The h Bar - 2015-01-27

« first day (1545 days earlier) ← previous day next day → last day (3388 days later) »

1:09 AM

@dmckee Don't worry, I'm not offended! If it stops ridiculous nonsense, feel free to do what you need to do

5 hours later…

5:45 AM

I think we need to make it a rule that downvoting questions (or answers) by a new user (who has been around for only a week or so) must be disallowed. A user who finds his way here through google and tries experimenting with the site will not be very happy if he recieves a load of downvotes on his/her first question (or answer). You may imagine his/her first impressions of the site. This turns away a lot of new members.

I usually refrain from downvoting such questions, but there are people out there who do.

Just wanted to drop by to give this suggestion, I gotta attend class. Bye.

2 hours later…

7:41 AM

@Gaurav if a user posts a crappy question that accumulates a lot of downvotes, then they'd better not be happy! If they decide, based on that, never to come back, that's on them. I can't imagine we lose many valuable contributors that way.

So, let me put it this way: identifying the reason downvoting exists and using it as an argument for why we shouldn't be downvoting (certain posts) isn't going to get much traction. ;-)

8:14 AM

This is why I really dislike Griffiths's quantum book

1 hour later…

9:37 AM

Everyone teaches bra-ket notation like that though

its that stupid transition from integrals to vectors

for quite a respectable amount of time you compare everything new you learn in bra-ket notation with everything you learned before using funtions

its innevitable

2 hours later…

11:22 AM

@DanielSank Would you rather they start with a rigorous treatment of Hilbert spaces? I don't think that's feasible: Many people struggle with the 'trivial' mathematics of physicist-oriented QM 101 already.

Oops... misclicked on this review: physics.stackexchange.com/review/close/71360

please vote to close if you see this :P

11:39 AM

@Danu I think the struggle with the rather trivial linear algebra is precisely because it is not made evident that it is just linear algebra

11:54 AM

Hey, since the Fourier transform mixes up multiplication and convolution

is there some kind of nice probabilistic interpretation?

12:34 PM

@Danu People struggle with QM 101?

@JamalS That sounds pretty arrogant.

Well, you said it was 'trivial.'

(At least, the mathematics, that is.)

Conceptually, I would expect it to be initially quite confounding.

@Danu By the way, I found a great Chrome extension for academia.

It's called sidekick.

It integrates directly into GMail and allows you to send tracked e-mails which inform you when they have actually been opened by the recipient.

That way you know if a professor is ignoring you, or they're just busy and haven't had a chance to take a look :)

1:21 PM

@Danu Characteristic function?

1:36 PM

@Danu You can always go back to the page (not the review page) and VTC there, you know.

@KyleKanos I really havent been on here enough lately :P

@alarge ?

Meh. There's always backspace > click link > VTC

2:26 PM

Does this guy just keep adding questions to his question? oO

Hello, everyone.

@ACuriousMind Yes

More questions = better, right?

On a related note, it is strange how some people think we can read minds.

2

3:09 PM

@KyleKanos : There is a problem and I need your intervention. The question with the {integral}(physics.stackexchange.com/questions/161592/…) that you put on hold has to be transferred to the mathematicians (mathematica Stack Exchange). Although I solved it, and checked the proof in different ways, there is something pathological there. I would be glad that they also see my comments, and get their explanations on what happens.

@Sofia That integral is the propagator of a scalar field, without the proper treatments to make it one of the well-defined advanced/retarded/Feynman/reverse-Feynman propagators. I believe it has no proper solution as it is written there. Also, it is commonly taken over more than one-dimensional k, complicating the issue a bit further.

(Why has it no solution? Because the integral has poles on the region of integration, i.e at $k^2 = \mu^2$)

3:30 PM

@DavidZ Don't mistake me over here. I'm not implying we need to degrade the quality of posts. I never said we needn't close or put questions on hold, only that the downvotes are discouraging to new users.

@Gaurav That's, uh, kinda what they're meant to be. If I downvote a question, it means I want to discourage the poster from posting stuff like that again.

@ACuriousMind Did you read my earlier comment ?

The alt-text above the downvote icon says, "This answer is not useful"

So downvotes are meant to say that the answer isn't useful. Doesn't really matter if the user is new or not

@Gaurav Yes, I did. It's just that the effect you are complaining about is, well, one of the intended effects of downvotes.

