The h Bar

Mike Miller

17:00

I know I'm not

ACuriousMind

I...probably should've paid more attention to the cross-lists, then :P

Danu

Yeah, they are a pain sometimes

But there's a lot on SYK right now

(not that I read it)

John Rennie

@MikeMiller I think the point is that mathematical physicists are attempting to construct a mathematical model that may one day be of relevance to the real world.

By contrast mathematicians study the maths for its own sake.

ACuriousMind

@JohnRennie If you knew what strange types of models are sometimes studied you might revise that opinion...

Danu

But isn't that just what they tell the funding organizations? :P

David Z

17:03

Quick note: that's the end of the official chat session. Sorry not to have a plan this time, but I will for the next one in two weeks!

John Rennie

@Danu Hey, I got my PhD funding by claiming it was relevant to semiconductor fabrication (it wasn't :-)

Mike Miller

Without going into the arguments you've heard plenty of before, there is skepticism (from the outside and inside, as is visible) that there is any such relevance.

The virgin america safety video is going to make me kill myself.

vzn

@Danu [citation needed] (other than EMDrive plz)

John Rennie

@MikeMiller I think we can afford to indulge the meanderings of mathematical physicists in the hope they will one day wander back in from the cold. It's not as if it's a major drain on the economy.

Mike Miller

@JohnRennie My complaints are in name only. I don't mind the funding.

I just don't think they're physicists. :D

John Rennie

17:06

A rose by any other name ...

Danu

@vzn nasa.gov/centers/glenn/technology/warp/warp.html

ACuriousMind

@MikeMiller Maybe we should start saying you're not a mathematician because gauge theory is physics, after all ;)

Danu

@MikeMiller But as you also noticed, HEP guys are flocking to cond-mat now.

Mike Miller

Certainly the likelihood of my work doing anything useful is zip. I don't mind.

@ACuriousMind You can look at my first paper whenever it gets out and try to make that argument if you like.

@Danu With skepticism from the people who care about the real world about whether it makes any useful predictions or statements about the real world...

ACuriousMind

Oooh, a challenge!

Mike Miller

17:09

My goal is March but writing is so hard

vzn

@MikeMiller sounds like Hardy. who ended up being somewhat wrong about that. ps have you seen that new movie? :)

Danu

@MikeMiller Meh. The stuff they now mostly work on, it seems, has jsut not been explored yet by cond-mat guys.

Mike Miller

Yeah, it was alright. Good luck applying what I do. And poor Hardy.

All he wanted was for his work to not be used to hurt others...

At least I don't work in stuff like predictive policing like others in my dept.

vzn

@MikeMiller yes, like his (antwar) friend Russell (also in the movie). it seems that war is one of those near-eternal certainties close to death & taxes...

John Rennie

@MikeMiller hey that could be useful for managing the review queues.

We could use it to ban people before they offend.

Mike Miller

17:13

It's not going to be eternal if we all kill each other.

ACuriousMind

@JohnRennie The SE version of Minority Report? :P

vzn

SE: sledgehammer to crush a fly™

John Rennie

@ACuriousMind yes. When people click the New Question button they get a prediction of how likely their question is to be closed :-)

vzn

@JohnRennie actually SE already has something like this based on keywords afaik, or was investigating something like it, trying to give an indicator of question quality, cant remember where saw that....

John Rennie

Hey, here's a proposal for the next chat session - the New Question button should take you to a quick questionaire along the lines of John Baez's crackpot index.

David Z

17:18

@vzn They do pop up a notification when certain keywords appear in the question title, advising you that the question appears likely to be closed.

John Rennie

If you fail it your question is refused, your bank account is hacked and your boss gets an email explaining how deep your carnal desires for them are.

vzn

86

Q: Heuristics for detecting a bad answer?

A lot of bad answers are slipping through the cracks on Stack Overflow. You can see some examples at Thanks a lot for this post ... and other first time user curiosities I've been paging through hundreds of these first answers by new users and I've identified a few heuristics that, when applied...

