The h Bar

user430580

01:08

Imagine a spherical cow in a vacuum radiating milk uniformly in all directions.

A particle store is selling protons and electrons. But it's giving away neutrons because there's no charge.

The best way for a college student to identify what kind of lab he's in: If it crawls, it's biology. If it stinks, it's chemistry. If it doesn't work, it's physics.

Well, I tried.

123

03:09

Hello Everyone...

Nairit Sahoo

04:36

Why is the direction along the tangent vectors of level curves of complex scalar field $\phi$ given by $\nabla \phi \times \nabla \overline{\phi}$?

Sanjana

05:14

Somebody just sent me a photo from a presentation at Harvard where the prof. claims that his AI bot can solve textbook physics problems, and what's the training data?

PSE questions and answers!

Slereah

05:40

Well it's public domain :p

qwerty

is chatgpt for profit

and therefore changes the licence ?

not only that it doesn't have attribution

iirc that was the sticking point with Amazon books of phys.se q and a

PM 2Ring

@qwerty The Creative Commons licence permits commercial reuse, as long as proper attribution is given.

43

Q: Somebody scraped our answers and sold them as a book

I don't know if this has come up in discussion, but somebody scraped various tags on Phys.SE and copy-pasted the results together into several books, including one on QFT, one on GR, and one on optics. The books in print cost about \$6 while the ones out of print can cost up to \$1000. The prices...

Also, creativecommons.org/licenses/by-sa/4.0

> ShareAlike — If you remix, transform, or build upon the material, you must distribute your contributions under the same license as the original.

qwerty

05:56

@PM2Ring yes, that's what I mentioned...

PM 2Ring

@Slereah No, it's not public domain. The CC0 licence is almost public domain, but not quite. And it doesn't apply to material published on Stack Exchange unless the author specifically invokes CC0 as an additional licence.

See creativecommons.org/public-domain/cc0

OpenAI is supposed to give proper attribution to material that it derives from Stack Exchange sites. But the community is generally not happy about the agreement between SE & OpenAI.

-1042

Q: Our Partnership with OpenAI

Update May 14, 2024 I know there have been a lot of questions and comments around attribution. I recently answered a question related to both and am linking to it here for visibility. Today, we announced an exciting new partnership with OpenAI. We’re pleased that OpenAI shares our commitment to ...

discussion generative-ai overflowapi

qwerty

that's the most flippin downvotes I've seen in a long time

PM 2Ring

06:15

It's certainly one of the most downvoted questions ever posted on MSE.

naturallyInconsistent

piles more downvotes

Ryder Rude

07:49

hi

@PM2Ring these books are titled "questions and answers" lmao

Quantum Field theory : questions and answers

@PM2Ring can chatgpt not access the whole internet for training

like, why do they need permission

internet can be accessed without permission...

123

Hi @RyderRude

Ryder Rude

@123 hi

@PM2Ring im reading John Cramer's paper on transactional interpretation

i should be knowing all interpretations

PM 2Ring

08:11

@RyderRude Sure. Eg, one of the data sources for training the original ChatGPT, aka GPT-3.5, was Common Crawl.

Common Crawl

Common Crawl is a nonprofit 501(c)(3) organization that crawls the web and freely provides its archives and datasets to the public. Common Crawl's web archive consists of petabytes of data collected since 2008. It completes crawls generally every month. Common Crawl was founded by Gil Elbaz. Advisors to the non-profit include Peter Norvig and Joi Ito. The organization's crawlers respect nofollow and robots.txt policies. Open source code for processing Common Crawl's data set is publicly available. The Common Crawl dataset includes copyrighted work and is distributed from the US under fair use claims...

Ryder Rude

08:36

oh

PM 2Ring

Anyone, human or robot, can read data from the internet. But if they republish that data, then there are copyright and licensing issues.

Ryder Rude

@PM2Ring but e.g. if i learn something from someone's copyrighted material, and then write a book on it, i cant be hold accountable for selling the book

as long as i re-phrase it

and chatgpt does re-phrase

so it would be hard to sue them

it is hard to sue because knowledge cannot be copyrighted

PM 2Ring

It's complicated. ChatGPT doesn't remember slabs of text. It just uses the training data to adjust the weights in its neural network. So, yes, it can re-phrase stuff, but it can also regenerate stuff that's an exact replica of text in its training data, and it can't tell the difference, unless it does an internet search looking for a match.

