did, I could go upstairs and fix it. Now, I'm a little farther away.

@rcollyer Do you have the rest of the definitions? That does look pretty convenient (from what you've described), but don't have a feel for its use at the moment

Wasn't this or something like this used in your paper? I mean, choosing the label based on the state

Actually, I did the stuff in the paper by hand, and that's why I started to put this together because it is damned annoying. I'll email you the notebook, and let you break it.

heh, thanks :)

You still have some fiddling with ArrayPlot in LS to do, but I've been holding back on sending you a minimal example because 1) it isn't important and 2) you need to write your thesis :)

I'll stop, lest I turn into @acl :)

@yoda the more, the merrier :)

Well, I'd hate to be at the receiving end in about a year from now... :P

Since you both (and a few others) use macs, here's a list of useful settings

github.com/mathiasbynens/dotfiles/blob/master/.osx

not all, but a lot of goodies in there

I should send it to Leonid, too, he always manages to find a way to shred my code in a new and interesting fashion.

Hi @MarkMcClure

Didn't notice you pop in.

blinks... wipes glasses... yup, it's @MarkMcClure. Hello there!

Later, all.

gotta go too

@rcollyer, the bounty question on splitting the word in chemical elements. His answer, and my suggestions on the comment, specifically the suggestion on wrapping Block

I'm not sure everyone had seen this, so here it goes, the Clippycorn…

@Fx Yeah, I got one of those earlier too. It did actually prompt me to accept an answer that I should have gotten around to accepting a while ago. But really, what the?

This is what you get looking at this question on Meta.SO

We're at 1048 visitors/day at the moment. I wonder what made it jump from around 950 to 1048 in two days time.

@Fx I haven't seen it in the wild yet.

@Heike Well, it must be that two low-traffic days dropped out of the 14-day window, and the two most recent days have traffic above 950 visits. But that's the proximate cause; it's hard to say exactly what's attracting more visitors. We aren't actually getting new users any faster.

Interestingly, the only other beta sites that have more visits/day than users are Fitness and Nutrition, Motor Vehicle Maintenance and Repair, and Gardening (just).

Surely there is a more functional approach than the code (in comments) here?

@Verbeia One could do something like Flatten[Reap[Sow[#, Sign[# - p]] & /@ lst, {-1, 0, 1}][[2]]]

I'm currently posting it as a question. Feel free to answer it. I'm off to bed.

@rcollyer did you see my comment?

@SjoerdCdeVries I also just quickly checked now and the limit is what the question says it is, not 2; 2 is the $\alpha\rightarrow 0$ limit

@SjoerdCdeVries I did, and I fixed it up.

@acl I disagree. There are two branches, as $\sqrt{\alpha^4}=\pm \alpha^2$, and you have to be careful you don't jump from one to the other. Assuming that alpha \[Element] Reals chooses the positive root. But, the limit of 2 requires the negative root. If you distribute the $k + 2$ term in the numerator, a limit of 2 is straightforward to achieve without making any assumptions about the branch we're on. That said, using Apart forces you to make a choice.

@rcollyer disagree with what?

@acl I think the limit is 2, and care must be taken due to the branch cut around the square root.

oh i see. let me have a look again

mma assumes that $\sqrt{a^2} = |a|$ not $\sqrt{a^2} = \pm |a|$ as it should.

I wouldn't have thought that this question would generate so many comments.

I was as surprised as you.

@rcollyer Why should it be \pm |a|? You're free to choose the sign of \sqrt{a} as you like.

@Heike actually, you're not. Choosing one or the other changes which branch you're on, so care must be taken.

@rcollyer It depends how you define \sqrt{z^2} I guess. If you define it as $\zeta =(z^2)^1/2$ then the complex plane minus the imaginary axis will be mapped to the half plane $\re(\zeta)>0$ is you take the principle branch. But you could also define $\zeta = (z^1/2)^2$ which for the principle branch will map the complex plane minus a branch cut along the negative real axis onto itself.

@Heike Very true. I had a friend who spent several weeks moving the standard branch cut which is along the negative real axis to the lower half plane only intersecting the real axis at $\pm 1$. In the original form, his calculation would not converge, but with the odd changes, it would.

@rcollyer I had a problem where I had a branch cut along a circular arc going through $\pm i$, which was a pain to implement.

We crossed 1000 questions!

@nixeagle you're late at the party.

The 1000th was an off-topic question that was deleted later on. Bummer.

