The h Bar

Sir Cumference

10:00 PM

@EmilioPisanty "Record"?

Emilio Pisanty

@SirCumference As in, it happened.

AccidentalFourierTransform

yeah, who flagged that?!

Sir Cumference

@EmilioPisanty You mean that one time I argued with you about PSE's logo?

Emilio Pisanty

@SirCumference That's the time.

Sir Cumference

Quite a short record, I'd say

Emilio Pisanty

10:01 PM

Given that the same pattern just recurred, maybe you can entertain the possibility that the problem is with parts of your attitude and not with other people?

Just a suggestion.

Sir Cumference

@EmilioPisanty I just don't really get the rationale for questions like "What real-world impact would your proposed change have?"

AccidentalFourierTransform

"this is not a useful conversation."

WWRS

Emilio Pisanty

@SirCumference I don't understand why you think those comments are anything other than people trying to help you improve your question.

but

AccidentalFourierTransform

what would rob say?

3

Emilio Pisanty

I agree with AFT

Sir Cumference

10:03 PM

Same

Emilio Pisanty

@AccidentalFourierTransform who's Rob-with-a-capital-R?

LegionMammal978

Just curious, what's the difference between 1 mol^2 (of dimension N^2) and N_A(1 mol^2) (of dimension N)? Is this the correct place to be asking?

Sir Cumference

@EmilioPisanty I'll gladly admit that your comments were less aggressive than the ones I received on the latter post

AccidentalFourierTransform

ninja-fix

Emilio Pisanty

@AccidentalFourierTransform good

now I can star it

AccidentalFourierTransform

10:04 PM

lol

DanielSank

Someone teach me a physics.

Anything.

Emilio Pisanty

@DanielSank did you see the IBM "quantum volume" thing?

DanielSank

Oh also one more thing: if anyone here is interested in giving a technical talk/seminar at the quantum computing lab in Santa Barbara, email me.

5

Sir Cumference

@DanielSank Light moves fast

DanielSank

We just started doing that.

Sir Cumference

10:05 PM

Congrats, you passed my physics class

doublefelix

@CRDrost Hi there :) You recommended talking here, I think it's a good idea

DanielSank

@EmilioPisanty Briefly.

Emilio Pisanty

arstechnica.com/science/2017/05/…

DanielSank

@doublefelix Physics, math and CS?

o_O

Emilio Pisanty

@DanielSank how worthwhile is this volume notion thing?

DanielSank

10:06 PM

@EmilioPisanty I haven't read it carefully.

Emilio Pisanty

@DanielSank fair enough

I'll quiz you later

no, serious question, though

can anyone make heads or tails of this?

DanielSank

Glancing over it, this is very much not a new idea at all.

doublefelix

oh, whoa that was a while ago. I ended up doing just physics and math

Emilio Pisanty

pnas.org/content/114/19/4920.short

Abstract
Intuition from our everyday lives gives rise to the belief that information exchanged between remote parties is carried by physical particles. Surprisingly, in a recent theoretical study [Salih H, Li ZH, Al-Amri M, Zubairy MS (2013) Phys Rev Lett 110:170502], quantum mechanics was found to allow for communication, even without the actual transmission of physical particles. From the viewpoint of communication, this mystery stems from a (nonintuitive) fundamental concept in quantum mechanics—wave-particle duality. All particles can be described fully by wave functions. To determine w…

DanielSank

@doublefelix Update that profile, sir.

Emilio Pisanty

10:07 PM

Significance

Recent theoretical studies have shown that quantum mechanics allows counterfactual communication, even without actual transmission of physical particles, which raised a heated debate on its interpretation. Although several papers have been published on the theoretical aspects of the subject, a faithful experimental demonstration is missing. Here, by using the quantum Zeno effect and a single-photon source, direct communication without carrier particle transmission is implemented successfully. We experimentally demonstrate the feasibility of direct counterfactual communication …

doublefelix

i am ;)

Sir Cumference

Sigh

Emilio Pisanty

> "direct communication without carrier particle transmission is implemented successfully"

does this ring huge red bells for other people as well?

I mean

ring huge flags

DanielSank

It's just entanglement stuff.

Emilio Pisanty

@DanielSank but you start off with shared entanglement, right?