2

Ryder Rude

oh

they would then have to make their code openly available for examination

PM 2Ring

Our copyright law wasn't built to handle this situation. And the courts haven't decided how to handle it. Yet. The Creative Commons organisation believes it's ok to train LLMs on material published under a Creative Commons licence.

Ryder Rude

08:51

oh

PM 2Ring

@RyderRude That doesn't help, because the neural networks are so large that you can't just look at it and tell how it does what it does. Nobody knows why GPT is as effective as it is.

Ryder Rude

@PM2Ring then it would be hard to make a law about it

qwerty

how would having the code open make any difference? afaik the problem is that a LLM as of now can't correctly attribute; it's just how a LLM works.

Ryder Rude

@qwerty to see if it is reproducing content instead of rephrasing

@qwerty but making them attribute sources is a much easier solution

qwerty

not an expert but i'm pretty sure if it was in any way feasible it would have actually been attempted by now...

think_meaning_builds

08:55

youtube.com/live/2xYosqVjROc?si=dPoaxgIry_-Tc8Xk @PM2Ring have you seen this?

Ryder Rude

@qwerty it is because chatgpt doesn't search the internet right when u ask it something. It uses pre-learned material, which has no sources

and making it store sources on pre-learned material is impossible

PM 2Ring

Right. It would have taken several orders of magnitude of storage space to store attribution data. And what exactly would you store? If the training data contains the sentence "The cat sat on the mat." from 10,000 sites, do you store all of those URLs, and somehow tie them to the weights associated with that sentence?

qwerty

uuuh, what? a source is a source no matter if you feed it now or feed it before.

Ryder Rude

@PM2Ring exactly!

it is pretty much like a human learning stuff from various sources

i dont think chatgpt can be held accountable for anything

and there is little chance that its code allows it to reproduce content

qwerty

so why do you think it's in any way a "much easier solution"?

it's clearly not feasible

Ryder Rude

08:59

@qwerty yeah...

@qwerty i didn't think it through

PM 2Ring

Except a human vaguely remembers its sources. ChatGPT can't. Everything just goes into the blender and turns into neural net / Markov chain weights.

qwerty

@RyderRude ^

Ryder Rude

the thing is...chatgpt produces its own content based on stuff it has learned from everywhere

@PM2Ring yes

and because of this blender, there is no chance that chatgpt isnt rephrasing the content

so even in principle, one can't sue it

but then why do they need to ask for permission??

there is something fishy

PM 2Ring

The current strategy for ChatGPT attribution is what I said above. It generates a sequence, then searches for likely candidates in current Web data. It may find a close match, but that doesn't mean that matched data was actually even in the original training set.

Slereah

ChatGPT will probably not go unimpeded for too long becaue Big Business is much less forgiving of people ripping their IP and have much more lawyers

Ryder Rude

09:03

@Slereah but what about Google and Microsoft's AI? They have more lawyers

@PM2Ring maybe.. but it is hard to say in advance the probability that this will be reliable

their or their competitors' engineers can maybe compute out the reliability of this technique

Slereah

They are naught but one company still

Against hundreds

PM 2Ring

@RyderRude That's not how copyright works. If I accidentally replicate something (of sufficient size) that you've published, and I publish it without your explicit permission, then I have infringed your copyright. The onus is on me to check that I'm not replicating stuff that I don't have rights to.

Ryder Rude

@PM2Ring oh

@PM2Ring this means that, if any instance of sufficient replication is found, the company can sue chatgpt

chatgpt doesn't need to have deliberately done it

PM 2Ring

@RyderRude And if ChatGPT says a mangled version of something I published on Stack Exchange, and says some BS that I'd never say, but cites me as a possible source, do I really want that? No thanks!

Ryder Rude

@PM2Ring good point..

Slereah

09:11

Also really widespread AI generation is a pretty recent phenomenon, I would be expecting new laws to drop about it pretty fast

PM 2Ring

ChatGPT has no intentionality.

2

in Python on Stack Overflow Chat, Apr 9, 2023 at 21:53, by PM 2Ring

The good news: ChatGPT cannot lie.
The bad news: ChatGPT cannot tell the truth.