John Rennie

Could we borrow Deepmind off Google to check the questions before they're posted?

vzn

217

Q: Can we prevent some of the low-quality questions from entering our system?

Stack Overflow has been wildly successful. And maybe in some ways too successful. I am concerned that Stack Overflow is being inundated by a stream of low-quality questions from users who are accidentally poisoning our well -- by turning off and turning away the core answerers who do all the rea...

discussion status-completed questions new-users

DanielSank

@JohnRennie don't see why not.

John Rennie

17:26

Although it was meant to be a facetious comment, I wonder if there is some interest in applying that sort of analysis to the SE.

David Z

What, using Deepmind?

I'm sure there's lots of interest, it's just a matter of resources.

John Rennie

After all it's a machine readable data source with a quality indicator (of sorts)

Mike Miller

@JohnRennie Frankly, I think he and I would hit it off.

vzn

fyi Deepmind is a (mostly research) company recently acqd by google, not really a particular technology yet. deepmind.com the company has achieved notable successes on a few different (AI) prjs, eg recently language translation & Go, & originally video games!

DanielSank

Blarg

Kyle Oman

17:40

@ACuriousMind If you're a fan of Witten... have you seen this one? On the hypothesis that cosmological dark matter is composed of ultra-light bosons

Mike Miller

I want someone to post a paper like that

"On the hypothesis that ___"

text: It's real dumb

ACuriousMind

There's that one paper whose entire abstract is "No."

Mike Miller

link?

ACuriousMind

The title is some silly question I forget, which

Mike Miller

also, who are you now?

Kyle Oman

17:42

@MikeMiller I saw a recent one that comes pretty close The ΛCDM simulations of Keller and Wadsley do not account for the MOND mass-discrepancy-acceleration relation, it's scathing

John Rennie

@MikeMiller Well dark matter research is getting a bit of a pickle at the moment, so there's no harm in exploring the more unlikely scenarios.

DanielSank

@JohnRennie I bet some folks at Google would be interested in a bad question filtering system.

I'll ask around.

John Rennie

> Keller and Wadsley (2016) have smugly suggested, recently, that the end of MOND may be in view.

Mike Miller

rather unprofessional

John Rennie

Niot at all tendentious then :-)

@MikeMiller It's written by Milgrom who originated the MOND idea.

Kyle Oman

17:44

Well, Mordehai Milgrom did invent MOND

ACuriousMind

@KyleOman No, didn't know he also did that sort of cosmology

John Rennie

Milgrom seems to be getting a bit unravelled these days.

ACuriousMind

@MikeMiller You mean my picture? It's Durance from Pillars of Eternity

Kyle Oman

Perhaps, but his rebuttal is also broadly correct. Opinionated, but he raises the right counter arguments. Not saying MOND is right, but K&W certainly didn't prove it wrong (yet).

Mike Miller

Any good?

ACuriousMind

17:46

@MikeMiller Pillars? Yes, very, although it of course lacks the nostalgia factor Baldur's Gate has

Mike Miller

I want them to stop with their isometric games and tell me they're going to make Fallout: Two Sun or something like that.

I've got a friend running through New Vegas. He just met Caesar and is having a good time running through HH right now. I want to see if he just makes Caesar 2.

ACuriousMind

@MikeMiller Well, I don't think Obsidian knows how to make non-isometric games ;) In the cases where they "made" one (F:NV, KoToR 2) they were already handed a pretty complete engine

Mike Miller

Sure, but Bethesda can hand them anotherr.

ACuriousMind

I'm...not sure the current iteration of the engine (Fallout 4) is able to handle dialogue more complex than "Yes/No/Sarcastic Yes" :P

Mike Miller

Never touched it. Probably never will.

Dialogue systems are probably not hard to enhance in the engine.

ACuriousMind

18:01

No, probably not

But I also don't know what Obsidian's next project is now that Tyranny is out, anyway

Mike Miller

I want another Fallout so bad.

SO BAD.

I should probably write my talk but man, not feeling it

Danu

What's it about?

Mike Miller

uncountably many exotic R^4s

I know what to say, but I need to get it on paper.