DanielSank

10:09 PM

Saying there's no particle exchange is kinda... I dunno.

@EmilioPisanty Probably. I'd have to read it more carefully.

I doubt they've broken physics.

Emilio Pisanty

@DanielSank yeah, me too

I tried reading it but it got boring =|

AccidentalFourierTransform

Emilio Pisanty

or rather, I tried reading the PNAS, couldn't get access, looked for it on shady russian servers, didn't find it, switched to the Salih, Sih, Zubairy paper they reference in the abstract, didn't get any of it

nbro

What's the derivative (with respect to time) of the linear acceleration?

Emilio Pisanty

@DanielSank if you do get round to reading it I'd appreciate a ping

@nbro the jerk

AccidentalFourierTransform

10:14 PM

be nice.

nbro

(Just asking this in order to make this seem a serious chat about physics)

Emilio Pisanty

after which come snap crackle and pop

AccidentalFourierTransform

gotta leave

bye chaps

Emilio Pisanty

Jerk (physics)

In physics, jerk, also known as jolt, surge, or lurch, is the rate of change of acceleration; that is, the derivative of acceleration with respect to time, and as such the second derivative of velocity, or the third derivative of position. Jerk is a vector, and there is no generally used term to describe its scalar magnitude (more precisely, its norm, e.g. "speed" as the norm of the velocity vector). According to the result of dimensional analysis of jerk, [length/time3], the SI units are m/s3 (or m·s−3); jerk can also be expressed in standard gravity per second (g/s). == Expressions == Jerk can...

nbro

to raise the levels in the same way that the glaciers do when they melt

nah mean?!

and I drop more metaphors than the bibble

DanielSank

10:18 PM

@EmilioPisanty

> It gets even worse when you consider that, unlike classical computers, you need a certain number of qubits to even carry out a calculation of a certain computational size

That's... wrong.

nbro

yeah

DanielSank

If I give you a classical computer with two bits, you can't use it to add 8-bit numbers.

nbro

@EmilioPisanty Thanks, anyway!

:D

Emilio Pisanty

@DanielSank yeah, well, when's the last time you saw a two-bit classical computer?

... that sounded unexpectedly demeaning w.r.t said perfectly-respectable classical computer possessing two bits

Mithrandir24601

@DanielSank If you've got a memory as well, then it should be possible

(as in, a memory big enough to store the numbers, then some more)

DanielSank

10:21 PM

@Mithrandir24601 Mehhhhhhhh ok but in a quantum computer you're doing logic directly on the "memory".

There's no separation between memory and logic units in a quantum computer (yet).

This might change later, but for now we're going with a system where all the qubits are just there in a grid and the logic operations happen directly on them.

There's no separation between qubits storing data and qubits processing data.

heather

@DanielSank ooh, have you heard of CRNs?

DanielSank

no

heather

k.

so CRN stands for chemical reaction network

the basic idea is that you have a bunch of chemicals (the "states" of the system)

and the reactions between them are the "commands" of the programming language

i.e., you can use it to solve problems.

molecular computing =)

you can also store information using chemicals

for example, if you use DNA, you can store approximately a petabyte of data in a vial the size of your pinkie finger

DanielSank

That sounds extremely impractical. Has anyone done it?

Oh well, yes, living things have done this amazingly well.

That's a good point.

heather

people have stored info on DNA and computed on chemicals, yes

DanielSank

10:24 PM

It's not programmable though... yet.

Is it?

heather

for example, someone solved a version of the traveling salesman problem using molecular computing

Emilio Pisanty

@heather that sounds remarkably like a biological brain

heather

@DanielSank - well, I guess I don't quite know.

Emilio Pisanty

(hence programmable)

heather

I think yes? @DanielSank

Mithrandir24601

10:25 PM

@DanielSank Weren't you one of the authors on a paper a few years ago entitled "Implementing the Quantum von Neumann Architecture with Superconducting Circuits" where you demonstrated "a quantum central processing unit that exchanges data with a quantum random-access memory integrated on a chip"?

heather

but anyway, I was reading this paper recently

DanielSank

@EmilioPisanty Ehhhhh, programmable? Kiiiiind of.

heather

well, currently.

