The h Bar

So I'm having a bit of confusion. If there's a Lagrangian that has higher-order terms $L(q, q^{(1)}, ... , q^{(n)}, t)$, varying it gives a differential equation $\sum_{i=0}^n (-1)^n \frac{d^n}{dt^n} \frac{\partial L}{\partial q^{(i)}} = 0$. But we have the freedom to add a total time derivative to the Lagrangian and not change things. So wouldn't the derivative terms drop out?

If it makes a difference, I'm trying to figure out what the commenter is trying to get at on this answer and want to make sure I'm not arguing something ridiculous

Actually nevermind. That was a dumb question

Secret

12:30 AM

0

Q: Are contravariant basis vectors and basis 1-forms identical?

The reason I'm asking this is because I am trying to develop a set of notes from my reading of MTW (and Wrede, Menzel, Bergman, etc.). I represent covariant basis vectors with $\mathfrak{e}_{i}$, or if they are orthonormal $\hat{\mathfrak{e}}_{i}$. In cases where the author is explicitly usin...

The definition of infinitesimal is problematic, but the definition of tangent vector is precise, thus tangent vectors provided a way to capture the notion of infinitesimal changes

Emilio Pisanty

1:35 AM

wait, what?

apparently the Teichmüller in Inter-Universal Teichmüller theory was a nazi

Oswald Teichmüller

Paul Julius Oswald Teichmüller (18 June 1913 – 11 September 1943) was a German mathematician who introduced quasiconformal mappings and differential geometric methods into the study of Riemann surfaces. Teichmüller spaces are named after him. Born in Nordhausen, Teichmüller attended the University of Göttingen, where he graduated in 1935 under the supervision of Helmut Hasse. His doctoral dissertation was on operator theory, though this was his only work on functional analysis. His next few papers were algebraic, but he switched his focus to complex analysis after attending lectures given by Rolf...

@ACuriousMind is this widely known in German mathematicksy academia?

Semiclassical

2:12 AM

it is

I think Nevanlinna is questionable as well?

Teichmuller is the go-to example of someone who was undeniably brilliant -and- an enthusiastic Nazi

danielunderwood

I don't know how they compare, but I always think of Lenard. And maybe von Braun

Cows

2:38 AM

Will be dropping moola on food at Costco in a bit

Cows

3:57 AM

Bought some dress shirts for 🏫

user157860

6:16 AM

@JohnRennie, Hi, can you spare a minute sometimes to explain yor comment physics.stackexchange.com/questions/417841/… , and compare your formula to the the one given in the new answer by BetaDecay? Thanks in advance

Cows

@AvnishKabaj hehe, usually tacos and milk lol :P

Secret

6:36 AM

[Random]
A brief history of nothing

It all begins with $\varnothing$

Slereah

6:57 AM

@bolbteppa sound quality is awful

Slereah

7:26 AM

Although you get the feeling of authenticity certainly

Professor breaking his chalk in half while writing

user1414

8:21 AM

I once had a teacher throw a chalkboard eraser at the guy sitting next to me for talking during the lecture.

Emilio Pisanty

@ACuriousMind @rob I just got a red review indicator for a Close queue at 64. Do you guys know if the behaviour changed in the last couple of days?

it's been higher recently without triggering any dots at all

ah

the full queue is at 100

maybe that triggers the dot?

user1414

prepare to have your mind blown

Slereah

9:00 AM

Also the phone rings during the lecture

user351417

@EmilioPisanty In an interesting coincidence, I came across this article :researchgate.net/publication/… while reading about wavemeters... it's the source for the diagram in that mirrors question. I threw together a sloppy answer now; I'll go about fixing it up tomorrow.

user351417

It looks like a wavemeter's probably a bit too expensive for a school lab to procure; I thought I'd need one for the photochromism stuff I'm working on now.