Danu

Ah, okay

That's a comfortable position, haha

Mike Miller

I also don't know what's going to go into the last 20 minutes yet. But I'll figure that out.

Danu

18:13

Skip dinner or take a break now to get food?

ACuriousMind

@Danu Food

Mike Miller

Skip dinner but take a break now anyway

Danu

Hahhaa

ACuriousMind

Only skip dinner if you have some food in the house if you get hungry later

Nothing worse than getting hungry at midnight and not having anything proper to eat

Mike Miller

the state of politics worldwide

ACuriousMind

18:29

Huh, rxiv.org is an alias of viXra.org.

Typosquatting - not exactly what you expect from a "scientific" pre-print archive.

Mike Miller

0

Q: "Branching" topological spaces

I am trying to understand why in general relativity we require spacetime to be a Hausdorff space. (I am a physics student not a mathematics student by the way.) The 'handwavy' reason that is usually given is that non Hausdorff spaces can have "branching", which we do not want to have for spacet...

general-topology

Simon

Hello everyone. I need some book recommendation. I figured the chat room is a better place to ask that kid of question than the Q&A forum. I'm trying to study classical mechanics beyond what I learned in grade 12 and 13 (in the "strong" class in a private school). I got classical mechanics by Goldstein, but it says in the preface that it was written for a graduate course, and after reading the first 100 pages, it is definitely above my current level. What book would you recommend?

On amazon, I've seen people recommending Thorton Marion "Classical dynamics", Keyth Symon's "Mechanics" and Kleppner Kolenkow's "An introduction to mechanics"

from the words "An introduction to..." I am guessing the 3rd one might be the easiest one

bolbteppa

18:58

@Simon basically read Kleppner, which is local, calculus-based physics, and learn 'calculus of variations', which is a calculus where your variables are functions, and then have a look at Goldstein, which is global, calculus of variations based physics.

Physics Meta

0

Q: Reliable sites for science research paper references

I recently posted a question based on: "spherical solutions to the Schrödinger equation" and had linked a research paper that formed a basis for my question. I was later informed via a comment by the moderator that the paper I had linked was from rxiv.org a.k.a, viXra.org, and that it was not a v...

discussion

ACuriousMind

> @ACuriousMind I'm not sure but maybe you are an experimental physicist [...]

lol

Simon

Thanks @bolbteppa . I've had an introduction to calculus of variations in "Mathematical methods for the physical sciences", but I'll need to study it more.

dmckee

19:27

@Simon Boas is generally a good book, but I recall not liking the treatment of the calculus of variations. f you are happy, then good. If not you might look at some other treatments.

Danu

in Mathematics, 2 mins ago, by Danu

(yes, I'm going around different chat rooms showing this off)

dmckee

@Danu So ... what does it represent? I mean I presume this is a multiply connected space and the non-intersecting paths show us something important, but ... what?

vzn

@mike instantons huh? can you sketch out their connection to solitons?

Mike Miller

Nope.

vzn

@MikeMiller (dont know or dont care to?) are you new to this chat room?

dmckee

19:40

The instructor at a graduate summer school told once told a class I was in "an instanton is a topological soliton".

Mike Miller

vaguely

dmckee

Which was only so helpful.

Mike Miller

@vzn The former.

vzn

@dmckee thx yeah they seem to be connected (closely?) but dont know much about instantons, have heard a tiny )( bit more about solitons over the years :|

dmckee

He used a string of coupled transverse pendula as a working model and talked about putting a twist in the string.

ACuriousMind

19:41

@vzn How often do I have to tell you that instantons are just a special kind of solitons?

vzn

@ACuriousMind as often as you like :P

@MikeMiller ok, welcome (anyway) :)

@dmckee do you have any interest?