@KyleKanos Do you think you can generalize that to the whole lot of users here ? People just downvote on questions by new users because they think it is not useful. Fine, that's going by the rules; but imagine the emotional impact on that new user.

3:35 PM

Also, the one unassailable freedom on SE sites is that of voting - it is completely at the discretion of the users to vote up and down how- and whatever they like.

Why do people feel it necessary to bring up the "emotional impact" of downvotes

So even if I agreed that new users should not be downvoted, it would never even stand the slightest chance of becoming a policy

It's not just you saying that Gaurav. Bobie & PeterHorvath have been lamenting about that for some time as well

It's dumb

There is zero evidence of an emotional impact on a downvote

I've been the recipient of many downvotes

All that's made me want to do is improve my post

@KyleKanos Well, that's because the people negatively impacted supposedly leave, and hence we don't hear from them.

Convenient, isn't it?

Too convenient

3:38 PM

@KyleKanos To be fair, PH and bobie have more of an issue with the close than with the down votes, so it's a bit different.

I kinda agree with Kyle, My first post in maths was closed because of"no prior effort", but it only made me explore the site than leave the site !
I think maybe a compulsary tour of the site for a new user can be more beneficial than all that has been done

Peter doesn't like downvotes either. WolframJohny is also big on anti-downvotes

He's even got the statement in his profile that he likes to give upvotes to unexplained downvotes

well i have never seen sensible questions with some or any effort closed , in my short time here..

@Gowtham The tour and help buttons are not hard to find. Maybe it should be more obvious that the help includes also an instruction manual how to use the site, and is not just there in case you have problems?

@ACuriousMind I understand that , but a majority of new users wont possibly use the help , before a moderator actually points it out to them .. i just thought maybe a tour that is non-skippable will make the new user ready when he posts his question or answer, thats all

3:45 PM

@Gowtham Arguable. Most of the questions here that get closed or put on hold are those by high school students or amateurs of physics. My argument here is that the emotional maturity of many of those students are low, meaning that they are easily put off by the downvotes. Leave the high school students, anyone for that matter who is emotionally sensitive may find it aggressive that their first question or answer got -4 or maybe -10 downvotes.

@Gowtham This has been brought up on the mother meta many times. The sentiment there is that we don't want to put too many barriers in front of people asking their first question, so there is a minimal click-through suggesting that you read those things.

@Gaurav That's highly subjective with absolutely zero evidence showing this

Now, that makes sense if you assume that most incoming users understand how to ask good questions.

May seem stupid, but just putting my view across, using nothing but my experience.

That might be slightly more true for Stack Overflow than for Physics SE, but it is not going to be much more true.

And Stack Overflow is now the biggest help site on the internet. Bar none.

Physics hasn't been that successful, but I think we're doing pretty well.

Because of, not despite, the strict rules we operate under.

3:48 PM

So what I say is that why take a chance ? Even if such cases might be a minority, if that user really might be intelligent and a good contributor, we're really losing something.

>42k members, >47k questions, >49k visits per day, >70 questions per day seems to be pretty damn good to me for a Physics site

We're the #2 science site network wide in most categories (only math is better than us):stackexchange.com/sites#questionsperday

@dmckee dint know it was discussed already, just my idea that any new user(not all) would be looking to post the question than going through help/tour.

Would it be a really big job to add such a feature barring downvotes on questions by new users ? If it is, my suggestion doesn't hold.

@Gaurav I imagine it is non-trivial. But beyond that, it goes against the intent of the SE network

We want to get rid of bad questions and answers

@dmckee I thought the moderators here are supposedly "kitten" :P

3:56 PM

@KyleKanos How do you get data on the frequency of new users registering here ?

@Gaurav I'm not sure the details on SQL, but you can try on the Data Explorer

@KyleKanos Will try.

I've found one that separates by week and forked it for use on this site: data.stackexchange.com/physics/query/268232/new-users-by-week

@KyleKanos Seems you just now composed a new query

Oh, let me see

Yeah, I forked it and made some minor changes (e.g., defaults to Physics instead of SO)

4:01 PM

530 this week ? seems a lot

Okay, i gotta go guys. I'll post a comment on that data later, figured i don't have time now.

