Wolfram Mathematica

halirutan

12:00 AM

@mfvonh I guess yes

mfvonh

@halirutan Thanks. This gives me some stuff to think about.

halirutan

@mfvonh No problem.

@Kuba Are you around?

Kuba

7:36 AM

@halirutan I am now :)

Öskå

11:36 AM

@Kuba Are you around?

Kuba

11:55 AM

@Öskå YEs I am

Öskå

COol. I would have a mission for you, if you don't mind :)

Kuba

@Öskå we will see :P

Öskå

Since you have v9 can you please check the matrices in the two answers here and tell me if I did something bad? Just to check if the matrix from RunnyKine == mine with a given SeedRandom

But brb, I'm going to each lunch :)

Kuba

@Öskå Well,
mo@"FaceCoordinates"

is not working:
{<<1>>}[FaceCoordinates]

halirutan

12:14 PM

@Kuba I already wrote a comment under my answer concerning your Block question.

Kuba

12:35 PM

@halirutan oh, ok, I will take a look in a min. Why haven't you @Kuba me? ;)

Öskå

1:07 PM

@Kuba so this is not working either?

Kuba

@Öskå I dont want to be oracle, I have problems with my mma version lately. I will reinstall it after weekend.

Öskå

mhm, I don't know either :) I just would have liked to compare both answers with a given seed.

But I'm out now, time to vote 8-) Thanks for trying @Kuba :)

halirutan

1:28 PM

@Kuba As OP you should always be pinged when someone comments under his question!?

Kuba

@halirutan yes, but the comment was under the answer.

so it seems it does not apply to the whole topic

halirutan

@Kuba I thought you get a ping anyway.

Now I know that you doesn't.

Kuba

@halirutan I'm struggling with Evaluate, want to help me? ;)

halirutan

@Kuba Evaluate isn't working either

Kuba

@halirutan in different context :)

halirutan

1:32 PM

@Kuba ah, ok. Shoot.

Kuba

Take a look, I was suprised it works:

ClearAll[eq, x, a, f]
eq := x^2 + 2 x + a;
f[a_] := Evaluate[With[{e = eq}, FindRoot[e == 0, {x, 0.5}]]]

?? f

so ok, but why if the above works

the following

Hold[Evaluate[ReleaseHold[Hold[1 + 1]]]]

gives Hold[2] not Hold[1+1]

I mean, FindRoot was stoped to evaluate even though it is in Evaluate.

I don't claim it's a bug, I just need a proper explanation of evaluation process, I don't quite get it now.

halirutan

@Kuba It's quite easy: The reason is that FindRoot returns in its unevaluated form when it cannot calculate the result.

Kuba

@halirutan oh, so it's a side effect

that makes sense

halirutan

@Kuba So if FindRoot would return $Failed this approach would not be successful

Kuba

@halirutan great. my interpretation was that somehow Evaluate acts only on the first level of the expression

halirutan

1:38 PM

@Kuba @Kuba A counter example would be

ff[a_] := Evaluate[Plot[Sin[a x], {x, 0, 2 Pi}]]

Kuba

@halirutan ok ok, I get it :) Thank you :)

Is far as I'm concerned such functionality is available in V10.

halirutan

@Kuba Btw, to get what you wanted originally, you can use Function[f[a_] := FindRoot[# == 0, {x, 0.5}]][eq] to define your f.

(it's strange the inline code using backtick code backtick does not work with linebreaks in chat)

Kuba

@halirutan Please, add this as an answer to mathematica.stackexchange.com/questions/48521/…

halirutan

@Kuba Ah, I just saw that you asked a question. Done.

Kuba

@halirutan and tell me why Function prevents renaming of a while this:

With[{e = eq},
Hold[f[a_] := FindRoot[e == 0, {x, 0.5}]]] // ReleaseHold

is not

halirutan

1:48 PM

@Kuba I remember talking to @Leonid about this a while ago and his answer was something simple like you have to get used to it :-)

Kuba

-.-

mma sucks

:P

each day I have about 10 questions about quite basic things in mma. why why why I've written pages of code already.

@halirutan so Function is some kind of special function or what?

@halirutan maybe it's a good material for a question. maybe then Leonid would say more?

what do you think?

halirutan

@Kuba Yes, and this is good because you are at a stage where you notice those things and try to understand them. This is by far more experienced than most Mathematica users who simply circumvent such stuff by copying the code into the right place.

@Kuba Let me check whether we already have something about this.

@Kuba For instance you find this

> The semantics of With is similar to that of rule substitutions, but an important difference is that With cares about inner scoping constructs (during varible name collisions), while rules don't. Both behaviors can be useful in different situations.

in Leonids answer here.

Kuba

@halirutan so it is With what renamed a_ to a&_?

halirutan

@Kuba Yes, afaik.

Kuba

that's crazy

:P

@halirutan the problem with Leonids answers is they contain a lot of subtle informations that is tough to realise/understand unless you are currently strugling with particular problem.

halirutan

1:57 PM

@Kuba Right. What you should read carefully is this post which contains many information.

Kuba

@halirutan But that's not Leonid's fault. It is mma's ;) and it's great that folks like Leonid participate here.

@halirutan oh, great, thank you for the second link, I have not seen it.

Leonid Shifrin

2:22 PM

@Kuba Usually renaming is always done by external scoping constructs, which analyze the inner ones and rename variables mercilessly if they sense any kind of possible collision. Such renamings are often excessive, in the sense that often also renamed are symbols which don't have to.