Anonymous

@user1414 Happened in our class too, but the teacher missed the target :P

Abcd

@user1414 lol

I thought it only happens in India

user1414

@Blue he actually nailed the guy right in the face :-/

Anonymous

9:13 AM

Must have been a memorable day for him ;)

user1414

@Abcd aggressiveness is universal

Anonymous

Boring teachers are universal too :P

Anonymous

There are some classes in which sitting quietly and listening to the teacher in itself is a torture XD

user1414

:-D

bolbteppa

@Slereah you say all of that like it's a bad thing :p

user1414

9:18 AM

how was your first day of uni? @Blue

bolbteppa

Sound was ok for me

Slereah

@bolbteppa well it is part of the professor aesthetic

bolbteppa

But it's not good quality yeah

Slereah

Yeah it sounds like it was filmed with an old camcorder

Sound is like SHHHHHHHHHHHHHHHHHHHHHHHH

bolbteppa

There's a few scattered perimeter lectures on it too with way better quality but I'd say those ones are better

Slereah

9:19 AM

plus the indian accent is quit thick

'course

The lagranyan

Anonymous

@user1414 Going well so far :) The digital logic class was interesting

Anonymous

@Slereah lol

user1414

cool :)

@Loong do you really think Croatia would have won without the video reply?

Emilio Pisanty

@Chair it's a nice answer

it'd be nice if you can give a summary of how the device works

(I can mostly see it, but it'd be nice to have it on the record)

I'm tickled by your surprise that the paper is "free", though

Loong

@user1414 I didn't say that.

Emilio Pisanty

9:27 AM

it isn't - it's just that one of the authors has set up a homepage on an academic social network and posted the pdf there

the "real" home of the paper is at doi.org/10.1119/1.19335

user351417

I have no idea... I didn't need to scihub it

user351417

I'm not too familiar with the whole concept though, to be honest.

user1414

@Loong I know, but it was the turning point...

user351417

I just visited that researchgate link and clicked download because I know researchgate's a reliable place...

user351417

And it worked

Emilio Pisanty

9:29 AM

> researchgate's a reliable place

well....

up to a point

Let me put it this way

Slereah

Researchgate is basically a blog

It's as reliable as the person

Emilio Pisanty

if a paper has a free eprint on the arXiv, I add a link to the arXiv

user351417

I guess a lot of the resources I've been needing lately have been on sciencedirect/elsevier... and those are always paid access/institute axis.

user351417

@EmilioPisanty Sure, that'd be great

Emilio Pisanty

if a paper has a free eprint on a university repository like e.g. spiral.imperial.ac.uk, then I add a link to that version

user351417

9:30 AM

I wasn't really looking for that article... it's a strange coincidence. I wanted wavemeters, and then I sudedenly saw a familiar-looking image

Emilio Pisanty

if a paper's only free version is on researchgate, then I might use it to get access, but it's not suitable for inclusion on a reference in a publication

Anonymous

@EmilioPisanty The pre-print can sometimes have a lot of errors compared to the actually published version though :P Some authors never care to update the arXiv version after publication

user351417

Eh whatever... I have the name of the article, along with the authors' namess, so it can't be too hard to find.

Emilio Pisanty

@Slereah to be fair, researchgate is less terrible than academia.edu, whose T&Cs are absolutely catastrophic

I'm not touching that with a ten-meter pole

Slereah

vixra is best, 'course

Emilio Pisanty

9:33 AM

@Chair well, there comes a point where being able to find it isn't enough

Slereah

Also really

How often do you end up on academia.edu

I don't think I've ever googled a thing and ended up on academia

At least for physics

Sometimes I do for humanities stuff

Emilio Pisanty

you might also care about e.g. writing a citation that ensures that your readers will always be able to find the resource you're pointing to

@Slereah when academia.edu sends me spammy emails

"your paper has been cited X times!"

"pay us $20 and we'll tell you where!"

Slereah

has it

Emilio Pisanty

absolutely despicable stuff

user351417

Also, the answers to the same question on quora are incredibly terrible

user351417

9:34 AM

quora.com/…

user351417

Quora actually has several measures which almost encourage people to write about things they don't know.

5

Emilio Pisanty

@Slereah On the other hand, researchgate was the first to pick up this citation researchgate.net/publication/…

which, frankly, I'm over the Moon about

I thought that paper was dead and buried and nobody was ever going to read it

Slereah

Who's that dork

Emilio Pisanty

@Slereah that's a younger me

used to be my PSE profile pic, too

Slereah

neeeeeeeeeeeerd

Emilio Pisanty

9:37 AM

meh

but anyways @Slereah is arxiv.org/abs/1807.04052 your kind of stuff?