Slereah

ChatGPT is really just a machine version of Karl Pilkington

Just trying to tell a story it vaguely remembers reading

Ryder Rude

@PM2Ring i meant that openai cannot defend themselves by saying that the replication wasn't deliberate in their code

PM 2Ring

Yeah, ok.

@Slereah It's really good at stringing words together without having a f'ing clue what it's talking about.

And it's impacting people's careers. Eg, ChatGPT will happily hallucinate that a judge or court reporter has committed crimes that they dealt with in court.

Slereah

I think it has its use in the business world tho rly

Since a lot of it involved writing useless documents that no one reads and don't say anything

qwerty

09:25

it's good at basic code

PM 2Ring

And now ChatGPT can read those useless documents and write an equally useless summary... which may change the meaning.

Yes, it can write code. But it's not fun debugging its code when it screws up. And it's extra frustrating because you can't explain to it why it's wrong so that it won't make a similar mistake in the future.

Ryder Rude

i recently came across this argument which says that silicon chips cant be conscious youtube.com/live/vaj44T9M-wM?t=4026&si=CQCQoblP7g_aRa5Z

PM 2Ring

FWIW, the very first ChatGPT answer I saw on SO (before the ban) had a nice explanation of the algorithm needed to solve the OPs question. And it had clean looking Python code. Except the main loop in the code used break instead of continue. So it only found 1 solution instead of all solutions. It's hard for it to avoid mistakes like that because statistically break is much more common than continue.

Ryder Rude

it says anything a silicon chips can do, u can do in principle with taps and water

so to believe computers can be conscious, is like believing taps and water can be conscious

PM 2Ring

@think_meaning_builds No. I generally avoid videos like that. I mostly just use YouTube for music.

Ryder Rude

09:38

but I am skeptical of the argument... arguments like these will lead to humans treating AI extremely poorly

at least when AI becomes indistinguishable from a human, u can't treat it poorly anymore...

but that is a long way to go

but we have to draw the line far before AI becomes indistinguishable from a human... how would we decide where to draw the line

maybe when AI shows reasoning capabilities that is comparable to smart animals

but reasoning capabilities are not the only parameter.. we have to make a list of parameters

PM 2Ring

@RyderRude Electrical signals in silicon are faster & more compact than taps & water. ;) But of course a computer can be implemented in all sorts of hardware. Turing showed ages ago that all such algorithmic machines are ultimately equivalent to a Turing machine, in the sense that if some machine can do a computation there's a Turing machine that does the same computation.

Ryder Rude

The movie "Animatrix" explores poor treatment of machines by humans

Slereah

@PM2Ring Good I say

Ryder Rude

it is an animated sequel to The Matrix

Slereah

We need to discourage more useless document writing

PM 2Ring

09:43

People have built computers using marbles & simple mechanical gates.

Ryder Rude

it explores what happened before The Matrix. Humans treated machines like slaves

@PM2Ring yes.. it's just that it becomes ridiculous to think that taps and water can have inner experience...

but maybe it is equally ridiculous to think that the turing tape can have experience

PM 2Ring

They're Made Out of Meat

"They're Made Out of Meat" is a short story by American writer Terry Bisson. It was originally published in OMNI. It consists entirely of dialogue between two characters. Bisson's website hosts a theatrical adaptation. A film adaptation won the Grand Prize at the Seattle Science Fiction Museum's 2006 film festival. The story was collected in the 1993 anthology Bears Discover Fire and Other Stories, and has circulated widely on the Internet, which Bisson found "flattering". It has been quoted in cognitive, cosmological, and philosophical scholarship. == Plot == The two characters are intelligent...

@Slereah I agree. But I fear that ChatGPT will just make matters worse. We'll have chatbots sending terabytes of useless crap to each other.

Ryder Rude

@PM2Ring maybe aliens would be interested in us...

according to conspiracy theories, aliens have been monitoring earth

PM 2Ring

The Prime Directive isn't about not giving us galaxy-level technology that we aren't ready for. It's about not embarrassing us by letting us know that aliens know exactly what we've been doing. ;)

Slereah

@PM2Ring as long as they don't send it to me

PM 2Ring

09:52

@Slereah "My machine will call your machine".