Kuba

@LeonidShifrin Thanks :)

Leonid Shifrin

@Kuba But when one uses rule replacements with Rule / RuleDelayed, usually no renamings happen. This is the source of the functional abstraction leak for Function with named arguments, and some other problems of lexical scoping emulation, which many of us mentioned in various places (WReach and myself many times, perhaps others too).

@Kuba Actually, this is a real problem, because strictly speaking, one should not use Function with names args at all - you can never be sure that it will work fine if passed to other functions as an argument, to be applied to something there. I was bitten by this many times, last time very recently.

Kuba

@LeonidShifrin but it's not the case of halirutan's answer, right?

Leonid Shifrin

@Kuba This is in fact a big deal, because it makes it harder to use the style of programming where closures are heavily used - while this style is quite natural to manage large stateful projects.

Kuba

Function with names args - Function[x, stuff] right?

Leonid Shifrin

2:28 PM

@Kuba Yep. In the case of halirutan's answer, it's fine, because he users slots, which don't have this issue.

@Kuba But functions with slots are more limited, because it's not as easy to nest them inside one another.

Kuba

right

@LeonidShifrin Thank you, and @halirutan too, I've learned a lot in this chat today :)

Leonid Shifrin

These scoping issues don't come up terribly often. But when they do, it can be pretty bad. I'd certainly prefer to have a more robust lexical scoping / ways to construct closures. Daniel (Lichtblau) developed a solution, which can be enabled by setting SetSystemOptions["StrictLexicalScoping" -> True], but is is not clear yet whether this covers all possible issues.

@Kuba No problem ) I wish there would be less to learn, would've been much better (since in this case, a lot to learn isn't a good thing) :)

Kuba

@LeonidShifrin Indeed :P that's a price of high level language, isn't it?

Leonid Shifrin

@Kuba Not necessarily. Lisp is also high-level, but doesn't have such problems to such extent. In some sense, yes, though: Mathematica takes the symbolic paradigm to the extreme, where everything is transparent to rule application, even things like variable bindings, scoping etc. This is cool in some cases, since allows you to manipulate scoping programmatically, but it also leads to this sort of issues.

@Kuba It might well be that the requirement of such openness to pattern-matching / rule application, and that everything is an expression (including pure function calls), can not be reconciled with strict and fully robust lexical scoping even in principle.

Kuba

@LeonidShifrin that came to my mind but I'm not a programmer/it specialist so I do not talk about such thoughts ;P

but if you say so.

good to know.

Leonid Shifrin

2:39 PM

@Kuba I can't be sure. I don't have a sufficient CS background to make these thoughts more formal / qualitative - these are just some speculations.

Kuba

@LeonidShifrin Ok, I will not quote you :)

Leonid Shifrin

@Kuba I guess, one way to find out would be to construct a toy Mathematica implementation (just the core language, very minimal), and try to keep both current semantics and fully strict lexical scooping, and see what issues would come up.

@Kuba But that's probably a lot more effort than anyone has free time for :)

Kuba

:)

@LeonidShifrin ok, thanks again, I have to finish reading those topics :)

Leonid Shifrin

@Kuba Talk to you soon ) Gtg myself too.

Kuba

3:05 PM

@halirutan do you think I could undelete my answer? It looks silly next to yours. I don't know. maybe not without proper abstract of todays chat it will be a messand I'm not interesting in elaborating the topic ;p

interested in*

Sektor

4:05 PM

@LeonidShifrin Leonid do you have any exp using Xeon Phi and Mathematica ? Or for this purpose any co-processor ?

Leonid Shifrin

4:32 PM

@Sektor Sorry, but no. I wish I had though.

rm -rf

4:49 PM

@Gabriel Cali is big...!

Sektor

5:00 PM

@LeonidShifrin Why :D ? What's on your mind ?

Leonid Shifrin

@Sektor Nothing in particular right now. Just meant to say that I'd be interested in trying this eventually. I had a few occasions to write some CUDA code, but have not yet tried using it with Mathematica. Generally the topic of using together Mathematica and some massively parallel processor / graphics card has certain appeal to me, and I can think of a few problems where this could be beneficial. No time for this now, anyway.

Sektor

5:20 PM

@LeonidShifrin Yes, I know time is scarce :)

Mind sharing what problems you are referring to ?

Leonid Shifrin

6:44 PM

@Sektor I was once experimenting with parallelizing the computation of various financial indicators on large sets of historical data. It was rather tricky because what I wanted was not all threads computing a single indicator, but dispatching threads so that many different indicators (even on different datasets) were computed in parallel. This was 4-5 years ago, and if I had the time, I'd certainly like to use a more high-level environment (like Mathematica) to drive these experiments.

@Sektor I was writing purely in C / CUDA C, and having some tedious boilerplate code automated by say CUDALink would make my work easier. OTOH, it is not clear to me whether this would be feasible, given that I might need more control than CUDALink can currently provide, such as using memory-mapped mode between host and GPU memory space, using parallel kernels (supported by modern CUDA-enabled GPUs), etc. But in any case, using Mathematica as a high-level dev. environment would've been useful.

Sektor

7:14 PM

@LeonidShifrin Computational finance as a hobby, research or trying to make money out of it ?

If I get my Xeon Phis later this year I would certainly let you use them :D

Leonid Shifrin

7:30 PM

@Sektor I was participating in one project, initiated by one of the local financial companies, which was supposed to provide a web service to test trading strategies on historical data, for Russian stock market. Initially, my role was supposed to be to write the "engine" part, which would be an interpreter for trading strategy language that would compute them on those large datasets.