DanielSank

@Mithrandir24601 Link?

heather

called reachability problems for continuous chemical reaction networks

DanielSank

10:25 PM

Oh Matteo's paper?

heather

arxiv.org/pdf/1508.04125.pdf

Emilio Pisanty

@DanielSank you can tell people "follow this algorithm" and have them follow a pre-set run of operations

heather

and the paper has come up with an algorithm that runs in polynomial time

Emilio Pisanty

(if you're nice enough and you pay them a salary for it, of course)

DanielSank

@EmilioPisanty Programmable to me means "No solve this problem".

heather

10:26 PM

that checks if there is a path between a state c and a state d, and if there is one, produces one possible path

DanielSank

You can train a biological brain, but I wouldn't say you can "program" it.

heather

(the path, of course, being in terms of reactions)

Emilio Pisanty

I mean, that's what "computer" meant before the late 1950s

heather

so i'm reading and trying to understand that.

Emilio Pisanty

@DanielSank you can load programs on it

DanielSank

10:27 PM

@EmilioPisanty That's a cute point.

heather

@DanielSank, well, couldn't you cause certain reactions to happen?

Emilio Pisanty

and have it execute them

DanielSank

@EmilioPisanty Not... really.

Emilio Pisanty

@DanielSank no, seriously

DanielSank

@EmilioPisanty ehhhhhhhhhh

Emilio Pisanty

10:27 PM

you haven't seen Hidden Figures?

DanielSank

I'm not comfortable calling that "programmable".

heather

Hidden Figures is a great movie

DanielSank

@EmilioPisanty Yes I have, and I know that the word predate's electronic computers.

Emilio Pisanty

Human computer

The term "computer", in use from the early 17th century (the first known written reference dates from 1613), meant "one who computes": a person performing mathematical calculations, before electronic computers became commercially available. "The human computer is supposed to be following fixed rules; he has no authority to deviate from them in any detail." Teams of people were frequently used to undertake long and often tedious calculations; the work was divided so that this could be done in parallel. Since the end of the 20th century, the term "human computer" has also been applied to individuals...

DanielSank

Yes yes I know.

I object to the term "programmable".

Mithrandir24601

10:28 PM

@DanielSank science.sciencemag.org/content/334/6052/61 I get what you're saying that 'logic operations happen on the memory' (to paraphrase), but it's not like you're not able to have any separate memory, surely?

Emilio Pisanty

@DanielSank how is what a human computer does any different to loading a program on a general-purpose electronic computer?

heather

@DanielSank to make this overly meta, wouldn't you call someone teaching me something "programming" me?

Bernardo Meurer

Humans computers are buggy af

heather

(sort of)

Bernardo Meurer

Worse than those dreaded PDP-8's

DanielSank

10:28 PM

@Mithrandir24601 Yeah that was Matteo's paper. I do not necessarily agree with all decisions made with respect to title or tone of the discussion.

Mithrandir24601

(it was 2 years ago since I read this, so I've actually forgotten most of it and might be completely misremembering things)

heather

@BernardoMeurer especially this one =P

DanielSank

I stand behind the science, of course.

Emilio Pisanty

@DanielSank "do not necessarily agree" ftw?

heather

there's a professor at Iowa State University that has a lab that researches molecular programming

which is kinda cool.

DanielSank

10:29 PM

@EmilioPisanty On an electronic computer, I can write a program to do a specific task, and it gets done. Humans can't do this in the same way. There's no input format I can give a human and be guaranteed that they do the required task immediately and without error.

You can train humans, like you can train neural nets. Not an accident.

Mithrandir24601

@DanielSank Fair enough, I suppose. It was interesting to read about though

heather

@DanielSank you can't exactly guarantee it always works for a computer either

DanielSank

oye veh

You guys... ;-)

heather

hey, you're making these arguments, i'll argue right back =P

Emilio Pisanty

@DanielSank there's no input format you can give an electronic computer and be guaranteed that they do the required task immediately and without error.

Bernardo Meurer

10:30 PM

@heather In any reasonable architecture you can, for all practical meanings of the word, guarantee it will work

Emilio Pisanty

there's always some give and take on standards

nbro

The following question was marked as duplicate, but that isn't a real duplicated question, so I would foster a moderator to remove that (e.g. @DavidZ)

7

Q: Why don't we consider jerk in physics classes?