ACuriousMind

@dmckee Yeah, that's not a useful explanaton because the notion of "soliton" encompasses various things and is rarely more precisely defined than "some solution with a topological charge (or that's "topologically stable")". And strictly speaking, instantons are not "really" solitons in the sense of being solutions to the equations of motion because we usually look at them in the Euclidean theory, so they're not solutions to the actual Lorentzian equations of motion

dmckee

@vzn IN principle, yes. In practice I have no time for it.

vzn

ahem. (building courage to be shot down again.) solitons seem to be mostly defined with diffeq's. so maybe a instanton is trying to generalize that somehow in a topological sense.

dmckee

I mean, I have this stack of short essays from students in my Gen Ed class to read and all.

Mike Miller

19:48

An instanton is a solution to a certain differential equation.

vzn

@MikeMiller are they defined in 3d + time or is it (further) generalized somehow?

ACuriousMind

@vzn An instanton is a solution to a differential equation, although calling $F={\star} F$ that is a bit of a stretch.

Mike Miller

huh? no it's not

ACuriousMind

Ah, no it's not :)

Mike Miller

it's a nonlinear second order elliptic pde

ACuriousMind

19:49

Forgot about the instanton being the solution for $A$, not $F$ :P

Mike Miller

@vzn In my context they're defined in 4d, not 3d+time, though I guess one could define it in the Lorentzian setting.

vzn

@MikeMiller 3 space dimensions + (what?)

Mike Miller

4 space dimensions.

vzn

@MikeMiller do 3 of the dimensions correspond to "classical" ones?

Mike Miller

That doesn't parse. There are four dimensions. None of them are in any way priveleged.

ACuriousMind

19:51

NB: Mike is a mathematician, his instantons are not supposed to relate to physics per se.

vzn

uh ok

thx for playing

Mike Miller

The appearance of 4 dimensions is related to the equation ACM posted: $F_A = -* F_A$. (We generally use the anti-self-dual equations, not the self-duality equations, due to standard orientation choices on complex manifolds.) $F_A$ is a 2-form, and the Hodge star spits out an $(n-2)$-form. Thus the equation makes sense on a 4-dimensional manifold.

The mathematical relevance is "That's where the equation makes sense". The physical relevance is that 3+1 actually means something.

vzn

so in other words they could exist in 3d spacetime (4d manifold right?) but its a more general mathematical formulation...?

Mike Miller

Yes. One nitpick is that the distinction is between Riemannian (lengths of vectors are positive, this is what I work with) and Lorentzian (lengths of vectors can be negative in the timelike direction). When one says spacetime, 3+1, they mean the latter.

ACuriousMind

Well, the physical notion of an instanton being a classical topologically non-trivial solution to the equations of motion makes sense in all dimension. The specific equation $F={\star} F$ makes sense only in four.

Mike Miller

19:55

What equation do you want to use?

vzn

@MikeMiller re your talk. we have intermittent sessions in this room, currently dormant for awhile. any possible interest? no pressure. :)

ACuriousMind

@MikeMiller The Euler-Lagrange equations for the action of the physical theory under consideration

Mike Miller

@ACuriousMind Oh. Fine.

@vzn I don't do or know anything about physics.

ACuriousMind

Which is $\int F\wedge {\star F}$ for Yang-Mills, and in Euclidean 4D $F={\star}F$ implies that $A$ is a stationary point of that action.

vzn

@MikeMiller lol, that zen attitude. right. mathematicians approach it. anyway we had another one by a mathematical physicist, it was pretty mathematical. but see your point. but on other hand it seems like instantons are definitely physics afaik. am even guessing they were invented by a (math?) physicist....

Mike Miller

19:57

So I'm told.

Danu

21:06

@dmckee It's called the a "handlebody stabilization". If you have a surface with some intersecting lines (the red and blue line initially cross over one another on the genus-2 surface), you can add a handle to resolve the intersection.