Sektor

@LeonidShifrin Niiice :)) Did that work out ?

Leonid Shifrin

@Sektor We planned to use CUDA to be able to process many strategies in parallel, perhaps for different users, and be able to have a fine-grained control over load-balancing (basically, be able to attach a given number of CUDA threads to compute a given indicator etc). This proved to be pretty hard, because in this formulation this is not a problem perfectly fit for GPU. So, we decided to not use CUDA at that time, and the interpreter was written in C by another person.

@Sektor But I learned some CUDA programming in the process. Ironically, I ended up as their javascript developer, and developed a pretty sophisticated javascript UI (8000+ lot of rather nontrivial code), which was a kind of a compiler from strategy language to the UI (js + HTML), and contained a lot of parsing technology.

@Sektor This stuff is now in production, but whether by now the project is alive or not, I can't tell, since all original developers (myself included) left it to do something else. This was a side job for me anyway. Made me learn js and CUDA, though, so not too bad. And I got some decent money for this js stuff, too.

Sektor

Sounds good :) So you have never been into FOREX, etc ?

Leonid Shifrin

@Sektor Nope.

Sektor

In any case - u got good money, u used something new- win win

Sticking to physics, uh :D

Are you still researching problems or you are full time working on other projects ?

Leonid Shifrin

7:39 PM

@Sektor Well, it could've played out better, but all the main devs for that project worked part time, and that's not good for the project. They should've had more budget, would allow us to get it done much quicker. And also, there were a lot of things I'd have done differently, but I wasn't the lead developer there.

@Sektor These days, my work at WRI occupies pretty much all my time, so I don't have much time left for other things.

@Sektor What branch of Physics you are working in?

Sektor

@LeonidShifrin Would love to ! :D But I am actually in Informatics/Mathematics

Currently working with a chemistry professor on differentiation smoothing and resolving uv-vis spectra

using wavelets

and that has become my primary research interest

Leonid Shifrin

@Sektor I see. Well, I think what really matters is that you do something you like, and it works. I've left Physics because I started to like programming / CS more, and for a few other reasons.

Sektor

@LeonidShifrin That's what I always say to myself :D :D

@LeonidShifrin I think Physics would have been a better choice

But who said I can't start all over again :D ?

And there's the love for the supercomputers and/or massively parallel machines

BTW You can check this out when you have time

parallella.org

Leonid Shifrin

@Sektor Sure, you can start all over, if you later decide so. Whatever you pick, I think the important thing is freedom. What I disliked about my situation in academia was that in many ways, I wouldn't have the freedom I wanted. Perhaps paradoxically, software developer has more freedom if manages to work on own projects (even part time), because no one can stop you and tell you that something is wrong, off topic, does not fit the format, etc.

@Sektor Thanks for the link, I'll take a closer look when i can.

Sektor

@LeonidShifrin No prob :) But weren't you in a situation where you can actually choose what is interesting to you ?

I mean you certainly made it to PhD, so why were you restricted ?

Leonid Shifrin

7:52 PM

@Sektor I did my PhD in Quantum Field Theory, which I learned already during my PhD. The problem I was working on was very popular (even quite famous) some time ago, but by the time I was doing my work, most of the interesting stuff there was done. And, it was an isolated problem, in the sense that it wasn't connected to some actively developed branch so that I could easily broaden my scope.

Sektor

So you had no other choice, but finish it :) ...

Well, after that ? Don't they have open positions to offer or actually receive a grant and continue ?

Leonid Shifrin

@Sektor I completely solved it, but then I ended up switching to a field my advisor was doing, during my first Post doc. This was interesting, but not interesting enough for me to spend the rest of my life doing that stuff. And to do what I really would like at that point, I'd have to spend enormous effort. By that time, certain other things about academia started to annoy me quite a bit, and other things in life were happening (I got a baby, had to think about money, etc).

Sektor

Hm do you think those problems are local or they apply in general to the academia ?

Leonid Shifrin

@Sektor The right scenario is that you come to start your PhD, de facto already having qualifications of one, and knowing precisely what you want to do. Anything else is a lottery: you are lucky with your advisor and your project - good for you. Not so lucky - and things can be much harder (has nothing to do with the adviser's qualities and a person and a scientist - has to do with whether you are on the same wavelength with him / her). Alas, I didn't realize that when I started.

@Sektor I think they apply in general. The only people very successful in academia, from those I saw around me, fell into one of two categories: either they had done serious research / self-education before starting PhD (being de-facto already very qualified), or they had complete freedom, so they only had formal advisors, but worked completely on their own. In retrospect, perhaps I should've tried to arrange for the second situation for myself.

Sektor

That got me thinking :D ... Unfortunately I still got beef to cook, so I have to interrupt the really interesting conversation ...

Please, if you have not finished your thoughts - write them

I will certainly read them later :)

Leonid Shifrin

8:05 PM

@Sektor Bon appetit! We'll talk later )

Sektor

Thank you :) !

acl

8:40 PM

@Sektor apply to most fields, more or less

@LeonidShifrin no, that would have been disastrous. extremely bad advice.

Sektor

9:01 PM

@acl What do you think it is a bad idea ? And what did work out for you ?

acl

@Sektor Long story.

Just avoid being independent if you want an easy life! (I'm not being cynical)

Sektor

Why* Oh my .. I can't even write a proper sentence today .. :D

@acl So you think it is better to receive a grant, stay quietly in an office and do my thing ?