What do all these people do here? One could almost think it was chat time or something...

3

@ACuriousMind : but all the calculus I did is correct step by step. I see no mistake. The two integrals into which I split the initial integral overcome the problem of their own poles. Though, as you say, each one is conditioned by avoiding the pole of the other. So, do you imply that the splitting into two parts, as I did, is incorrect?

Is it chat time?

Actually, we're late on that, aren't we

@KyleKanos Yes, chat session started ~7 ago :P

The unit of 7 is left as an excercise for the reader

I'm used to them starting at noon

Stupid daylight savings

4:09 PM

@Sofia Your argument for the integral over $\frac{\cos(q)}{q}$ vanishing because of its oddity fails, since it has a pole at zero.

@ACuriousMind : If my calculus is incorrect I would gladly remove it. But how?

@ACuriousMind : simply erase it? I can't delete.

user image

2

@Sofia If you want to remove it, you should see a "delete" button next to the "edit" button on the bottom of your answer

^ new users registered on Physics.SE per week for the entirety of Physics.SE

You can always delete anything you have written here (unless it is a question with upvoted answer, or an accepted answer, I think)

4:12 PM

@ACuriousMind : no, it doesn't work.

@KyleKanos Unnecessary last data point is unnecessary

50

Q: Best of MathOverflow

This is a place to collect MathOverflow success stories! Was some of your research inspired by something on MathOverflow? Do you know questions & answers that led to interesting research? MathOverflow citations? Open problems solved on MathOverflow? Then add your story in an answer! (One story ...

discussion site-promotion

@ACuriousMind But I like cliffs

Does physics.se has a similar post on meta?

Otherwise, nice graph

4:13 PM

@ACuriousMind : you have the privilege to delete whatever you consider appropriate. Can you do it?

I would like to browse through the linked posts, just out of curiosity...

@Sofia Aside from the answer being wrong not being sufficient reason to delete it, I do not have the priviledge to cast deletion votes on your answer. (That comes with 20k reputation, I think)

@ThomasKlimpel I don't think we have something like that.

@ThomasKlimpel: A related Meta.Physics post:

12

Q: Does physics.SE help research?

Does some concrete evidence exist that the physics.SE site offers any help with research? (Note that research in physics may not be confused with theoretical physics, that is a specific thing.) Were new works published there that later obtained recognition from the professional scientific communi...

discussion content-level

With zero answers :(

(and it references that "Best of MO" post too)

Oh, best of need not necessarily be research in the form of published papers.

$Mathematics$ In praise of Math.SE site and its users

achievements, milestones, interesting statistics

I just would like to browse interesting questions and great answers

So do we need to discuss anything about "Serial Voting"?

2

Q: Careless (serial) voting

If you see a few comments engaged in a pointless discussion and try to flag them in a quick succession, you'll notice that the system will slow you down: you are compelled to wait 5 seconds before you can flag another comment. I imagine that it is not casual and suppose you agree that it is a wis...

(sorry, wrong link before)

No?

4:21 PM

That particular question is going to keep popping itself up for a while.

But I don't think that anything needs to be fixed.

Well, this has been a productive chat session.

They've been rather "productive" lately

deadest of dead chat

Yeah, seems I done killed it

Better than Astronomy, I'll admit.

Astronomy's chat is often dead?

4:25 PM

It's been frozen a couple times. I've tried to sort of wake it up.

@ACuriousMind : a wrong answer is misleading. I don't want to mislead. I will just erase my answer and explain in brief why I do it. But I want to get rid of the upvote that I don't deserve. Can you remove it? Also, can you post YOUR explanation?

@HDE226868 o.O wow

The last time it was up and roaring was for about +/- 5 minutes of New Years UTC.

I said we should resolve to be more active in chat.

Engineering has been productive, though.

@HDE226868 That's good.

Meh. It's not as many people as we'd like.

4:27 PM

user image

@Sofia: Are you sure you don'T see a delete button there?

I've posted a screenshot of the bottom of one of my answers - there's a "delete" right next to "edit", and it should be there for you, too

@HDE226868 260 is still a good start for a Beta site. We ought to promote it too

I hope we can make a better promotion ad than we have on HSM, although that's still a work in progress.