When i got more into physics, i started asking myself if just like acceleration represents the growth of speed, something else could also represent the growth of acceleration itself. And it came that is exists and is called "jerk". Before i thought about this, i thought that acceleration was the ...

Emilio Pisanty

but once those are agreed

heather

not really - edge cases and such

Bernardo Meurer

@EmilioPisanty Input NULL :^)

DanielSank

10:31 PM

@BernardoMeurer I write C code. Computer does what I want.

heather

or poor programming

or running out of memory or something

brb

DanielSank

Surely you see a difference between how a human learns and how a computer takes input?

Bernardo Meurer

@heather That's semantical faults, you're cheating

Emilio Pisanty

like they are for a professional human computer

Bernardo Meurer

@DanielSank She always leaves when things get interesting...

Emilio Pisanty

10:31 PM

you just agree on a standard, and use that as input format

DanielSank

@EmilioPisanty Humans don't take procedural program instructions!

Emilio Pisanty

@DanielSank they do if you pay them

DanielSank

You can't take any human and have them run your algorithm.

You have to train them.

@EmilioPisanty Are you being cute now?

Emilio Pisanty

@DanielSank so?

@DanielSank no, I'm being dead serious

I don't see the distinction

Bernardo Meurer

@DanielSank Speak for yourself

Emilio Pisanty

10:32 PM

with an EC you need to agree on standards

with a HC you need to agree on standards

same thing

Bernardo Meurer

What is this discussion even?

DanielSank

Standards... are beside the point, I think.

Bernardo Meurer

I just came in because I felt like fighting

Emilio Pisanty

@BernardoMeurer I claim that humans are programmable, finite-memory, (relatively buggy), Turing-complete machines

computationally speaking

DanielSank

Turing complete?! That's a helluva claim.

Bernardo Meurer

10:34 PM

@EmilioPisanty If the implementation is buggy then they're not Turing complete

I see your point though

Emilio Pisanty

@DanielSank not really

nbro

@EmilioPisanty I would not call humans programmable at all, not even computationally speaking

Emilio Pisanty

If required, you can dry run a Turing machine's workings on paper and tape

that's not an accident

nbro

The most accurate description of a human you gave was given in parenthesis

Humans learn, they are not programmed!

Bernardo Meurer

Most accurate description of humans: aspogijaspoaefjocjo2rc98984rnfqchasd f

Emilio Pisanty

10:37 PM

@nbro sigh

see above discussion

Bernardo Meurer

I crafted that with my elbows :^)

nbro

@EmilioPisanty Yours is a very risky claim, which I would bet that most of us don't really agree with without evidence.

DanielSank

I'm kinda starting to agree with you @EmilioPisanty, but I still think there's some qualitative difference.

heather

hello

(again)

nbro

@BernardoMeurer Make sure you know about floating-pointing arithmetic and related problems...

Emilio Pisanty

10:43 PM

@DanielSank so... elaborate on what the qualitative difference?

heather

consider also that computers are beginning to operate more and more like humans.

Mithrandir24601

Rytsas @heather

heather

@Mithrandir24601 rytsas

Mithrandir24601

That's the first time anyone has actually said that back - thanks :)

Bernardo Meurer

@nbro What about it? The IEEE standard for floating point arithmetic works OK?

Emilio Pisanty

10:44 PM

@nbro I don't really see why this is controversial. Turing machines were modelled on humans. Of course humans can perform Turing-complete calculations.

with the only limitation being size and memory

same as any other physical implementation

nbro

@BernardoMeurer Take a course online or read a book about computational science... any serious programmer needs to know about roundoff errors, and things like that.

Bernardo Meurer

@nbro I do know about roundoff errors and the problems associated with floating point. Heck I implemented a floating point ALU in VHDL; my point was that it's an issue with how we implement floating point, not a fault inherent to computers

nbro

But it isn't the easiest subject as you delve into the details/advanced topics.

Bernardo Meurer

Well, not how we implement it, but the whole idea of a floating point

and perhaps not an issue but a quirk

David Z

@nbro It's not clear to me that it isn't a duplicate. Both questions are asking why we don't generally consider third and higher derivatives in physics. (Besides, five people voted to mark that question as a duplicate; even if I did think it wasn't, I would be somewhat hesitant to undo their decision on my own.) Maybe you could make a meta post arguing for its reopening, to get the community's eyes on it.