'cause I've got no idea what's really going on there =P

Slereah

I'm not great at SUSY

Emilio Pisanty

you're probably more in tune than me =P

Slereah

Sure but so's a monkey

Emilio Pisanty

it'd be nice to have a sense of what they're actually doing and how the truncated harmonic oscillator fits into it

@Slereah well, I mean, I couldn't even have told you that it was a SUSY paper

Slereah

You can tell from the $\mathcal N$

They love the big N

Emilio Pisanty

9:41 AM

ah

Slereah

(it's the number of SUSY generators)

Emilio Pisanty

also probably the fact that they call a 2012 paper "recent"

Slereah

Well yes

it's rare for a paper you reference to be less than a decade old

Emilio Pisanty

heh

yeah, I guess formal quantum optics doesn't move that fast anymore

though partly that is because the central core of the subject is pretty mature

Abhinav

10:33 AM

If I have a two spring i n series with one of them connected to ceiling and other attached to a mass then why is the total force experienced by the mass equal to their combined spring force instead of just the spring force of the spring attached to the mass

Can somebody help me on this

Looks close to newtwon's third law

user929304

11:01 AM

@ACuriousMind Hi :), let me know if you're around in the chat today.

Anonymous

@user929304 If you have a question for him why not write it down now instead of waiting for him to come online? :P Saves both of your time(s) in that way. Chat doesn't necessarily have to be in real-time!

Anonymous

@Abhinav The force gets exerted by a single spring only (the one directly connected to the body). However there's another spring pulling it from above. Draw the FBDs.

user2236

Wow! @Loong it looks like no World Cup champion has had less ball possession than France?

user2236

11:26 AM

I think the ball possession statistic should be divided into having the ball on your half of the field compared to having it on the opponent's half.

user929304

@Blue indeed, good idea :)

@ACuriousMind I'm looking for a post, I admit I may be mistaken, but I think it was answered by you, on the topic of how to correctly explain what is meant by spacetime curvature, as opposed to the common misconceptions where one gets the impression, the manifold is curving into "something else". Any chance your remember which post it was?

Abhinav

@Blue yea...the upper spring is also pulling the lower string with spring force.And since the springs are massless they are equal in magnitude. Thanks for the help

user351417

1:34 PM

When do you formally study hamiltonian mechanics? Maybe second year of undergrad or something?

user351417

I've been working on it lately, but I'm not really getting anywhere... Does anybody know of a good resource for that?

Anonymous

@Chair Probably depends at what level you want to learn it. There are some basic undergrad level classes, yes, but also a lot of graduate level classes possible. If you're a beginner maybe get a basic idea first :). It's something I've been tearing my hairs over for a while (the grad level course) :P

Anonymous

@Chair For a basic introduction, see Goldstein chapter 1, 2 and 3

Anonymous

@Chair Also Balakrishnan has some good undergrad lectures. See his CM playlist

user351417

1:55 PM

@Blue It's not a common undergrad thing? Then what all's there in undergrad? Presumably an introduction to calc-based physics, because we do algebraic in school... that'll have separate electromagnetism and mechanics components presumably. What else?

user351417

Maybe lagrangians?

user351417

I'm quite sure that would come before more specialized modern physics (20th century) stuff.

user351417

@EmilioPisanty Rollback, please.

Anonymous

@Chair You have some of the topics covered in undergrad courses like - action principle, Hamilton's principles, action angle variables, Poisson brackets, solving some mechanics problems using those two formalisms etc. But that's not all of it. There's a lot more to CM which isn't covered in undergrad courses

user351417

@Blue Sounds fun. Thanks!

Anonymous

2:11 PM

@Chair "Sounds fun"...wait till you reach the principle of least action. It seemed like complete nonsense to me at the beginning (combined with the fact that the unknown lagrangian basically appears out of thin air)...and sort of still does :P There are some nice PSE answers on it though by joshphysics and QMech btw

user351417

@Blue From what I've read about action as a quantity in general, it seems pretty random. But apparently it's got some important applications in quantum mechanics so whatever... I've been working with a nonlinear dynamics prof in a nearby university over the summer, and one of the things was to make a simulation+phase portrait generator for a double pendulum. He told me that the principle of least action would come in handy, but eventually I gave up on that and used the lagrangian.

Anonymous

2:27 PM

@Chair Interesting :). BTW where/how did you use the lagrangian? (And what's the professor's name, if you don't mind saying?)