It should be easier on Engineering, though.

@HDE226868 Uh, why?

@ACuriousMind : haaa ! It worked. But, please, it is not enough. Would you care to post YOUR explanation? Why my answer is wrong and your one correct, that should better be explained. Do you want?

@ACuriousMind Because history of science and math isn't exactly a well-known discipline. It's hard to draw on past figures or ideas.

4:31 PM

@Sofia The question is (rightfully) closed, so I cannot post an answer. I'll leave a comment, though

You've seen Academia's ad, for example.

@HDE226868 Pictures of famous scientists?!

@ACuriousMind Plato?

I know, it's not really what the site is about

@ACuriousMind Maybe.

@KyleKanos I think Academia already has him.

4:31 PM

@KyleKanos Yeah, one or two from each epoch or something

Oh. I see now

I was reading the intent wrong

Those seem like some interesting choices.

@ACuriousMind : I got a FAULSE upvote. As I am upset when I am deprived of the upvotes that I deserved, I don't want to stay with non-deserved upvotes. Help me, can you?

@ACuriousMind : what to do, I like justice. Can you help? Do you want to?

@Sofia If you delete the answer, you will lose the reputation from the votes on it a few hours after that, no need for me to do anything about it

There's an automatic routine that checks once in a while if you really deserve all the reputation you got, and if the reputation came from posts that are deleted now, it will remove it

@Gowtham No big deal. A whole lot of things have been discussed on the mother meta and there are few people (even among those who have been around since Stack Overflow's beta) who know more than a little piece of it all.

4:37 PM

@Danu Forget rigorous Hilbert spaces. Griffiths starts on page #1 with the Schrodinger wave equation. Think about what students have learned up to that point: in classical mechanics do we have fixed $x$ and $p$ and a time varying "other thing" which contains the information of the system? No! We have time varying $x$ and $p$. Why not start quantum mechanics with the Heisenberg picture instead of the Schrodinger picture? It's much closer to what students have learned prior and more useful.

Alas, if you are serious about trying to get a major change rolling you should probably read the relevant posts on meta.stackexchange.com and on the blog.

@DanielSank If only more people thought like you :)

Furthermore, Griffiths spends the first few chapter focusing on wave function crank-grinding instead of learning physics. I think one of the first things you should learn in quantum mechanics is black body radiation or something like that. No wave functions, easy math, and completely blows classical physics out of the water.

@ACuriousMind: I'm working on that. I evangelize in my research group frequently.

@dmckee sure !

I figure if even one of my colleagues becomes a teacher and leans in my direction that is good.

4:40 PM

@DanielSank We do that in "modern physics" which they take before QM. But our student don't always have the preparation to be confident about the classical result. ::sigh::

Terry Bollinger

Hmm, looks like a physics discussion going on...

@TerryBollinger The last thing you'd expect in the h bar, right? ;)

@dmckee: This is always an issue, but I find that the real value of school is learning what ideas are there and where to find out more. It's unreasonable to expect students to download everything the first time (although that's no excuse for bad teaching!).

Terry Bollinger

:)

@DanielSank Yeah. When they're stuck I back and fill. But it cuts into the fun topics at the end of that semester.

4:42 PM

@Danu: Then Griffiths does the hydrogen atom. This is a huge mistake. It's a very hard problem requiring advanced mathematical techniques. I think that instead one should focus on interesting phenomena which don't require so much math.

@DanielSank: I've always thought that it would be best to teach as much of QM in finite-dimensional spaces as possible before finally moving to wavefunctions and these "weird bra-ket integrals" and such.

@dmckee: I personally don't think "fun topics" are all that important in undergrad. I found that I was usually more too being annoyed at not understanding the basics to wish I were learning something "neat".

There's a reason we call Heisenberg's picture matrix mechanics, but it is almost never made explicit.

@ACuriousMind : A (hopefully) last question. I still see that answer of mine. That's bad. But, could it be that I am the only one that sees? In this case it is O.K.

@ACuriousMind: I think infinite dimensions is fine but that Schrodinger picture is bad.

@Sofia: Link to answer?

4:44 PM

@Sofia Only you and other users with the priviledge to see deleted stuff can see it now. It's fine :)

Terry Bollinger

Came in late: Has Feynman's "QED is actually easier" idea been discussed?