Bernardo Meurer

10:47 PM

@nbro docs.oracle.com/cd/E19957-01/806-3568/ncg_goldberg.html

nbro

@DavidZ Two perspectives make them not duplicate, in my opinion.

Bernardo Meurer

This is how I learned floating point

David Z

@nbro Asking the same thing makes them duplicates, in my opinion.

Bernardo Meurer

Woah, a true clash of minds here

David Z

I mean, different perspectives isn't enough.

nbro

10:50 PM

@BernardoMeurer No, it's also inherent to computers, no matter how much memory you put into them.

@DavidZ How can you consider "Why is the Lagrangian a function of the position and velocity (possibly also of time) and why are dependences on higher order derivatives (acceleration, jerk,...) excluded?" a duplicate of "why isn't jerk widely used/taught in physics"?

Bernardo Meurer

@nbro How about you don't use floating point

Use GMP

or OpenMP

nbro

lol

David Z

@nbro I don't think this is a productive way to start a discussion about it. I really think it's better if you make a meta post, if you feel strongly about it.

nbro

No, I don't want to bring this further... I just thought I was helping, but, apparently, I'm seeing dead birds flying

(bring this further?) maybe not the best expression for the purpose, or is it?

Monad

Hello, I have only learnt high-school physics but would like to understand how Earnshaw's theorem makes passive magnetic bearings (bearings that use magnetic levitation from permanent, inactive, magnets) impractical? Any places I could do this or would anyone be kind enough the explain the gist?

David Z

11:03 PM

@Monad You could probably ask that on the main site. Some people may answer in a way that's above your level, but perhaps some others will explain in a way that makes sense to you.

Monad

Should I insist for a simplified answer?

heather

"insist", no, but do state your level.

CR Drost

@doublefelix sorry I missed your message :(

Monad

okay, albeit, having narrowed down on statements such as "[It is] not possible, to stabilized all degrees of freedom of a body by passive magnetic levitation, alone", would the question not be too specific?

doublefelix

@CRDrost no prob :)

I still have time if you do

nbro

11:08 PM

Interesting project.

argo.ucsd.edu

CR Drost

@doublefelix sure. so you're totally right to identify vectors and covectors; there's a bijection there between them.

But it is important to define what this \hat e_i and \hat e^i are, and how you're relating them to each other.

Monad

@DavidZ In any case, I will try asking and thank you.

Sir Cumference

@AccidentalFourierTransform Jeez, 11 downvotes, 13 upvotes

CR Drost

the bijection here is that given an inner product $\cdot$ then covectors are just linear maps from vectors to scalars, so each vector \vec v defines some covector as (\vec v \cdot), if that helps.

nbro

kakapo.ucsd.edu/pub/argo/slides/argo.avi

Not a virus ;)

doublefelix

11:16 PM

In my mind, I have the following scheme for defining the vector and covector bases: define the \hat e_i as $\partial_\mu$, then define a metric using $g_{\mu \nu} = e_{\mu} \cdot e_{\nu}$. In euclidean space this was not an issue, as any basis vectors could be expanded in another orthonormal basis and then the regular dot product could be used. Then use that metric to "raise the index" of the e_mu. You'll need to find the inverse of the metric to get the version with upper components

I don't know enough to say whether the fact that you work on the tangent plane lets you work as if you're in euclidean space in this example

Slereah

So I found Geroch's paper on the spacetime example that is geodesically complete but still has singularities

I think it's the worst spacetime I've ever seen

It is literally made if an infinite stack of spacetime

CR Drost

Are you assuming g_{\mu\nu} is given to you and using it to define \cdot in that expression? Because if so that makes some sense.

It probably falters a bit in that it now requires you to know what $\partial_\mu$ is which will suck when you try to transfer to general relativity, but if you're just in special relativity then you can use the fields w=ct, x, y, z as coordinates and they cover the whole space and you can use \partial_w, \partial_x, \partial_y, \partial_z as basis covectors.

doublefelix

I was wondering about that \cdot... in the embedded case where I can define the basis vectors as \frac{dr}{dx^\mu}, it can just be the usual dot product (sum of squares of the components, with a minus sign if working in SR). But here I'm not sure how to form a dot product out of just partial derivatives

Emilio Pisanty

@Monad if you're quoting from somewhere specific, always cite the source

CR Drost

Well yeah, that's basically the problem you're going to face. Really \partial_\mu is an operator which turns a scalar field into a covector, because it operates on a vector field to produce a scalar.