@acl Expand a little bit more, if you have time, of course :D

acl

9:24 PM

@Sektor No I meant, it's better to work closely with a senior person rather than work independently. The reasons are a) you will most likely produce better work, since he/she can probably know what a reasonable problem will be (whereas you'll have to work it out yourself and it takes trial and error), b) people will recognize the name of your supervisor on the paper and it gets more citations, attention etc, c) he/she will probably support you more strongly this way later, and that is crucial.

Sektor

Actually what is your field ? Can read something yours :D ?

acl

So, a) you need someone to support you (certainly in very crowded fields like condensed matter physics, but to some extend everywhere), b) you are judged on successes, not potential, independence, or anything like that, no matter what you're told; and you will obviously be much more successful if you are doing technical stuff for an idea suggested by someone who's been working (most likely) on the same thing for 20 years, than if you try to actually come up with a problem yourself.

Sektor

Can I read *

acl

@Sektor haha yes, ask me for a direct link to a pdf with my name immediately after I badmouth all my colleagues en masse :)

theoretical physics (mainly quantum dynamics nowadays)

Sektor

@acl Well, I dont think you are offensive or anything like that ... :D

Thats just your opinion :D

@acl I have to dig into arXiv

acl

9:30 PM

@Sektor Well not offensive no...

@Sektor How come? You said you work on mathematics

Sektor

@acl I have been keen on physics and I love to read :D

@acl No matter what comes before my eyes - if it is interesting - I will read it :D

acl

@Sektor Oh OK sure. There's a huge amount of stuff on there so it's hard to decide what to read. I look at the new preprint in my area every day (I just go through the titles to see if there's anything interesting), and even then there's too many.

Sektor

@acl Oh, I was gonna look for your work :D

@acl But on the hand I constantly search arXiv, so it is true what I said :D

other hand*

my god

acl

@Sektor oh right!

Sektor

I gotta go to bed I dont even know how we end up here. I cant even write my name properly ... :D :D ..

how we ended up :D

that's enough :D :D

acl

9:36 PM

@Sektor ha indeed! night then

Sektor

night :)

and thanks :D

Leonid Shifrin

@acl By being independent, I didn't mean to work with no one. I meant to be free to pick who you work with, be it someone in your institution or outside. Most fellows of mine who ended up getting some really good results, worked either together (teamed up), or sometimes with external people. A few made good stuff in collaboration with the faculty, and I've seen a lot of cases of people getting completely demotivated when they stuck working only with their advisors.

@acl In my case, I de facto worked independently, and that's the only reason that I got the results I got. But I wasn't independent to a degree I really would've liked - for example, I wasn't able to publish my main results for a couple of years, because my advisor wanted to add some more stuff, while being himself busy with his own projects. And, he is both a great person and a great scientist - it's just that we've had very different approach to things.

acl

@LeonidShifrin OK, what I meant though that if you do work close to what someone famous is doing (ie, take his research and push it a bit under his guidance) then you're going to be much better off than if you work alone, by which I mean if you work on your own thing, perhaps in collaboration with others. I think we are in fact saying similar things

Leonid Shifrin

9:51 PM

@acl If you are / were lucky to work with the person with whom you are on the same wavelength, so much the better. All I was saying is that finding such a person should not be a game of the blind, or done by trial and error. Those who made some research in their area of interest before PhD, have better chances to find a better match for them among more experienced colleagues. Often, they already have some collaboration before actually coming to a PhD program, where they merely continue it.

acl

@LeonidShifrin Well, for my PhD I ended up just working independently, just because that is how I normally operate (ie, my supervisor is a great guy and very smart, but I ended up doing most of my PhD on something he didn't know anything about--we still keep in touch, he's a great guy actually). And then I kept doing this for a few years, I think the first paper on which I wasn't the only or oldest author was my 8th or 9th. And they were not garbage, there were two PRLs in those etc.

But despite this I had a harder time than my peers who had similar records but well-known names on their papers. Partly, as I said, name recognition, partly that those well-known people are going to vouch for you on applications, etc.

Leonid Shifrin

@acl Yes, sure, you are describing what I'd consider a close to ideal situation. Also, it is good to keep in mind that in the US the system is rather different from Europe, and PhD students are much more on their own. I've seen people struggling with their PhD all around me.

acl

@LeonidShifrin Right, that is true. Not so easy to do in the UK though (not much research experience as an undergrad).

@LeonidShifrin I guess it depends where in Europe! I had an officemate in London who was told during the viva "hmm, no, not very good, try again in 6 months" (or a year, don't remember). She was not happy!

Leonid Shifrin

@acl I've had similar experience, except I didn't have so many papers done completely on my own. But in most of my papers my results were either the meat of the paper, or at least a crucial contribution. One big blocker was that I had to waste close to 2 years taking courses, passing all sorts of stupid exams and tests, etc, and teaching. Then, I got my results not seeing light for at least a year and a half for what I think were quite subjective reasons.

@acl My advisor is a great guy, but like you, I was doing stuff he wasn't involved in at the time (he did some work on that in the past though), and our ways of thinking and approaching problems were quite different. I thought I needed that at the time, since I was a bit too informal in what I was doing, while he was very mathematical. But in retrospect, I think I could've been better off working with someone who had an approach closer to mine (we had such people in our group too).

acl

@LeonidShifrin Well that is part of the problem though, you just don't have the experience to do all the stuff that needs to be done after the calculations. Of course if the supervisor delays you then there's not much you can do.