@TerryBollinger: No, but that's nonsense. :D

Pardon my frankness.

Terry Bollinger

'Tis not! :)

Read QED, seriously.

@TerryBollinger: It's "easier" until you want to solve areal problem.

I've read it.

I don't think confronting QM newcomers with renormalization and such is a good idea :P

As Daniel says, it's only easy as long as you don't try to actually do anything

Terry Bollinger

4:48 PM

Interesting! Simplicity of concept doesn't matter? (I agree fully on the computation part.)

@TerryBollinger: It matters, but I hesitate to arm my students with an easy to understand shovel that can't dig straight holes.

Terry Bollinger

Heh! Nicely put.

I don't think it was until I was in grad school that I wanted to use a path integral to calculate something.

@TerryBollinger "Simplicity of concept"? The concepts involved in QFT are not really simple - they are only when explained in natural language by Feynman, which is not quite an accurate portrayal of what you actually do in the theory

The vague idea is simple and elegant. Doing actual physics is messy and difficult

^ that

Terry Bollinger

4:52 PM

@ACuriousMind, by that I mostly mean that Feynman pares back the concepts required as a foundation for the math. That aspect if anything justifies the subsequent formalization, which of course was mostly by Dyson, not Feynman.

It saddens me that simple and elegant has been lost as drivers for physics. Seems to me that a lot of the early 1900s work emerged out of just such silly beliefs.

@TerryBollinger Which is wonderful for giving short introductions or talks about it, but if you want to train physicists in a lecture, what would you do? Start with the concepts of QFT, then backpedal to teach actual QM, then teach actual QFT?

@ACuriousMind : just mathematics now. Where did my treatment become no more correct? I believe that up to the point where I assumed that the odd integrand brought no contribution, everything was fine. Am I right? Was before that, some other step incorrect? Can you tell me, or is it a too much pathologic case for deciding?

@TerryBollinger: Indeed. Unfortunately, despite all of that, if I ask you what is the state of a calcium atom after subjected to so-and-so laser pulse you are sure as the sun rises not going to do it by "adding up the little arrows".

Why do you think simplicity and elegance is lost in physics?

@TerryBollinger Oh, "simple and elegant" is still a driving force in theoretical physics. It's just that simplicity and elegance as understood by the practitioners of a theory cannot be assessed by laymen anymore

(...and nothing Feynman did could be understood by a layman anyway)

He was not known for being a good teacher.

Terry Bollinger

4:56 PM

@DanielSank, wow, no objections at all! I was about to ask the goal of the particular type of physics being taught, and you just gave an excellent example. Feynman's QED would just confuse for that goal, wouldn't it?

@TerryBollinger: We were talking about undergrad level physics in the abstract. No particular example.

@Sofia Everything in your answer is correct except for assuming that the integral over the cosine vanishes, which it doesn't - the integral is simply ill-defined because of the pole at the origin. To "fix" that, one would do precisely the same complex-analytic trick that people to for the original integral.

Terry Bollinger

@DanielSank: Why lost? String theory. I am a computer scientist, not a physicist, and I know horrible programming with a complete absence of verification when I see it.

@TerryBollinger What would QFT before Feynman have looked like to you? ;)

@ACuriousMind : a lot of thanks for your kind help. Now it's O.K.

4:59 PM

@ACuriousMind: I have a primal need to make a comment here:
What many folks call a "complex analytic trick" is really just adding some dissipation to the system and then taking it to zero at the end.

Terry Bollinger

Mired in the present instant of time.

Feynman exploited the functional perspective nicely.

@TerryBollinger That's like me finding the worst C++ code and pointing at it as evidence that computer science isn't elegant.

Physics is not a proper subset of string theory ;)

Terry Bollinger

It isn't, ohmygosh, it's a bit of a travesty!

@DanielSank Yes, that is a good physical interpretation of it.

@ACuriousMind Here is a reasonable reference to that idea.

Terry Bollinger

5:02 PM

What I like about QED is not its teachability -- yes, you are all right, and it's not even a complete work in my mind -- but its focus on extreme simplicity generating fascinating large-world physics.

@ACuriousMind This is another thing I evangelize about.