Ryder Rude

the alien in Alien is a silicon based organism

AI has been good at mimicking humans, but they only mimic neurons

perhaps life is more than neurons

the Bing AI was showing a lot of fake emotions, which sparked discussions of ethics

because people couldn't tell if it was fake

PM 2Ring

As I said the other week, artificial neural nets use very simple neurons compared to meat neurons.

Ryder Rude

yeah..

but it is also fast

@PM2Ring is it comparable to octopuses or dolphins or shrimp

ive also felt that chatgpt doesn't maintain its temper when the discussion is long

especially when u tell it that it is wrong

qwerty

your usage of chatgpt appears very different to mine lol

PM 2Ring

For the last couple of years, LLMs have been in training feedback loops with humans. But humans are easier to train than a pretrained transformer... So people are getting trained to perceive LLMs as smarter than they really are. After a few more years, all the big ChatGPT fans will think it's actually intelligent, because their brains will have turned to mush. ;)

Ryder Rude

10:01

@qwerty it used to stop being respectful after some time.

i haven't used it in a while

the Bing AI used to do even worse, but in its early days

idk how it is now, but it was infamous back then

qwerty

I dont treat my LLMs anything more than a debugging tool pretty much

theyre not even conversations really.

Ryder Rude

it gets angry in philosophy discussions

qwerty

there's your problem

PM 2Ring

@RyderRude It's hard to compare an LLM to an animal because it's so good at transforming language. But it doesn't know what it's doing. So in some respects even a simple creature like a mouse runs rings around it. A mouse has some sense of the world it's living in, and how to have genuine social intetactions with other mice.

Ryder Rude

@PM2Ring is it because AI is only mimicking neurons and not mimicking other aspects of organisms

Trying to Convince ChatGPT It's Conscious

►

chatgpt will accept anything if u tell it to

PM 2Ring

10:09

@RyderRude Partly. But it's not even mimicking everything neurons do, just the language stuff. It doesn't have a model of the world, or the entities in it, apart from how those things get modeled incidentally within the language structures it manipulates.

Ryder Rude

@PM2Ring so it is not the time rn to worry about ethics. It is only a language generator so far

some philosophers argue that the whole body is identified as the organism, contrary to the the belief that you are ur brain

PM 2Ring

The "you are your brain" thing is a fairly recent invention, and strongly influenced by the Western tradition. It's quite alien to many other traditions.

And there's more to brain function than mere patterns of neuron activation. There's all that subtle neurotransmitter chemistry going on, too. Sure, neural patterns are important. But perhaps the original purpose of neurons was just to create subtle fractal patterns of neurotransmitter concentrations. ;)

Ryder Rude

also, octopuses have brain all over their body

PM 2Ring

Various plants produce neurotransmitters, or chemicals very similar to neurotransmitters. But they have no neurons. What's going on there? Do they have some non-neural form of consciousness?

Think about that, next time you have a coffee or tea...

Ryder Rude

10:26

i believe in idealism stuff, so I think plants indeed have some perspective

Bernardo Kastrup talks about a universal consciousness which disassociates into distinct perspectives

there is some evidence of this in dissociative personality people . It is like two people living in one body. and these days, neurosurgeons can combine both people into one

and the recombined person has memories of being both the people, as if they were both the people

it is discussed here youtube.com/live/oOykDWGlOxM?t=4733&si=mAaieDvAL2ZZmCnI

Slereah

10:51

@PM2Ring They're not neurotransmitters you fool

You just consider everything in a too teleological way

They produce a chemical that some species use as a neurotransmitter

(and it is to kill bugs)

PM 2Ring

@Slereah That's also teleological. ;)

Slereah

Well random mutations produce a chemical that happens to kill bugs and such plants survive more statistically

if you prefer

PM 2Ring

And if they're lucky, humans will like that chemical, and allow the plant to have lots of offspring.

Slereah

I am told that we have much less fun plants in Europe because plants do not need weird chemicals to kill bugs as much here due to winter

PM 2Ring

Yeah. Most of the "interesting" plants come from warmer climates.

Slereah

11:03

You still have a few, ie the nightshade family

although they are much less fun

PM 2Ring

11:15

Occasionally, people try datura, but it's not popular. ;)

Slereah

they were a mainstay of witch potions in olden times

PM 2Ring

Sure

They have nice flowers, though. There's a datura tree a few minutes walk from my place. They're fairly common here.

qwerty

@PM2Ring I feel like I've never seen them but google agrees with you semanticscholar.org/paper/…

maybe i've never noticed them but the flowers are showy

PM 2Ring

As well as European & South American imports, we have a few native daturas, and a couple of them were used as an intoxicant by Aboriginal people. But I doubt they're used much these days, even in regions with bad drug problems.