@SjoerdCdeVries brilliant! which question was it?

@acl I don't have a link anymore and since it has been deleted I wouldn't know how to get at it. It was a math problem about a poissondistribution. I solved it easily using MMA but the questioner admitted he didn't know this site was about the Mathematica language

@SjoerdCdeVries ok, the usual maths question

@acl this one: mathematica.stackexchange.com/questions/3772/…

(I have no idea if that works)

@heike how did you find it?

@SjoerdCdeVries By clicking on tools in the banner at the top of the page

@Heike thanks

tools... I can't find them!

@acl I think you need 4000+ rep to get access to those.

@Heike I think I did have access to them at some point...

@acl It's a switch. It maybe called Review on your bar

If you click on Review you get a window with both tools and review and then you can click on tools

@SjoerdCdeVries that's it, it switches between review and tools

thanks

welcome. Signing off for today. Bye!

bye @SjoerdCdeVries

bye

We're on the homepage of stackexchange again.

does NMinimize not work over the complexes?

@acl I guess you could always write z=x+I y and minimize over x and y.

@Heike yes :)

I am looking for shortcuts

guess I'll have to do that. well, tomorrow, rewriting things after midnight is not a good idea

Generally not or you'll end up with a mess, like in Gremlins.

@acl I'm glad I'm not the top user any more actually :-) And elegant answers are worth a lot more actually

never watched it...

@Szabolcs "I, Don"? as in I, Robot?

:)

I'm being dragged awat, η αγάπη μου :-)

καληνύχτα!

will be back after Easter!

We passed 1000 visitors per day, did you notice?

Have a nice break.

I wonder if we'll graduate after 90 days

(though I'd prefer to have a bit more time)

Bye!

@Szabolcs I think it has been mentioned before, today.

@Szabolcs night!

@Szabolcs Yes, me too. It would be nice to have a few more people with >10k points by the time we graduate.

since some of you might use macs, I'd appreciate if you could do me a favor and review some of my answers and up vote ones that are useful, especially the lower voted ones.

I'm inching closer to winning an iPod Touch =)

o.O

I thought that contest was over.

:(

Tomorrow is the last day

I see.

It's a good contest — kind of sad in some sense because voting is being done only in cliques as everyone is conserving their votes to edge the other one out — but for the site, it's been great in reducing unanswered questions and publicizing it

Over the last 2-3 days, I got a few revival badges for answering old questions and a bunch of announcer and booster badges. We should definitely do something like this when v9 comes out

What do you propose we offer as prizes?

Home edition of mma ~ $250 and Workbench $145. So about 20 of each should be short change for SE

just home edition would be nice, but then again, if someone answers questions they have mma...

amazon gift cards or whatever they're called?

but not v9 and most probably not something they can keep between jobs/univs, etc.

true. but credit at amazon is more flexible

Some sites (tex.se, programmers) etc., get funded for a conference, so a few get to go - all expenses paid. Mods could perhaps bring this up with SE late summer in time for WRI conference in Oct

Alright, I'm out. Bye all.

Hmm.

Bye.

@CHM btw, I think we all came down quite hard on you the other day on the blog issue. I would still encourage you to keep posting to that blog as if it were your own and maybe share your interesting finds in chat

Sjoerd runs a twitter bot - twitter.com/#!/mathematicaatse

Basically, he just gets a feed of new questions and gives it to twitter, which does the rest. That has gotten 11 followers already without any human interaction.

next step, we write mma clients to subscribe to it. then we're all set

@RM Maybe the personal blog is the right choice. But I don't have too much time for blog writing, that's why I envisioned a many-contributor approach.

Ah nice, twitter bot. Good idea.

@CHM y'know, I've never seen that go well without an editorial team. I've seen friends group together to write about their undergraduate program experiences — originally with the vision of sharing unique experiences. They all end up in Prof. X was a jackass, which is hardly unique. Then they go to grad school and start a different blog group. Rinse and repeat.

what would a blog be about?

I understand. What I'm thinking about is not a traditional blog where people share their experiences, but where people just share code. Much like a source repository, but à la blog: "I wanted to do that, here's how I did it with MMA".

The idea's not fully mature.

@acl Gee... I don't know. That itself is worthy of a meta post when we're more mature -- the scope of our blog.

@RM what do you use MMA primarily for?

@CHM to answer questions here =)

Don't use it very much outside of here, although I'd like to change that.

Really? Are you a programmer then?