@DanielSank We have even talked in the comment thread there already ;) (And you already have my upvote)

@TerryBollinger For whatever it's worth, statistical mechanics offers similarly large things-you-can-understand/assumptions-of-the-theory ratio.

Terry Bollinger

It occurs to me that someone reading back over this conversation might have trouble connecting the various threads... :)

@TerryBollinger: That's why we use the reply feature. If you mouse over the right side of a message you can click an arrow. This will link your next message to that one.

Try mousing over one of my messages. You'll see that it highlights the message to which I was responding.

Super useful.

You can also respond to a user's last message by using the @ notation.

Terry Bollinger

5:05 PM

@DanielSank I am such a newb!!

@DanielSank And isn't it absolutely fascinating that statistical mechanics emerges from all them little arrows and dials and such at the path integral level?

I feel guilty on eating time -- is @Sofia getting her math questions answered?

@TerryBollinger I think her concerns have been adressed :) (see her last message)

@TerryBollinger Not sure I understand this.

Terry Bollinger

@DanielSank Understand statistical mechanics emerging from QED?

@TerryBollinger: Yes. Isn't it actually kind of difficult to recover statistical mechanics from first principles equations of motion? I thought this was an active area of research.

Terry Bollinger

@DanielSank First principle equations of motion... OK, since in the path integral approach even the concept of a particle moving in a straight line for any non-trivial distance emerges only after you do a lot of summing of a lot of non-straight options, which comes first?

5:21 PM

@TerryBollinger Your "emerging" is not the Wick rotation, is it?

Terry Bollinger

@ACuriousMind Heh! I'm a computer scientist, I don't even recall what a Wick rotation is, though I seem to recall seeing the term.

"Emergence" is a huge, scary, and important topic in networks, AI, and biology these days.

'kay then, carry on ;)

Terry Bollinger

@ACuriousMind Looked it up, that's intriguing and I'll have to read more, but no: I was only referring to the very standard integral-of-all-possible-histories and how even the concept of "straight" requires summing enough probabilities to make it likely. Again, my perspective is odd: That to me just looks like a form of network emergence, one with a lot of nicely precise rules that make it even out at large scales.

BTW, a lot of QED can be remarkably easily justified if you simply assume a finite memory that begins to lose detail (on average) at the quantum scale. I'm sure someone somewhere has done papers on the idea, it's just too obvious. You can instantiate more detail, but only by applying more memory (mass/energy) to the issue.

The QFT framework in general assumes infinite spatial and temporal precision, which of course cannot happen in real physics. I always found that dissonance fascinating.

@TerryBollinger yes, thank you for your concern, I got answers and they are clear. What I cannot understand is how is it possible that such a homework be given to students. It seems evil.

Terry Bollinger

The best time to teach a new concept is right after you personally went though the pain of learning it. After that, everything starts to merge together in terms of perceived levels of difficulty. It makes it hard to determine the right point at which to connect.

5:34 PM

@Sofia I think the asker probably misunderstood what "calculating" an amplitude means, and thought they were supposed to solve this integral when they actually weren't (or have done it in class, or whatever)

@ACuriousMind : I was just about asking you of your last comment. What you mean by The probability density doesn't always have that form for ψ ?

@Sofia Ah, I see how that is bad phrasing. I meant that the probability density is not always time-independent, because psi is not always of that form (i.e. an energy eigenstate)

@ACuriousMind : see my comment, will you? It's a trivial standing wave.

Terry Bollinger

Ladies and gents, it's always a pleasure watching the interactions here. I hope I didn't distract too much with tangents and/or sines over cosines. Till next time...

Terry Bollinger

6:00 PM

Oh hey, one last comment: The formal systems we call "math" -- and I could get into so many fascinating AI and neural processing issues just with that one comment -- are indeed very precise, within the limits that Godel noted (which is basically that all sufficiently powerful formal systems must contain paradoxes).

But beware! If you try to make e.g. a robot understand and do math, a curious problem emerges: You must be very careful and precise in describing how the formal symbols map into real world phenomena.

My computer is shutting down for an Adobe update, goodbye again...

6:22 PM

0

Q: Opinions: When are or are not they answers?