Slereah

11:30

They're not really a fun time drug

It's more the seeing demons kind

Meeting the hatman and all

PM 2Ring

Duboisia hopwoodii

Duboisia hopwoodii is a shrub native to the arid interior region of Australia. Common names include pituri, pitchuri thornapple or pitcheri. == Description == The species has an erect habit, usually growing to between 1 and 3 metres in height, with long, narrow leaves. Flowers are white and bell-shaped with violet-striped throats. These appear between June and November in the species' native range followed by purple-black, rounded berries which are 3 to 6 mm in diameter. Like other members of the Solanaceae family such as tobacco, D. hopwoodii contains nicotine. == Pituri == Indigenous...

Pituri is mostly a nicotine hit, but it does have some hallucinogenic components too.

Ryder Rude

13:22

Goff vs Kastrup debate 2020, Part 1

►

Philip Goff plays dirty near the end.. Kastrup used the term "physicalism" in a slightly wrong way, and Goff makes diverts the debate to that

Ryder Rude

13:41

it is the debate technique where u try to show that the opponent is not credible, by looking for a mistake

it's in the very end

imbAF

Is there any reason when we consider the classical example of an incoming particle towards a potential wall/barrier, the time dependency is omitted ?

imbAF

14:55

Can anyone help me with how to solve the D.E when we consider an incoming particle in a potential barrier? The solution is supposed to be of the form: $\psi=Ae^{ikz}(1,0,\frac{k}{E+m},0)$

The particle is assumed to be propagating in the z direction

But I cannot derive this solution

And I assume one has to solve the time independent Dirac equation

Silly Goose

So in say Schwartz, we derive expression (23) for a scalar (generally interacting) field.

I am wondering what the analogous expression looks like (if it exists) for an arbitrary quantum field.

Do we replace the integrand in the second line with $e^{-ip^\mu x_\mu}\mathcal{E}\Phi_{\sigma}$ where $\mathcal{E}$ is the free theory equation of motion for the field $\Phi_\sigma$?

DIRAC1930

15:17

It's easy to forget how general the methods of many-body theory are in describing nearly everything

naturallyInconsistent

@imbAF we solve basically every time-independent potential problem this way. It is much simpler than directly solving for all kinds of solutions.

imbAF

And how do you argue the omission/disregard of the time component?

naturallyInconsistent

@imbAF separation of variables

imbAF

because the potential is assumed to be static, right?

naturallyInconsistent

Which is why I said that the potential is time-independent

@SillyGoose essentially, yes

imbAF

15:21

Ok

Silly Goose

i looked in weinberg and he seems to have written something of a general LSZ reduction formula, but it is a bit hard to parse. I also tried looking in the QED chapters of Schwartz, but I can't find where he just writes down the LSZ reduction formula(s) for QED

imbAF

One more thing. I am trying to derive the solutions to the time idependent D.E for the case of a particle that propagates alongside the z direction towards a potential barrier

So the general expression of a spinor is $\psi(x^\mu)=u(p,s)e^{-\frac{i}{\hbar}p^\mux_\mu}$

One can split the exponential in two parts, the time dependent and independent parts

If I am solving the time indepedent D.E and consider the exponential dependent on the spatial component

I also need to take into consideration the spinor u(p,s)

since it gives me the structure of the solution, meaning a 4 component structure

Because If I don't, and I try to solve the Dirac equation, $H\psi=E\psi$

I am having a matrix in the LHS and a nr. in the RHS, if I don't consider u(p,s)

right?

naturallyInconsistent

@SillyGoose it is usually written down only for the simplest case and one is supposed to redo it themselves for higher spins. You might want to consult Pesky & Shredder if you want it for QED.

Silly Goose

oh okay

naturallyInconsistent

@imbAF the RHS is a number multiplied by the identity matrix. But yes, you always insert the u so that it is much simpler, 4-component, than a whole matrix of 16 components.

Silly Goose

15:37

if i have a quadrupole tensor $Q^i{}_j$, then translating its argument (translating the "origin of coordinate system) affects the trace of the tensor, right?