Leonid Shifrin

10:01 PM

@acl Re: Europe - you are probably right.

@acl In my case, the biggest problem was to constantly translate my ideas into the language he would understand. It turned out that in all major points of my work on my thesis problem(s), I was right despite his opinions, not with the help of them. It was pretty hard to convince myself that I am on the right track when I was constantly hinted that I was not.

acl

In any case, working in academia, you're a lot more constrained than people realize. Especially if you're worried about getting a job at a good place etc, then you don't have any freedom to play, despite what it looks like. And if you don't care about where you are, you'll most likely end up unemployed because there are others that do and not all can get a job at Harvard/Cambridge/whatever, so they just trickle down.

@LeonidShifrin I see, self doubt. Can't say I ever had any :)

Leonid Shifrin

@acl That's not to say I didn't learn from him - at the end, it was this communication which taught me how to separate valid arguments from hand-waving. I am just saying that there is more than one way to learn this, and I learned it the really hard way.

acl

Well sure. On the other hand, a lot of this is just sheer luck. Only in retrospect do you see what happened and how.

Leonid Shifrin

@acl Re: a lot more constrained - completely agree. That was one of the things which turned me away from it. I did quite well on my first postdoc, could find another one easily, I am sure. I just lost the drive.

acl

@LeonidShifrin Yes, usually, at some point it takes a very dedicated masochist to not give up.

Leonid Shifrin

10:07 PM

@acl Re: sheer luck - but that's precisely my starting point. I've seen people coming well-prepared in the sense that they knew the field before they started PhD, and they all had great head-start and most of them used that to always be on top of the wave. They kind of eliminated this luck factor. For others - yes, pure luck.

@acl Well, I've spent 15 years on this stuff, and I liked it a lot (still do). I just had other things in life happening, and I realized that I won't be satisfied by the level of freedom I can get in academia. I might still return to some research, but only after I get my financial situation totally under control. For example, I'd like to spend some time reading a number of original papers from the past, and re-learning some parts which I originally studied more like a good student ...

@acl ...rather than an original thinker asking questions and answering one's own questions. I knew that I won't have this luxury if I stay in academia.

acl

@LeonidShifrin Sure. But if you do a 3-year undergrad in the UK plus a one year taught masters, it's not at all comparable to a 5 year diplom in germany or a 3 year undergrad plu 2 year msc in NL (where one year is also full-time research). Obviously. But you don't know that if you have done your undergrad in the UK and nobody told you, do you?

Leonid Shifrin

@acl Besides, I dislike many current trends in science. The quality of papers dropped dramatically, on the average, in comparison with 70-s and even 80-s, while the number of them increased just as dramatically. People stopped working on important problems, in favor of problems they are sure they can solve reasonably quickly, to get another paper out. Fashions drive research more than ever. Etc, etc.

acl

@LeonidShifrin Oh yes! Reading too much leaves little time for publishing! haha

Leonid Shifrin

@acl Agree on the UK.

acl

@LeonidShifrin That's what everybody says. I don't know how true it is though. Suppose there were garbage papers in the 50s. How would you have come across them?

Although it's true that the rate of publication you need to survive nowadays is much higher than it used to

(and we're still the same type of humans so obviously quality has fallen)

Leonid Shifrin

10:18 PM

@acl The problem is not that there wasn't garbage, but that the signal to noise ratio was much higher. And the reasons were, I think, that first, judgement of the quality of the publications was left to the scientific community alone, that people were trying to solve real problems rather than problems they can solve, and that there were not so many people involved in science.

@acl The problem with falling quality is that this isn't a linear function of the number of required publications per year. This is similar to a Quantum Zenon effect - continuously observed system never evolves (the "system" is a scientist).

@acl And the researcher is nowadays continuously being "observed" and "measured" by various committees, etc.

acl

I don't know, everybody complains about it yet someone uses these criteria when you apply for grants etc. Methinks some of those complaining aren't honest

And well if you select people based on criterion X, they'll probably think criterion X is a good way to choose smart and capable people :)

Leonid Shifrin

10:33 PM

@acl Well, since I no longer belong to the Physics community (at least currently), I have no subjective reasons to complain. But, it so turned out that consistently the best papers about things I was working on, belonged to the period of 70-s - 80-s, with a few from 90-s. So, at least in my case, my complaints are based on observations having not much to do with the "common complaints". And, when I talked to other people, many of them were having similar observations.

@acl You can even see that when you look at how the papers were written. If you look at some papers by Sidney Coleman, or people like Callan, Dashen, Gross, Polyakov, Weinberg, Schwinger, ... (sorry, I am mentioning people from my former field), you can see that they were talking about principles, doing the necessary computations along the way. Now people are just doing computations, most of the time. The context is gone.

acl

@LeonidShifrin I do agree with you, papers from the 50s, 60s, 70s and 80s are much better than later. But maybe that's just because we would otherwise not have read 50 year old papers! I don't know.

But overall I do think they probably just were overall better on average

@LeonidShifrin I've certainly read papers by Dashen (and Coleman). I don't disagree.

Leonid Shifrin

@acl Yes, part of that must be true. But in the past, people didn't bother to publish if all they had were computations, without a proper justification of why doing that in the first place. I think, strings played a bad role here, in popularizing the practice of publishing stuff which can't even in theory be tested by a contact with physical reality.

Kuba

Sorry to interrupt but isn't it "survivor bias"? About papers from the past?