So, I was going through and flagging new answers to old questions (because that's fun) and it dawned on me, are opinions really answers? The obvious answer is "yes" and that wrong opinions should be downvoted. But let me present an example and then I'll open the floor for discussion and laying of...

discussion answers

6:52 PM

@KyleKanos interesting! always wish se made these statistics more accessable. (yes have heard of Stack Exchange Data Explorer)... but that downward-to-zero spike at end is bogus right?

@vzn Yeah, it's an artifact due to the fact that today is Tuesday, so the "week" is only 2/3 days (depends on archival date)

so do you use a supercomputer for your computational physics experiments (refd on profile)?

Yes I do

a cluster? is it at a university?

presumably in the "top 500"?

My university has one, but it sucks terribly

6:58 PM

which one do you use?

I use Stampede

It was #7 in the Top 500 about 1 to 1.5 years ago

cool

reading

And it's apparently still at #7, according to the Nov 2014 list: top500.org/lists/2014/11

est only ~4yr lifetime. boy those things dont last long.

> ...one of the world’s most powerful computers devoted to “open science” research.

wonder what they mean by that.

Yeah, clusters are getting bigger and faster (partly due to GPGPU computing), so their life expectancy is kinda short

Which is what happened with my university's cluster

It debuted truthfully around 60 in the Top 500 about 5-6 years ago

7:02 PM

it says 90% use is decided by expert committee & 10% of the machine use is discretionary by the director. wonder what showed up in that 10%.

Quickly fell off about 2 years go (internet servers were ranked higher)

Then they bought 24 nodes worth of nVidia Teslas and are now back in the Top 100

we chatted about GPGPU awhile back on here & you said it wasnt working out well for the software/ usage....

hard to utilize/ code....

@vzn As far as I know, that 10% is pet projects for the employees at the compute center. It can be quite the variety, from what I understand

it would be very cool if there was some info on the 10% somewhere.

maybe not documented.

doom lol

I suspect you'd have to do a lot of digging to find the answer

Like looking for employees with PhDs and then their publications

7:05 PM

yeah ok.

have you heard of anyone getting a GPU based app to work?

See any Sony, Microsoft or Nintendo video game console :D

But in terms of science based research, there are a few GPU softwares out there that I'm aware of

Someone ported Romain Teyssier's ramses to a GPU (converting it all to C from F90 along the way!)

Someone's ported the cosmology code Enzo to allow for GPU computations

do you think anyone runs stuff like ruby or python on those clusters ever? is it possible?

mean a GPU based supercomputing app.

I don't know about Ruby, but Python is installed on the clusters I work with

cool :)

How many people use an interpreted language for serious computing work, I'm not sure

7:09 PM

yeah guess porting basic libraries/ apps would be the main direction.

Maybe for some dumb-parallelized work

presumably LINPACK has been optimized for GPUs a long time ago.

Like doing a monte carlo simulation where only the RNG seed is needed to be changed between runs

I think many of the @HOME projects use GPUs (molecular dynamics etc).

I think some people use clusters for Big Data (e.g., Hadoop). Whether that's done at a place like Stampede isn't known to me

7:11 PM

(reading bit on ramses...) could ramses model hydrogen clouds? eg the famous hubble "stellar nursery" region?

@vzn It's an n-body/hydro code, so it should be able to do so

hydro= hydrodynamics?

IMO: It's hard to actually figure out the inner-workings of ramses. I tried it for about 6-8 months before I realized that the only person who knows WTF is going on is the author, Romain Teyssier. Rather than have him on all of my papers, I figured I'd run to an alternative choice

@vzn yes

"dumb parallelized" also sometimes called "embarrassingly parallel" in CS circles

lol couldnt get it to work without him collaborating on the paper? citing him is no big deal presumably :)

do you have any papers out?

@vzn That's the word! I couldn't think of the phrase, so I just went with an alternative

@vzn Yeah, he has this (somewhat intelligent) rule that if you ask him for help on his code, he needs to be listed as a co-author (or acknowledgements if it was something trivial)

@vzn Not yet, I'm preparing one though

7:17 PM

wow thats a pretty low threshhold for coauthorship.

Yep. I, being a somewhat arrogant guy, didn't want to seem too dumb and decided to change codes

I asked him for help once and the help he gave still went over my head

hey it sounds like maybe he has a bit of the )( "arrogance" lol... but hey great devs are like that hah.