Bernard Meurer

    optirun ./bench_cufft
    [INFO] META trials: 10
    [INFO] FFT trials: 10000
    [INFO] Signal Length: 4096
    [INFO] Run benchmark...
    0.111315 sec
    0.111290 sec
    0.111313 sec
    0.111318 sec
    0.111462 sec
    0.111303 sec
    0.111324 sec
    0.111433 sec
    0.111297 sec
    0.111322 sec
    [INFO] Finished!
    [INFO] Average: 0.111338 sec

     ./bench_fftw
    [INFO] META trials: 10
    [INFO] FFT trials: 10000
    [INFO] Signal Length: 4096
    [INFO] Run benchmark...
    0.260865 sec

Some research on compute time for fourier transforms

GPU optimization seems to significantly improve timings

Wonder if I could implement Schonhage-Strassen multiplication with GMP making use of CuFFT

Since CuFFT seems much faster than FFTW, even on my Latop GPU

@heather

heather

21:25

@BernardMeurer, hello, good timing

Bernard Meurer

@heather Regarding our discussion previously about computing Collantz

I've been looking into it

I think if one implements Strassen multiplication with CuFFT it can be made tremendously fast

heather

really? what does the Cu mean?

Strassen algorithm

In linear algebra, the Strassen algorithm, named after Volker Strassen, is an algorithm for matrix multiplication. It is faster than the standard matrix multiplication algorithm and is useful in practice for large matrices, but would be slower than the fastest known algorithms for extremely large matrices. Strassen's algorithm works for any ring, such as plus/multiply, but not all semirings, such as min/plus or boolean algebra, where the naive algorithm still works, and so called combinatorial matrix multiplication. == History == Volker Strassen first published this algorithm in 1969 and proved...

^that is strassen multiplication, right?

Bernard Meurer

@heather Search for Schonhage-Strassen

I can't do umlauts on this keyboard, but the "o" in Schonhage should have an umlaut

heather

Schönhage–Strassen algorithm

The Schönhage–Strassen algorithm is an asymptotically fast multiplication algorithm for large integers. It was developed by Arnold Schönhage and Volker Strassen in 1971. The run-time bit complexity is, in Big O notation, O ( n ⋅ log ⁡ ( n ) ⋅ log ⁡ ( log ⁡ ( n ) ) ) ...

Bernard Meurer

That's the one

heather

21:31

and then FFT = fast fourier transform, Cu...*::googles::*

Bernard Meurer

Cuda

GPU optimization

heather

okay

Bernard Meurer

I still have to think about division though

And I have no clue whether I can use GMP with cuFFT

heather

multiply by 1/2 using the schonhage-strassen?

^for division?

Bernard Meurer

ijircce.com/upload/2014/november/3_big.pdf

I'm not sure if that's faster than a divide-and-conquer proper division algorithm on very large numbers

heather

21:34

oh okay

hmm

Bernard Meurer

I also wonder how the GPU will deal with memory constraints as the numbers grow

heather

is there anything i can do to help you with this project? i know i'm not very technically capable, but i thought i'd ask. =)

Bernard Meurer

@heather Of course there is, if you want to help of course :)

heather

@BernardMeurer, I'd be glad too =D

Bernard Meurer

Unfortunately we will reach a point in a couple of weeks where you will be unable to test the code for yourself, since afaik you lack an nVidia GPU. But I always need help with research and things of the kind

heather

21:45

what do you need right now? =)

Bernard Meurer

So, I'm working on implementing a dumb Collantz calculator right now, nothing fancy

and in the meantime we need to figure out:

1. Both Toom-Cook, Karatsuba, and Schonhage-Strassen are good algorithms for large number multiplications, but for maximum efficiency we might need to switch halfway through. Up until when are each of these methods the best of their class? When do we change to the next one?
2. How to connect GMP with cuFFT? How can we manage memory in the best way possible?
3. Should we only concern ourselves with very large numbers, and use GMP-types all along? Or should we start with small numbers and transition to GMP as the program reaches higher levels

heather

i have two dumb collatz calculators from previous experiments:

github.com/physicsnerd/Collatz-Conjecture/tree/master

I created a repository for it.

Bernard Meurer

Nice, let me port it to C++

heather

sorry, forgot that you were using C++

ACuriousMind

22:03

@BernardMeurer ...why?