Semiclassical

quantizing the double pendulum seems like it might make for an interesting test-case of constraint dynamics

Anonymous

BTW I should mention that Goldstein's chapter 2 has a lot of mistakes in the 3rd edition. Get a hold of the more recent editions if possible. Here's the errata page: astro.physics.sc.edu/Goldstein

Semiclassical

after some digging, it doesn't seem like the proper quantization of the double pendulum is actually known

Fawad

2:46 PM

Few important definitions. Energy :ability to do work. Work: transferee energy

Slereah

I am out of energy

I am unable to do work

Semiclassical

sounds legit

Fawad

Transfer of*

Emilio Pisanty

3:06 PM

@Chair oh, definitely not.

That's a liiiiiiittle too visibly associated with my IRL name now.

There's a number of considerations that apply when you do the right thing and participate with your real name instead of a pseudonym

user351417

@EmilioPisanty I'm going to put up a question on the main meta once I muster my guts... "Enforce rollback of profile pictures".

user351417

@EmilioPisanty I guess mine isn't a real pseudonym... my full name's on my about-me stuff, and mods can see my previous profile name since I changed it recently.

rob

@EmilioPisanty I think you're right that the dot is triggered by the size of the entire review queue, rather than just the items that you personally can review.

John Rennie

@Chair mods are obliged to keep your personal info confidential.

rob

A database person would probably explain that computing the size of your queue is too expensive an operation to perform in the header of every single page you view on the site.

user351417

3:11 PM

@JohnRennie Yeah, but I guess it's not confidential if I have it on my about-me for anyone interested/jobless enough.

user351417

I think I just wanted to evade google searches for my name, and apparently it works

Semiclassical

it's a sort of security-through-obscurity

which is always a bit shaky: if someone wants to learn something obscure, they usually can do so

John Rennie

It has never occurred to me to keep my name secret. Aside from the unfortunate incident of the goat and the watermelon I have nothing I wish hidden from the public gaze.

rob

@JohnRennie I didn't realize you were Welsh?

Oh, no, I'm thinking of the curious incident of the dog in the nighttime‌.

Semiclassical

I'm not exactly unknown---I've said enough about myself over my time here that someone could work it out without much trouble---but I just prefer the nickname.

John Rennie

3:17 PM

@rob in fact my mum's family is Welsh, though they left Wales (and the sheep) several generations ago.

@rob isn't there a Sherlock Holmes story called something like that?

rob

@JohnRennie That is a very Conan-Doyle title, isn't it? Perhaps that was intentional.

Semiclassical

"What was curious?" "That the dog didn't bark."

user351417

@Blue I used the Euler-Lagrange stuff... Took both pendulums' angles as my generalized coordinates. It was simplified because I got to consider 0 dissipation forces. Then you just do the $\frac{d}{dt}\big(\frac{\partial L}{\partial \dot{q}}\big)=\frac{\partial L}{\partial q}$ for each of those. You can simplify for second time derivative of each angle, then I approximated with $0.00001=dt$. I was hoping that I could do the runge kutta 4 or something but it got a bit too confusing.

rob

The Adventure of Silver Blaze

"The Adventure of Silver Blaze", one of the 56 Sherlock Holmes short stories written by British author Sir Arthur Conan Doyle, is one of 12 in the cycle collected as The Memoirs of Sherlock Holmes. Doyle ranked "Silver Blaze" 13th in a list of his 19 favourite Sherlock Holmes stories.One of the most popular Sherlock Holmes short stories, "Silver Blaze" focuses on the disappearance of the eponymous race horse (a famous winner) on the eve of an important race and on the apparent murder of its trainer. The tale is distinguished by its atmospheric Dartmoor setting and late-Victorian sporting milieu...

Semiclassical

(it's a nice literary example of where the absence of a certain event serves as evidence for a hypothesis)

rob

3:22 PM

@JohnRennie I know essentially nothing about Wales, but in the same year I happened to read two novels set in Swindon

The one I linked above, and one of Fforde's Thursday Next novels.

John Rennie

Swindon isn't in Wales ...

rob

@JohnRennie Next you're going to tell me that it isn't a gleaming megalopolis, either?

And that the library system doesn't employ armed enforcers who hunt down overdue books?

This is what I get for learning things from fiction.

John Rennie

@rob Swindon is near where I grew up, in the West Country. That's how I know where it is :-)

Anonymous

@Chair Cool, I see :)

rob

@JohnRennie That's my trouble: it's nowhere near where I grew up :-)

Abcd

3:38 PM

Can you raise the pressure of a gas without heating it, anyone?