I started beginning to think that it was by design that he made the code indecipherable to everyone else, so that he could get his h-index bumped up

But I think it's really how you have to design a cell-based adaptive mesh refinement code

exactly something like that thought was just in back of my mind. actually though if the code is so hard to use by others, thats not so great from a "courteous" dev pov.

BRB, important phone call

Actually, probably a BBL instead of BRB

7:21 PM

there are different ways to implement stuff. some of it is more computationally efficient. some of it is equally computationally efficient but harder to interpret. aka "code/ design patterns" etc

np

l8r

thx for details re cluster was curious. got to use supercomputer many yrs ago in college & also on a job. neat tech.

....

(lol had a feeling was gonna get a tumbleweed badge on that Q after a bit... yay!) :p

1 hour later…

8:40 PM

@dmckee I had never seen Community unlock a post before, so I popped over to Meta.SE to see why it did that. I guess there's some timeout period that mods can apply to locks so they automatically expire at a given time? And it defaults to an hour? Was that post supposed to be unlocked or was it just an oversight on changing the timer?

Not that it matters, I am actually just more curious about how the system works... Turns out there's a lot of things that Community user does that aren't easily found in documentation anywhere!

@tpg2114 I think that's right and I so rarely lock posts that I had forgotten.

Seriously, with 9999 out of 10000 users Physics SE is so easy to moderate that I could almost do it in my sleep and we don't need the big guns.

You guys are collectively great.

3

My thanks.

It seems odd that it would attribute it to Community and not the original moderator who locked it

On the other hand, I guess it makes clear what was done actively vs what was done automatically

@tpg2114 That could be it. Makes sense if you know to read it that way.

I know it's always a bit surprising when I see Community actions. The first time I saw a post bumped with no changes, I was really confused. Then approved edits by Community confused me, didn't seem like something that should be "hidden"

So the post unlocking was even more surprising.

9:22 PM

@KyleKanos why should we discourage self-answered questions like this recent one?

@glance Because it's a "do this calculation" homework problem. I don't see any physics concept there

@KyleKanos I get that an equivalent question asked as a proper question (i.e. to get an answer from someone else) would be an homework question with insufficient effort, but in this case he is just sharing something he figured out, it seems to me that this is indeed quite different from the former case

It's not even the "insufficient effort" aspect

53

A: Bite-sizing homework

Questions that can be summarized as "please solve this exercise" or "please plug these numbers into an equation for me" are OFF topic.

The question is literally, "Please solve this for me"

Whether OP answered it or not is irrelevant

@KyleKanos but why not? The point of not accepting these kind of questions is, to my understanding, to discourage the use of the site as a do-my-homework hub. This changes in the case of self-answering the question

but besides all of this, why do you think it is bad for the site to have such a question in it?

I disagree, it is quite the same with OP answering it or not. Homework problems are homework problems, regardless of who solves it

@glance Slippery-slope: allowing HW questions in this format opens the flood gates for other excuses

9:31 PM

@KyleKanos I imagined that reply. However, an exception for self-answering the question does not seem to open any gate, does it?

Note also that I'm not discouraging self-answered questions (I've got one I asked & answered)

(though I agree that in this case the OP could really have phrased the question/answer better)

I don't see how one can phrase that question into a physics concept. It seems to be "How do I solve this transformation"

It could be better by asking, "Why does this transformation hold?" (in which case OPs answer is useless) or "Why do we impose $U(\Lambda)|\mathbf p\rangle=|\Lambda\mathbf p\rangle$?" (in which case OPs answer is also useless)

"How does a scalar field transform under Lorentz tranformation?"

the answer by the OP would only cover the calculations needed, not the physics, but would still be something

2 hours later…

user54412

11:55 PM

@vzn Another thing discretionary time is used for is scaling studies: You ask for some nominal amount of time to prove your code uses the cluster efficiently. Some clusters won't give you normal time unless you do this to their satisfaction.

« first day (1545 days earlier) ← previous day next day → last day (3388 days later) »

Transcript for

Jan '1527

The h Bar

General chat for Physics SE (physics.stackexchange.com). For M...

join this room
about this room

00:00

06:00

12:00

18:00

all times are UTC

site design / logo © 2024 Stack Exchange Inc; legal

mobile