Bernard Meurer

@ACuriousMind Because I can

ACuriousMind

Yes, but other people have done so before. What's so interesting about the Collatz conjecture that you want to numerically check it yet again?

vzn

@BernardMeurer !!! congrats on inquiring into that. anyway no need to do fast mult to study it extensively unless its "poc"

Bernard Meurer

@vzn Typo, yeah :)

@ACuriousMind Because I can?!

vzn

oops lol messed up my editing

heather

22:10

hey @vzn, do you want to look at the programs I wrote a while back for the collatz conjecture? I forget if I ever showed them to you.

vzn

@heather sure if you want. prefer to read blogs tho... :| ... do you have a tumblr yet? its one of the easiest

heather

@vzn, no blog, sorry

github.com/physicsnerd/Collatz-Conjecture

^github repo with code

i think my parents would be against my having a blog

okay, scratch that, i'm 99% sure my parents would be against my having a blog.

vzn

@heather wow surprising. and youre all over chat lines. blogs are probably far less edgy yaknow.

heather

@vzn, well, I don't think they realize quite how much I'm in chat....and anyway, they're a little wary of the amount of time I spend on se anyway.

vzn

@heather ah geez thats a shame (any wariness wrt SE). think this is all usually PG rated at least :P ... seriously though as you are aware, theres lots of helpful experts around. tell em not to be afraid :) ... (will comment on the code later)

heather

22:15

@vzn, thanks for looking at the code!

vzn

@heather (sure!) did you ever figure out factoring? :)

heather

@vzn, sadly, not really, I kept trying to make it better and kept running into errors, and then i realized there was a fairly fundamental error, so i kind of left it to come back, yet never really did. i was going to try one of the more established algorithms and try implementing it.

=)

vzn

@heather ok. fyi think you had all the right ideas and were that )( far from finishing. sometimes debugging is the hardest part.

heather

@vzn, okay, I'll give it another go then =)

i needed to take a break, but not this long of one. hopefully i can get rid of the pesky bugs.

vzn

@heather yeah breaks are good. human brains cannot think ("straight") all the time :)

heather

22:18

its kind of ironic because i hate bugs like creepy-crawly things

=)

vzn

@heather fyi theres a possibility that they were 1st named by grace hopper. see her letterman interview if youre curious. which reminds me you (+parents) might enjoy this re Hamilton/ NASA/ moon landing code wired.com/2015/10/margaret-hamilton-nasa-apollo

heather

@vzn, oh, yeah, i've read that story (grace hopper) - found a moth causing an error, computer bug, etc.

wow that's a really cool article, thx for sharing

Andras Deak

22:34

hey

@heather I've been lurking around and I happened to see some python code:P If I may, I suggest that you get acccommodated with python 3, it's what everybody should be doing learning python nowadays:)

besides, at the beginning it just means using print('') and //, but later on it will make a difference

heather

@AndrasDeak, okay =) I'll start doing that more. I've done a little bit in python 3, but not as much as I should be.

Andras Deak

there are a few guys who are adamant in preserving python 2, especially a guy who built his python-teaching enterprise on python 2, but the end-of-life for 2 is 2020...

heather

yeah, i'll start using python 3 =)

Andras Deak

:)

good luck if you do;)

heather

22:50

i think i wrote that quantum computer simulator in python 3

Andras Deak

Good!;)

I didn't read too much of context to tell if this was new code or not

thought I'd drop by and be nosy, as I always am:D

good bye

heather

have a good day =)

Andras Deak

good night;)

Bernard Meurer

@ACuriousMind math.stackexchange.com/questions/2036370/…

Bernard Meurer

23:10

Bah, couldn't find a good implementation of Schonhage-Strassen in TAOCP

ACuriousMind

@BernardMeurer not the slightest idea

Bernard Meurer

#include <iostream>

unsigned long collatz(unsigned long n){
    unsigned long result;
    if (n&1){//Odd
        result=(3*n)+1;
    } else{//Even
        result = n/2;
    }
}

int main() {
    unsigned long x = 27;
    while (true){
        std::cout << x << std::endl;
        if(x == 1) break;
        x = collatz(x);
    }
    return 0;
}

ta-daa

not the best

but it works

Bernard Meurer

23:46

@heather github.com/bemeurer/collatz

Transcript for