Semiclassical

do you mean: without any heat being transferred, or without the temperature of the gas changing?

both are possible, but they're not the same process.

Abcd

(At constant volume)

Sorry forgot to mention constant volume

Semiclassical

not for an ideal gas, you can't.

Abcd

@Semiclassical nothing as such has been discovered till now?

Anonymous

It's less about discovery and more about theoretical impossibility :P

Semiclassical

3:41 PM

^

Abcd

I mean temperature should change as a result of changing pressure not the other way round

Anonymous

Why not?

Abcd

Pressure shouldn't change because of temperature change

Anonymous

$PV=nRT$

Abcd

@Blue I am asking a question

Semiclassical

3:42 PM

Don't tell ideal gases what to do.

3

Abcd

You are not getting my question

Anonymous

If $n,V$ is constant, if you increase $P$, $T$ increases

Abcd

I am asking how will you increase P to do that

Semiclassical

If you hold the volume, temperature, and the moles of an ideal gas fixed, there is no way to change the pressure.

Abcd

Do you know any such mechanism?

@Semiclassical Exactly

1 min ago, by Blue

If $n,V$ is constant, if you increase $P$, $T$ increases

But its not a thermodynamic impossibility

Semiclassical

3:44 PM

Okay? That means that you can't raise the pressure of a gas at constant volume without changing the temperature.

Abcd

17 secs ago, by Abcd

But its not a thermodynamic impossibility

Anonymous

@Abcd It's a bi-directional property

Abcd

4 mins ago, by Abcd

@Semiclassical nothing as such has been discovered till now?

Which means we havent discovered anything to do this till now

1 min ago, by Semiclassical

Okay? That means that you can't raise the pressure of a gas at constant volume without changing the temperature.

Semiclassical

To the extent that PV=nRT is a defining property of an ideal gas, it is impossible to change the pressure of such a gas without changing at least one of n,T, or V.

There's nothing to 'discover' there. It's simply a feature of what it means to be an ideal gas.

Anonymous

@Abcd Yes, physically increasing $T$ would be the cause and increase of $P$ would be the effect

Abcd

3:47 PM

@Blue Ya. so I want physically increasing P to be the cause and Increase in temperature to be the effect

Semiclassical

That's not hard, really. Consider a balloon rising through the atmosphere, under the assumption that the temperature surrounding it remains the same.

John Rennie

@Semiclassical then the volume isn't constant

Semiclassical

point. hmm

that's not the best example

Anonymous

@Abcd I don't think that's physically possible

Semiclassical

I think I got myself backwards there.

John Rennie

3:49 PM

@abcd if you have a sealed volume of gas then the only thing you can transfer in or out of it is heat i.e. change its temperature.

Abcd

@Blue Why ??

Semiclassical

By forbidding a change in volume, you're forbidding work to be done.

John Rennie

I can't think of any way to make a change in pressure the cause and a change in temperature the result.

Abcd

@JohnRennie we can even excite its molecules through EM waves

John Rennie

@Abcd that's just another way of adding heat

Semiclassical

3:50 PM

Ergo, any change of internal energy must be due to heat.

Abcd

@JohnRennie I don't know whats wrong with my laptop , suddenly too many 3333333333s get typed themeselves when I am not even touching it. Why is it so?

Slereah

Evening gents

John Rennie

@Abcd faulty 3 key?

Try tapping the 3 key a few times.

Abcd

3

Semiclassical

Actually, there is one system which comes to mind: a metal undergoing a superconducting transition

Abcd

3:52 PM

it works fine , the 3 key

Semiclassical

since then you can have work done by an external magnetic field

Anonymous

@Semiclassical That's not an ideal gas?

Semiclassical

sniiiped :)

Anonymous

@Abcd If you have a closed container, the only way to increase pressure is to forcefully compress it or add more moles of the gas. How else do you increase pressure directly without heating? Even theoretically, I can't think of a way

Anonymous

Let's go back to the basics

Anonymous

3:54 PM

@Abcd What is pressure of a gas?

Abcd

@Blue even I cant think of a way thats why I was asking

Semiclassical

I think the question is really: If the internal energy increases, is it due to work or to heat?

We're forbidding mechanical work here.

But are there other ways to do work on a gas?

For an ideal gas, the answer is simply no.