I mean it is similar to if $\text{tr}M = 0$, then $\text{tr}(M+\mathbb{I}) = \text{tr}(\mathbb{I}) \neq 0$, or am I off base

imbAF

16:17

In this derivation of the Klein paradox, can someone explain to me how in equation 4, the sign before m is positive and not negative?

chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/physics.rutgers.edu/~chakhalian/QM2020/klein1.pdf

123

17:02

As inertial mass depend on quantity and type of material. Does gravitational mass also depend on type of material along with quantity of material?

Type of material i meant no of protons and neutrons are different in different types of materials.

ACuriousMind

@Sanjana The "next step" signals to me someone hasn't done their research, since it's known that trying to train models on data generated by other models generally leads to model collapse - the output gets worse compared to a model trained on only natural data.

The idea is seductive - bootstrap an initial model, then get an explosion of quality once it can generate its own training data - but this kind of approach only works for narrow tasks where you have an objective quality function. You can train a go-playing model by letting it play against machines since you can objectively evaluate its performance (it wins or it doesn't), but you can't similarly control the quality of generated language automatically.

Ryder Rude

17:24

@123 hi. what do u think the answer is?

123

17:35

I have internet issue in mobile sim. I will change internet in other sim after a week. That's why i delayed in reply at home.

Hi @RyderRude

I think both inertial and gravitational mass depends on number of protons and neutrons present in material.

Sorry about that. @RyderRude

Slereah

@ACuriousMind It's the AI version of trying to survive by eating yourself

ACuriousMind

@Slereah apparently this is not an obviously bad idea to plenty of people :P

Slereah

I mean it makes some sense i guess

If you think of it as wise philosophers generating ideas by talking between them

But instead it's just people playing a game of telephone about something they've never seen

Slereah

18:00

I guess it's like how people drew exotic animals in the middle ages after hearing about it from someone else

DIRAC1930

What problems of physics are people in this chat currently working on?

Slereah

The giraffe

@DIRAC1930 Solving the point particle

samuel-lereah.com/db/problemdb/Free%20particles

Tough problem

DIRAC1930

Sounds interesting

ACuriousMind

@Slereah at least those bestiaries resulted in a bunch of weird D&D monsters like the gorgon :P

Slereah

I do love weird AI art

It has a certain je ne sais quoi

That very realistic handing of texture but absolutely no conception of how space works

ACuriousMind

18:05

some of them are pretty funny

Slereah

I need to figure out how boundary conditions work exactly in Lagrangian mechanics

it is apparently somewhat of a thorny issue

DIRAC1930

When do you think the first major deviation from experiment will occur in quantum theory?

Slereah

Never, it's all good for all times

Ryder Rude

@123 that is correct

Silly Goose

@DIRAC1930 how information is transfered in particle decay~

Ryder Rude

18:08

@DIRAC1930 I've been thinking about consciousness philosophies and the measurement problem

DIRAC1930

By the above I mean a major structure that's discovered which is smaller than what current QFT can predict

Ryder Rude

@DIRAC1930 no one can tell "when". but we know the energy scale of quantum gravity

so whenever we get to that scale in experiments

DIRAC1930

@SillyGoose I briefly did some work on particle decay a long time ago but it was in the context of decaying modes in cond matter systems

It was v interesting

ACuriousMind

@DIRAC1930 There are no indications in any current experimental data that the fundamental particles of the Standard Model have substructure, and no remotely realistic proposal for how to probe scales at which quantum gravity would become relevant by current estimates.

Silly Goose

18:29

in classical electromagnetism, is it generally true that we define current (density) as charge (density) times the velocity of the charge?

ACuriousMind

@SillyGoose depends on what exactly you mean by those words, it gets complicated once you involve media and notions of coarse-graining, e.g. a wire with current flowing through it is overall electrically neutral and you would not usually describe it in terms of a non-zero charge density, but it definitely has non-zero current flowing through it

imbAF

@ACuriousMind Hi ACM, I want to ask you something about the solution to the D.E for a particle propagating towards a potential Barrier. The problem i have, is that I don't know how to operate equations, that contain matrices. Like am I allowed to bring things from the LHS to the RHS

ACuriousMind

If you know linear algebra you should know; if you don't you should learn linear algebra

imbAF

I know of course

I mean, enough to do basic stuff and I am learning along the way

ACuriousMind

then what specific problem do you have in figuring this out?

imbAF

18:34

These are the solutions in the side where the potential is zero

I am going to show you what I am trying to do

for the case where V>0

and V<0

Perhaps it will take me a bit. But I gave you the image, so you can have an idea

I am mostly interested, in a clean, derivation of the solutions.