Emilio Pisanty

11:26 PM

@Monad ~~Otherwise,~~ your question is OK as is

CR Drost

The way that it operates on that vector field is by first obtaining that vector field's directional derivative \vec v \cdot \nabla, and then applying that on the scalar field to get a new scalar field.

Monad

@EmilioPisanty you mean the one I just posted?

Emilio Pisanty

@Monad yes

Monad

ok, thanks.

I just edited btw and cited.

doublefelix

So then, how does one construct the metric in the embedded case?

oops, in the non-embedded case

CR Drost

11:32 PM

well, you'll see geometry-minded people start with the scalar fields, define vector fields as derivations on those scalar fields, then define covector and tangent fields from the vector fields, and finally identify a [0, 2]-tensor as the metric tensor, where it serves also as a canonical way to raise and lower indices.

so the definition of the partial derivative operator literally is just what I said above, in Haskell you might write grad scalar = \vector -> vector(scalar) since a vector field maps scalar fields to scalar fields anyway.

Geometry-minded people don't like the fact that things have these components, you see, with their complicated transformation rules... so those come in later, you say "also in any neighborhood of a point you can define some scalar fields which work as coordinates; any other scalar field is a smooth function applied pointwise to these coordinate fields."

@doublefelix just in case I get torn away from the screen by other responsibilities, I gave sort of an overview of what this background looks like at physics.stackexchange.com/questions/312703/… .

doublefelix

thanks. I am going to work through the site and/or some book. I was thinking of using Nakahara, since I want to see somewhat from a physics perspective, but maybe something less "quick and dirty" would be better.

CR Drost

@doublefelix one thing that might help before you get into relativity too is to focus on the solid-state side of the picture.

Usually you've got tensors like a stress or conductivity tensor that in some crystal work out to be really simple along some principal axes.

You might think, "why don't I just choose my basis vectors to lie along those axes?" and the only thing that's stopping you is, those axes are not orthogonal.

So then you can think "well I will just define a dual basis to my normal $\hat e_i$ which is that the dual vector $\hat e^i$ satisfies $\hat e^a \cdot \hat e_b = \delta^a_b$ with the usual Kronecker $\delta$."

Then any vector has two sets of coordinates which describe it, the $v^i$ in the normal basis and the $v_i$ in the dual basis, and this restores your simple formula that $\vec u \cdot \vec v = \sum_i u_i v^i = \sum_i u^i v_i,$ but only if you pair the dual coordinates from one with the normal coordinates of the other, or vice versa.

doublefelix

I feel that I have a pretty good understanding of how things work in embedded geometry. I see how a dual basis arises, I know where the christoffel symbols come from, I can describe curvature and all that. It feels just like an extension of basic vector calculus. What's been baffling me is how much more I need to learn to speak in the language that people use in differential geometry

(of non-embedded manifolds)

CR Drost

11:48 PM

Cool.

I mean, I think there's a lot of that.

If you're embedded in a bigger space then you know that you can take for granted "hey, I've got these unit vectors." When you're not embedded in a bigger space there's much more of a push for "uh, ANY numbers that can be different in a neighborhood of points need to be 'good enough' because we don't necessarily know some nice choices up-front."

Mithrandir24601

@doublefelix Yep, I felt exactly the same about the language used doing differential geometry

doublefelix

I think the definition of \partial_\mu is somewhat natural because it comes almost directly from the embedded basis vectors \frac{dr}{dx^\mu}. Having dx^\mu be a basis for the cotangent space as opposed to \partial^\mu is where I am stuck at the moment

of course I still have to get used to the operator properties of the vector basis but it seems like a very reasonable extension

Does the reason for not using \partial^\mu as a basis for the cotan space have to do with the difficulty of constructing a metric that you could use to "raise" the index of \partial_\mu?

(without first defining a basis for the cotan space)

CR Drost

No, I mean, people raise the index of $\partial_\mu$ all the time.

Transcript for