Abcd

@Blue F/A

Anonymous

@Abcd The key point really is: temperature and kinetic energy are more fundamental than pressure for an ideal gas. (Hint: kinetic theory of ideal gases)

Anonymous

Pressure is always necessarily a side effect of the molecules hitting the walls (more the KE, more forceful the hits)

Anonymous

3:57 PM

And the KE is a result of T

enumaris

extracting data dun dun dunnnn

Semiclassical

The main point is that an ideal gas is a very simple model. It doesn't have a lot of variables to play with

As such, the only ways to have transfer of energy are by particles entering/leaving the system, by mechanical work, or by heat transfer.

Anonymous

@Abcd I'm precisely trying to explain why physically trying to increase P and thereby increasing T, as a result, doesn't make any theoretical sense. Not really a matter of "hasn't been discovered till date"

Anonymous

P is an averaged out property of the (rate of) gas molecules hitting the walls

Semiclassical

Now, you can consider systems with other mechanisms of energy transfer. But such a system is by definition not an ideal gas.

(you can find some discussion online about a classical ideal paramagnetic gas. in that case you can do work on the gas without changing the volume. but that's not what one typically means by an 'ideal gas')

Anonymous

4:06 PM

@Abcd I just cooked up an analogy: When you're hitting a football, the cause of it flying away is the force you exerted on it, and the distance if flew is the effect. Now does trying to increase the distance of flight of the football in order to increase the force the force of your kick, make sense? :P

Anonymous

Anyhow, I gotta leave now

Semiclassical

@bolbteppa given our discussions re: Bohmian mech, you might find this paper interesting: mathreview.uwaterloo.ca/archive/voli/1/oltean.pdf

it purports to give an account of path integrals from the Bohmian pov, including a derivation of the Feynman path integral from such

...i dunno how I feel about it yet

(the expression they give for the 'de Broglie-Bohm path integral' only makes sense if (q',t') and (q,t) fall along the same trajectory. but the feynman path integral is supposed to be valid for any pair of spacetime points...)

Abcd

@Blue yeah

Semiclassical

4:24 PM

(I think I know the fix for it, but I’m not convinced I like it...)

Anonymous

@Abcd What?

Emilio Pisanty

5:20 PM

any native speakers around?

I need a more formal version of "reckon", as in "we reckon that our work is innovative"

something that's safe to put in a cover letter to the editor

I'm reluctant to replace "reckon" with "feel"

@Semiclassical @JohnRennie

Semiclassical

"we believe that our work is innovative" is more formal but also seems a bit wishy-washy in the vein of 'feel'

though in any case you're giving an opinion

another one would be "we're confident that our work is innovative"

Emilio Pisanty

@Semiclassical oooh, nice

@Semiclassical well, that's a given, no?

it's why it's in a cover letter and not in the manuscript

Semiclassical

sure, i'm just saying that there's no way around that

Emilio Pisanty

oh, sure

Semiclassical

i imagine you're also giving reasons why you have that opinion in said letter

Emilio Pisanty

5:25 PM

it's not the opinionness of "feel", it's the washiness

@Semiclassical I kinda wish I could say "nope, this is the entirety of our cover letter" =P

Semiclassical

true enough

Emilio Pisanty

but we do

Semiclassical

lol

so you can easily frame this as "we're confident in our work being innovative, and here's why"

for better or worse, there's little room for ambivalence in that

Emilio Pisanty

@Semiclassical well, it's the concluding paragraph

> Given the above, we are confident that our work is innovative ...

enumaris

Or you could just drop the believe part entirely

and just go with "Our work is innovative"

Cows

5:54 PM

Might have to actually dress up professionally and nice tomorrow for a face to face with some big fat giant ceo. He wants us to meet in his office, I want him to meet me at Starbucks , I told his boy by text that it's Starbucks or nothing lol

come to think of it might show up in my shorts

jk , will be professional but

he has to come to Starbucks if he wants my code

enumaris

lol

Cows

lolz

enumaris

extracting data is so boring

sit here and watch a number tick up

lol

danielunderwood

My evening plans are to clean up and normalize a bit of data...I'm so excited

I've spent more time on handling the data than I have on actual analysis

enumaris

lol

I thought you did software dev

danielunderwood

6:11 PM

Well this is a Kaggle competition I'm doing in my spare time. My work also wants to do data stuff, but at the moment it's just creating software/devices to collect the initial data

enumaris

I see

danielunderwood

It's actually kind of thrilling to work with data again, especially compared to endlessly writing api code

enumaris

kaggle kernels are so shite

limited to 1 hour of computation time...easy to lose your work...easy to have computation interrupted...yuck

danielunderwood

lol I don't really look at the kernels much. I figure I have a better chance of figuring it out on my own than from a good number of them