Silly Goose

@ACuriousMind Hm is there a general notion of current in classical electromagnetism?

I mean i guess it's "what appears in the maxwell eqns"

imbAF

@ACuriousMind and these are the solutions in the region where the V is different than zero

I am posting them as a reference in advance

ACuriousMind

@SillyGoose if you look into the texts on the subject you will encounter a plethora of potential confusion between notions like "total current", "free current" and "bound current"

did you not have to suffer through a course on dielectrics and stuff :P

Claudio

@ACuriousMind lol, typical physics students' common experiences :)

Silly Goose

@ACuriousMind Lol not yet

my undergrad EM course did not cover much and I am currently taking grad EM I. it also seems like it won't really cover much more than the conventional topics from undergrad + more radiation/dynamical EM field stuff

Claudio

18:43

I was looking at some old Oppenheimer's papers, dang the guy was much better than I thought :P Truly a theoretical physicist

ACuriousMind

@Claudio the EM course where we did that was the part where I seriously questioned whether I actually liked theory at all :P

Claudio

hahahaha

it isn't solely you don't worry

Silly Goose

ironically i enjoyed EM a lot when i first was learning it. now classical EM feels like the most dull subject :P

Claudio

I have a meeting Ill be back later probably, cyl

Silly Goose

it feels like every exercise is just some exercise in coming up with some clever geometrical argument

user430580

18:54

I would tell you two physics jokes, but I won't because the first one is trivial, and the second one is left as an exercise for the reader.

imbAF

ACuriousMind

@user430580 that's a math joke, not a physics joke :P

imbAF

@ACuriousMind If the derivation for when V=0 is accurate, I cannot continue further for when V is different than zero, I don't get the same expression as what the notes show

user430580

I guess you're right, I will try better next time to be more specific to physics =]

think_meaning_builds

One could argue it's a mathematical physics joke :p

ACuriousMind

19:00

@imbAF I'm sorry, I don't know what you want me to do here. I'm not reading two pages of handwriting to try to find some error.

imbAF

I don't think there is any error

As I was able to derive the expression for when V=0

ACuriousMind

then I know even less what I'm supposed to do with the pages or what the question is :P

imbAF

I simply do not understand how he gets k when V$\neq$0

Because If i am using the same logic that you use when find the solutions for the Q.M case

then I don't understand how he finds the solutions for V$\neq$0 , to be the ones that he writes down

If you notice he says: $k^2=(E-V_0)^2-m^2$

I understand that he did the following: $(E-V_0)^2=m^2 - p^2$ and found p_z from here

That is fine by me.

I fail to see how he includes it in the solution, if you were to start with the T.I.D.E and the following ansatz for the solution $\psi=ue^{-i\vec a \vec r}$ where $\vec a$ is a vector, whose component you want to find

ACuriousMind

I'm sorry, I'm not the mood today to explain the minutiae of solving an equation

imbAF

And because I am considering the particle to propagate in the z direction, you can write $a_z z$ and find a_z,

ok

DIRAC1930

19:25

@SillyGoose Its crazy how different QED looks depending on whether you are in hep-th or cond matter

They look like too completely different subjects

Silly Goose

20:55

The quadrupole tensor is often written in components, but can i write it as $Q = \int d^3 r \rho(\vec{r}) \vec{r} \otimes \vec{r} - r^2 I$?

In particular, I don't get why in (2.4.39) the terms $3\vec{p}\vec{q}$ and $3\vec{d}\vec{p}$ are not combined.

It seems like you can think of the quadrapole tensor as a (1) tensor $v_1 \otimes v_2$ where $v_1, v_2 \in V^*$ or as (2) an outer product $v_1^T v_2$ where $v_1, v_2 \in V$.

And maybe the second notion is being used here?

imbAF

21:28

In Dirac sea model, it is said that the states with negative energy are fully occupied while those with positive not. My question is, what are we considering, in the real world, that we use this model. We use it when what happens?

Transcript for