Leonid Shifrin

@Kuba I don't think that can be explained just by that.

Kuba

I agree, Im curious what your opinion

acl

10:42 PM

@Kuba That's what I am suggesting. But to some extend he's probably right. The amount of publishing you have to do nowadays to survive is much, much higher than before.

Leonid Shifrin

@Kuba I rather think that science is just one example of modern trends, but if you look into other sides of life, you'll find the same things - over-consumption, excessiveness in everything, ignoring long-term goals, going after easy stuff, creating your own virtual worlds, etc.

@acl I think the h-index is a culmination of that: measure of being average.

acl

@LeonidShifrin Being popular

I imagine the actual explanation here is that you and I are misfits, not that everybody else is "doing it wrong" :)

Leonid Shifrin

@acl I doubt it. The questions to ask I think are what people are currently doing, and why. I don't think many people would be able to clearly answer those, taken as conceptual rather than technical questions.

Kuba

@LeonidShifrin @acl I'm not aware how exactly young scientists were founded in the past but now you may do a phD only because you like e.g. quantum mechanics but you are lazy

acl

@LeonidShifrin That's right.

Kuba

10:51 PM

I think people are the same, just now there is more folks in "science" that would have not been 30 years ago

But I was not parsing this subject so it's only an opinion.

acl

@Kuba Where? In most places I know competition is stiff, you aren't going to get a phd position if you're not dedicated

Leonid Shifrin

@acl I also think that the story is much more banal, and basically was are all just slaves of certain social laws, governing the behavior of large masses of people, and those laws depend primarily on the number of people involved, and the complexity of their social structure. In the past, the community was smaller, and "mean and lean". Now, it is fat and unwieldy

Kuba

Moreover, I have to go sleep :) so good night @acl @LeonidShifrin

Leonid Shifrin

@Kuba Night )

Kuba

But I was not parsing this subject so it's only an opinion.

acl

10:53 PM

@LeonidShifrin OK when you put it this way it certainly does sound like a good explanation, yes

maybe there is something to this

@Kuba night!

Leonid Shifrin

@acl All right, I got to be going too. Was a nice discussion )

Kuba

But I was not parsing this subject so it's only an opinion.

acl

@LeonidShifrin night!

Leonid Shifrin

@acl Night!

Kuba

sorry for spam, connection problems. anyway, I'm out :)

rm -rf

11:01 PM

@acl @LeonidShifrin Modern academia has also gotten to the point where there are a LOT more people fighting for an ever decreasing pot of money, promotion and tenure decisions being made by non-academics and people outside the field (not just the decision itself, but the related negotiations like salary, perks, etc.) to whom a person's "worth" can be reduced to just #of papers/yr (regardless of quality) and size of grants, etc.

Since there are very few openings in academia compared to the applicant pool (unlike in the 50s-80s), everyone values quantity more than quality in order to survive.

acl

@rm-rf said the most well-connected frog academia has ever seen!

(hi)

rm -rf

Hi :)

@acl "well-connected"?

acl

@rm-rf for a frog

rm -rf

Hehe

Leonid Shifrin

@rm-rf Yep. Couldn't let me leave in piece, huh?

rm -rf

11:08 PM

@LeonidShifrin I hope you're still intact! :D

Leonid Shifrin

@rm-rf Lol ) Re: quantity over quality, non-academic - this is what I meant by "being continuously measured" and "over-consumption" etc. I totally agree.

rm -rf

@LeonidShifrin Ah, yes. I'm sure we (the three in the room now) all have quite similar experiences... You've probably summarized my thoughts well above, I just hadn't read the entire conversation :)

@acl has somehow had the will to continue in academia... mine is diminishing real fast (perhaps for the good!)

Leonid Shifrin

@rm-rf So, you sound quite disappointed too.

@rm-rf I would actually like to get back to research, but on my own terms, and completely managing my own time, not being subject to many modern rules, regulations, etc. It is a path to become a crackpot, yes, but it also seems one of the few remaining routes to be able to do original research too.

Matariki

Hi, anyone of you know of a color scheme that works well in color and in B&W. Some publications are printed in color but others only in black and white. What are you guys using and is such a "safe" color scheme predefined in Mathematica?

acl

@rm-rf Well I am obsessed with physics, hate the rest (I most likely hate it much more than you do actually)

rm -rf

11:16 PM

@LeonidShifrin Quite. When slapping on additional names on papers for nothing becomes common place and then they turn around and "sell" it as their own idea to funding agencies (and undercut me because of seniority), it starts to feel like you're no longer working to solve interesting problems, but simply so that you can some day become a slave master.

acl

@Matariki I use "BlueGreenYellow" for this, but it is probably not the best choice out there

@rm-rf ah, OK I never had that. Although the flip side is you also don't have their support when you need it.