Ahh that. I don't really use them either

enumaris

some competitions force you to use their kernels though

danielunderwood

6:14 PM

I have a home server that I was running jupyter on, but I've preferred just using pycharm and writing code the normal way

Ahh I haven't run into any of those. Although most competitions I just join then never do much with. There is one from CERN on there now that I really want to try to do, but the data is quite a beast

enumaris

6:34 PM

I write python on Sublime text and then run it off command line...

I tried Visual Studio

but it was super slow for some reason

danielunderwood

VS likes to be slow. I usually use pycharm/other jetbrains ides for most things. I also use vscode some, particularly for stuff with platform.io

enumaris

suuuuper slow

like took my 3 minute run time and spent an hour without terminating lol

danielunderwood

Ahh I never had an issue with runtime slowness, but vs itself being slow. Like taking 5 minutes to load a project. Though I never tried using it with python

enumaris

I see

vzn

@danielunderwood which kaggle competition?

danielunderwood

6:42 PM

kaggle.com/c/trackml-particle-identification

vzn

wonder if kaggle is going to chg somehow now that google owns it, so far seems to be no chgs techcrunch.com/2017/03/07/…

danielunderwood

Their tos probably changed in some sneaky way. And I imagine that they'll eventually try to push the usage of GCP for analysis. I think google had a competition on there at one point when they did that

Don't know if that was before or after buying it though

vzn

@danielunderwood thats very cool have heard of a CERN/ LHC kaggle prj awhile back & blogged on it but didnt know if they were continuing those, really great to see it continue & have sizeable prize money.

enumaris

6:54 PM

welp, that tableau dashboard is finished

time to go back to working on my deep speller mmmhmm

danielunderwood

Can I ask what kind of structure that has? My initial thought would be a word input, some type of nn, then an output covering a corpus...but that seems like quite a large output space

enumaris

gotta eat lunch, will explain later

danielunderwood

Although I suppose it may be a smaller output space than some image stuff

My lunch includes beans. Thought you of all people may like to know

DanielSank

Is there a way to see that the Aharanov-Bohm effect isn't weird by arriving at the infinite solenoid from a non-infinite one and then taking a limit?

enumaris

7:35 PM

@danielunderwood it's a character level Sequence to sequence model. It inputs a phrase, encodes that phrase character by character, and then decodes + spell corrects that phrase character by character. So the input and output spaces are pretty small actually.

@danielunderwood also, what kind of beans?

Abcd

I am unable to understand how a current carrying loop is a dipole, what's the best way to understand this concept?

enumaris

@DanielSank what do you mean by "isn't weird"?. a finite solenoid would have to include edge effects which would be extremely hard to calculate.

danielunderwood

Huh that sounds neat. Looks like that's dependent on RNN, which I haven't looked at yet but was planning to in the next couple of days.

And pintos. I was going to just cook beans, but I had some unused veggies and kind of ended up with a vegetable bean soup type thing

enumaris

hmmm

danielunderwood

And for those of you that use pdf textbooks, how do you manage them? Just stick them in mendeley with papers?

enumaris

7:41 PM

always good

It's a deep LSTM network :D

danielunderwood

I've seen LSTM a lot, but haven't actually made it to what it is yet lol

"least squares tangent manifold"

enumaris

It's a RNN cell that has memory so as to alleviate the vanishing/exploding gradients problem

Long-short-term memory

danielunderwood

Hey it looks like an RNN is basically what I thought it would be. A node feeding backwards into the net it looks like

enumaris

basically yeah

it takes the output and feeds it back into its inputs

hence recurrent

does require a bit of care when doing back prop on that though

danielunderwood

Yeah I figured that's what recurrent would mean. I have no idea how you'd train that off the top of my head though. Looks like tf has some rnn contrib stuff, so I guess just let that do all the hard work

enumaris

7:56 PM

well TF does the backprop for you yeah

I'm talking about at a theoretical level the backprop is a bit complicated - it's called back propagation through time

Transcript for