Matariki

@acl. Thanks, I'll give it a spin. I sure I saw a paper on this topic but I can't find it for the life of me

acl

The grass is always greener on the other side

Leonid Shifrin

@rm-rf Sorry to hear that you've had these experiences. At least, I was lucky to work with exceptionally honest people, during all my time in academia. We may have had disagreements on some things, but nothing remotely close to the things you describe were happening. May be, because they just were the best in the field (although that is not always enough).

acl

@Matariki I also like using grayscale everywhere, but sometimes it's not enough

Matariki

11:20 PM

@acl True but some color translates badly to grayscale

I usually start in color and then look at it in grayscale but often it is a guessing game on what works well

Sektor

@acl What's your preferred color scheme for presentation/publication/report graphics ?

rm -rf

@Matariki In general, a color scheme that has an increasing/decreasing brightness from 0 to 1 works well with BW. So, "DeepSeaColors", "SolarColors", "SunsetColors". If you're only interested in showing the deviation from a central value, schemes that have the same brightness at the ends and decreases/increases to the center works well. E.g., "RedGreenSplit", "TemperatureMap", "ThermometerColors", etc.

acl

@Sektor I'm really no the best person to ask, I usually stick to "BlueGreenYellow" just because I like the colours! But there are systematic ways to do things better

Leonid Shifrin

@rm-rf But then I can understand your attitude against staying in academia even much more. For me personally, I just realized that achieving something in the field of programming / software development is more straightforward these days - if you become a truly great programmer, you can be unstoppable in what you do, and you don't depend on the rest of the social infrastructure much or at all.

acl

@rm-rf Yes that's it @Matariki, @Sektor

@LeonidShifrin I've thought of that a number of times. Shame I suck at programming.

Sektor

11:22 PM

@acl So, pretty much what I am using :D

Leonid Shifrin

@rm-rf That is, in theory ). As always, in practice that's more complicated )

rm -rf

@LeonidShifrin Not so much as "dishonest" as a systematic practice in the field where proof reading is equated to co-authorship... of course, depending on your PoV that could be "dishonest" or something "wrong with the field", but it's not uncommon.

Leonid Shifrin

@rm-rf That's equal sign to me

@acl You are being too hard on yourself.

Matariki

Thanks, @rm-rf I'll use those. When I have various graphs in the same plot It is hard to distinguish them in BW and I don't like using dashing etc

halirutan

@Matariki What you never should use is something like the famous Solarized scheme. Those colors are picked to have a certain equal brightness feeling in the eye. Printed in B&W you cannot distinguish them.

Matariki

11:26 PM

Hi, @halirutan how are you? Thanks, that color scheme I haven't used but I will try it next time I want to upset the reviewers. ;-)

acl

@halirutan I think what @rm-rf said is the "best" way of doing things, eg, "TemperatureMap" seemed to work well when I tried

Sektor

@Matariki Here's a nice link

blog.visual.ly/rainbow-color-scales

Leonid Shifrin

@acl One other thing I do like about programming is that there isn't a single right answer. I do like a "single right answer" situations, but not when all things you work on have to be like that.

acl

@LeonidShifrin not really, I can see that the way I think doesn't work for programming. But thanks :)

@LeonidShifrin Sometimes physics is like that too, for example, if you're trying to set up a framework (or viewpoint if you prefer) rather than just do a specific calculation

rm -rf

@Matariki If you want to tweak the existing color schemes a bit to suit your tastes, you can try using this package which provides interactive manipulation... no guarantees. A "BW simulator" for that is an interesting idea though!

halirutan

11:28 PM

@Matariki Hehe, I'm fine. Yes, nice for messing with reviewers.. As the red portion of the plot indicates... :-)

Matariki

@Sektor thanks I'll read it later

Leonid Shifrin

@acl Yep, that's true.

Matariki

@rm-rf Ok, I will download it and give it a spin tonight

Sektor

Okay guys I call it a day

:)

night all of ya :)

rm -rf

adios

acl

11:30 PM

@LeonidShifrin Of course each anonymous referee will probably have (or form on the spot) their own contradictory framework and consider it the absolute truth :)

Leonid Shifrin

@acl I agree that one have to think differently for programming - it is more an art than science. At least, that's what I find valuable about it - it is an expression medium. Although, a number of prominent physicists do Physics like an art.

@Sektor Night!

Matariki

@halirutan You certainly use the right phrasing there hehe. Isn't it quite early for you?

acl

@LeonidShifrin I think any who is worth anything do it like this

Leonid Shifrin

@acl All right, perhaps you are right.

acl

@Matariki probably early evening for him :)

halirutan

11:31 PM

@Matariki Had a nap with my daughter at 2000 and now I'm awake again :-)

Matariki

Bye @rm-rf

Michael Hale

Is it still possible to do physics research without programming?

Matariki

@halirutan Those tricks I have used to but now the daughter is too old

@acl time is relative

Leonid Shifrin

@MichaelHale I am sure it is, but then you won't take advantage of the technological progress. Perhaps, a questionable advantage, though.

acl

@LeonidShifrin ah, you might be amused to know that I am calculating level statistics for quantum systems as we speak! I had forgotten you'd worked on similar things

@MichaelHale sure. You're just more limited in what problems you can tackle (guess Leonid has a different viewpoint here though!)

Leonid Shifrin

11:33 PM

@acl I worked on eigenvalue statistics of Dirac operator in QCD

acl

@LeonidShifrin I'm looking at strongly interacting systems (ie, no small parameter) out of equilibrium

with disorder

Leonid Shifrin

@acl I also know that there are strong connections of this with quantum chaos and periodic orbits

acl

@LeonidShifrin yes, in my case it's a "quantum chaotic" system. but that is somewhat ill-defined as there is no classical analogue

Leonid Shifrin

I've taken a course in diagrammatics for disordered systems, by Igor Aleiner.

acl

@LeonidShifrin I've read their papers on many-body localization

Leonid Shifrin

11:35 PM

I was the only one in that year to complete all exercises )

acl

in fact that is what I am doing right now, but out of equilirium

Leonid Shifrin

I still remember a cool problem on Van Hove singularities

@acl Oh, cool

acl

Well in my case a) there is no small parameter so no perturbation theory is going to help you (or, no quasiparticles), b) I am in a time-dependent situation so you'd need to also do another expansion before you do diagrammatic perturbation theory. so it's hopeless and I just do it numerically

(I am being deliberately vague as the relevant keywords will immediately localize me via google and it may not be such a good idea to have this right after expressing negative opinions about academia!)

Leonid Shifrin

@acl Quite clever :)

rm -rf

@acl Do you have anything you need wiped from the transcipt?

acl

11:39 PM

@rm-rf ah no I'm joking, I say the same things at conferences

Leonid Shifrin

@acl Fortunately, I am free from such restrictions )

halirutan

@rm-rf Since you are here, a question: I had spoken about this already some time ago. Is it possible that we make a Meta thread for rating/measuring different things? I would like to know things like How many people use editors/IDE for package writing, what is operating systems are used, how many users write link code like JLink, MathLink, etc, how many people write packages in preference to evaluating notebook cells over and over again.

rm -rf

@halirutan Oh sure, go for it!

halirutan

@rm-rf @rm-rf One could make a general question on meta and then every answer is one specific topic with several comments as possibilities which you can upvote.

It really would give us the opportunity to evaluate user behavior to predict the impact of things like the IDEA plugin.

Leonid Shifrin

@acl Did you ever use the periodic orbits / WKB for your computations? This is one topic I am (was) quite interested in, but never actually did anything in that direction. I mean, the Gutzwiller's papers, and others?

rm -rf

11:42 PM

@halirutan Create a sidebar ad too, to alert users to the poll.

halirutan

@rm-rf That would have been my next question whether this is OK for you. Thanks

rm -rf

@halirutan Does JetBrains give you any download stats?

acl

@LeonidShifrin I am somewhat familiar with this, but in my case it is not applicable as I focus on systems which either a) have classical analogues but are deep in the chaotic regime, thus periodic orbits are a negligible fraction of phase space, or b) have no (obvious) classical analog

Basically I want to be as far from anything regular as possible

Leonid Shifrin

@acl Oh I see.

acl

(except one paper this year where I specifically looked at an integrable case, and I could do the whole thing analytically--that was nice!)

Leonid Shifrin

11:44 PM

@acl Does this generalize to other sides of life for you ? )

halirutan

:15723387

acl

@LeonidShifrin funny that you should ask...

:)

@halirutan nice :)

11:45 PM

@acl I very much prefer exact solutions / analytical stuff, perhaps too much

@acl Just kidding

acl

@LeonidShifrin oh, yes, so do I. So does almost everybody I think. But most people can't do it, and even if you can, it's too limited

rm -rf

@LeonidShifrin Is your colon key stuck? :)

acl

and too fine tuned. not "generic" (ie you don't know if what you found works "everywhere" or just for integrable systems)

Leonid Shifrin

@rm-rf Sometimes :)

acl

@rm-rf it's a russian smiley (

rm -rf

11:47 PM

@acl I think @LeonidShifrin is hinting to us that he's very sleepy, hence no open eyes... |)

halirutan

Leonid Shifrin

@acl Yep, I understand. I got spoiled by the world of 2-d integrable field theories and RMT. But I wasn't always like that.

acl

@rm-rf hmmm |-|

Leonid Shifrin

@rm-rf Yep, that's right. So, apparently you can read other's minds.

rm -rf

hehe

Leonid Shifrin

11:48 PM

@rm-rf @acl @halirutan Night, guys! Was good to talk.

rm -rf

Bye!

acl

@LeonidShifrin night! I'll keep demotivating @rm-rf for both of us

Leonid Shifrin

@acl Sounds good!

halirutan

@LeonidShifrin Night.

@rm-rf Another question, sometimes when I visit other SE sites, I get logged in automatically, sometimes I don't although I'm logged in at M.SE. Is this purely done with cookies or can I somehow influence it that I only have to be logged in at one site to be automatically logged in when I visit another?

rm -rf

@halirutan I believe it is done entirely with cookies, although the actual details are a lot more complicated:

100

A: How does SO's new auto-login feature work?

I'm going to provide more than just the technical details here, as I think there are alot of implicit assumptions about global login that aren't quite correct out there. Accordingly, this is going to be really long. Design Requirements A user having logged into any SE-site will be automatic...

Blog - Stack Exchange: Global Network Auto-Login

Jeff Atwood on September 10, 2010

We now support automatically logging in to any site in the Stack Exchange network.

By that I mean, as long as …

You have recently logged in to any Stack Exchange network site You hold an existing account on the target site you’re navigating to You are using the same OpenID credentials

You hold an existing account on the target site you’re navigating to You are using the same OpenID credentials

You are using the same OpenID credentials

… the site you’re navigating to will automagically log you in! You’ll see a notification bar at the top to let you know when you’ve automatically logged into a site. …

There might have been additional behind the scenes changes since 2010 (for instance, they support https now)

halirutan

11:57 PM

@rm-rf OK, I read through this later. Thanks, because this is sometimes annoying. I had to switch to Chromium (instead of Chrome) yesterday, because this stupid Mathematica C&P bug now infected Chrome too and I cannot simply paste code here without additional garbage bytes.

With the new browser I have to login again for each site.

rm -rf

@halirutan I haven't had any issues with Chrome (OS X)

halirutan

@rm-rf Nope, it's a pure Linux/X thing.

(Work = OSX, Home = Linux)

Transcript for