Just launched a new SE Proposal : Physics Educators for those involved in the field of teaching physics, a bit like the Mathematics Educators, if that interests anyone.

Nature to offer double blind peer-reviews. Wonder if anyone's actually going to elect to use it.

@Hakim Not sure if physics educators should really be its own SE considering that it is often (in early education) grouped together with chemistry and the other natural sciences.

ps innis you refer to "one of popper's important insights" as if it were, uh, ... an objective truth? :p

(maybe TK can now save me he has a precisely measurable 2.066K on Philosophy)

@Hakim looks interesting! Be good if it went ahead

(alarge thx) ... fyi interesting recent experiment on blind peer review in CS field

what problem is the nature policy attempting to address? scratching my head

reading... eg maybe better for/ preferred by younger scientists without as much established rep? dunno

Well, Nature has been retracting a whole lot of papers as of late.

really? interesting! in what fields? but how would double blind peer reviews help that?

And I guess that the referees were in part affected by the names of the authors.

Well the whole STAP fiasco to mention a big mishap.

It is my understanding, but I'm no expert, that with proper refereeing this would have been avoided.

nature / STAP retraction (biomedical/ stem cells)

> We have concluded that we and the referees could not have detected the problems that fatally undermined the papers. The referees’ rigorous reports quite rightly took on trust what was presented in the papers.

I followed the thing quite closely, and many working in the field said that it would've been easy to see the mistakes and to ask for more data. It is also my understanding that the manuscript had been rejected by Science earlier.

My memory is a bit hazy, but I think the rejections came before the Harvard professor was added to the list of authors. Don't quote me on this, though.

At any rate, I think the number of retractions Nature issued last year was the highest since the Schon scandal.

And the year before that Nature didn't fare too well, either.

And Nature really doesn't like to retract articles.

theres a lot of "2020 hindsight" wrt this kind of stuff.

peer review cannot really effectively catch fraud, it was never really intended for that...

Right. And STAP from what I understand was not a fraud per se. Just bad science: the kind of stuff that peer review is supposed to be able to weed out.

yeah. ppl sometimes forget peer review is a human process. it is not really a "stamp of authenticity"

its not unlike refs in sports!

there will always be "bad calls"...

Flags went up almost immediately in PubPeer, and were raised by people who would not have had access to any more material than the referees.

in some sense the system worked if the retraction is issued.

At a huge cost. Several people lost their jobs, and one of the authors of the papers killed himself.

it is possible there is more cutting-edge science going on leading to more errors. higher stakes. some increase in "cheating".

the peer review system cannot be blamed for some "consequences"...

are you talking about the STAP retraction?

had not heard of the suicide.

stem cells are indeed a "high stake cutting edge area" lately. notice schon fit that profile also.

billions of dollars aka "monetization" riding on these discoveries.

Well Nature only publishes cutting edge stuff, that's why people read it and that's why it's difficult to get publications through. When Nature publishes something, it endorses it, so of course it has to share the blame when its processes failed.

understood; yet publishing is not exactly the same as endorsing.

You can't blame, say, Bhopal on human error, these things happen, and say that the company should not bear any responsibility.

human (social) processes can never be foolproof. or cheaterproof.

in a sense, cheaters are fools...

there is prob also room for improvement.

The attitude "well, sometimes planes just drop out of the sky, it happens" is not very helpful. If this happens then obviously something went wrong, badly, and needs to be addressed.

@Danu U wot m8? ;)

@Jimdalf: Noooooooo, the Jimnosperm is winning.

alarge agreed bhopal/ plane crashes etc can also be regarded as costly/ disastrous failures in "peer review" systems :\

@ACuriousMind Please tell me you have access to Di Francesco's CFT book

@0celo7 Then I'd be lying.

As it is a Springer book, I could get a PDF of it, but I would have to visit the campus for that

(which I'm not going to do on a Sunday evening)

@ACuriousMind I'll reproduce the relevant equations then. First we have the Ward identity $$\partial_\mu\langle j^\mu_a(x)\Phi(x_1)\cdots x_n\rangle=-i\sum_i \delta(x-x_i)\langle \Phi(x_1)\cdots G_a \Phi(x_i)\cdots \Phi(x_n)\rangle$$

The goal is to show $[Q_a,\Phi]=-iG_a\Phi$

$Q_a$ is the Noether charge and $G_a$ is the transformation generator

He says to make a "pillbox". We assume that $t=x^0_1$ is different than all the other $x^0_i$s. Call the fields at $x_{2,3,...}$ collectively $Y$.

Then the pillbox is bounded by $t_-<t$ and $t_+>t$ and by spatial infinity.

Then we integrate and some how get $$\langle Q_a(t_+)\Phi(x_1)Y\rangle-\langle Q_a(t_-)\Phi(x_1)Y\rangle=-i\langle G_a\Phi(x_1)Y\rangle$$

The problem is that the Noether charge is not time dependent, so $Q_a(t_\pm)$ doesn't make sense to me.

Or how the integral getting converted into a surface integral gets this result.

If the pillbox is $\Sigma$, then I see the integral of the l.h.s. is $$\int_\Sigma d^{d-1}x\,n_\mu\langle j^\mu_a \Phi(x_1)Y\rangle$$

What is $n_\mu$? (I know it's the outward normal.)

This integral also gives me the correct r.h.s.

@0celo7 I think he just writes $Q_a(t_\pm)$ to indicate from which of the boundary terms which term comes from - of course, $Q_a(t)\equiv Q_a$, since it's a Noether charge, as you say

But then isn't the l.h.s. trivially zero?

@0celo7 Ah, implicit time ordering!

(I think)

Furthermore, how is the Ward identity not trivially zero?

@0celo7 Why should it be zero?

Divergence of the current?

The point of the Ward identities is, in general, that the classical conservation laws only hold up to contact terms in a quantum theory

(the $\delta$s are the contact terms)

This isn't particular to CFT, it happens in all QFTs

So is $n_\mu j^\mu_a=j^0_a$ in this case?

@0celo7 I think so

Yes, yes it is

Isn't the charge defined as $$\int_\text{all space}d^{d-1}x\,j^0_a?$$

If we're only integrating over the pillbox, that isn't all space

Well, if you integrate $j^0$ over any volume, you get the amount of charge inside that volume

If you integrate it over all space, you get the globally conserved charge

Is the $Q_a$ appearing in the commutator above the global charge?

Wait - isn't the pillbox whole space?

You said it is "bounded by spatial infinity"

:: hangs head in shame ::

I did say that...

Hah, it happens

Thats 30 minutes of my life, gone.

Not seeing the forest for the trees is a danger of trying to retrace steps in a calculation

@ACuriousMind It's also part of the reason that mathematicians, make formal lemmas: by studying the progression of lemmas you can get a sense of the whole, and then you can deal with the details in more bite-sized chucks to prove each lemma.

@ACuriousMind That's a wise sentiment.

@dmckee If they overdo it, this leads to the infamous French style of "proof by series of trivial lemmas", where you have a result at the end but no idea what was crucial about the proof.

So, you have to pick your way stations with care. Sure. But that's not easy. At least for me.

Unrelated question: If I cast a closevote and then retract it - is the question removed from the close queue? Is it removed only if no one else has voted yet?

Is the definition of a lemma something that is used in the proof of a theorem? Because something like Poincare's lemma is certainly theorem worthy.

@ACuriousMind No idea.

Check the mother meta, I guess.

::chuckle:: The line in chat popped up just as I clicked to respond ant that was initially directed to the wrong person.

@0celo7 There's no formal definition. Usually, the lemmata are easier to prove than theorem, or less broadly applicable, but it's essentially historical accident what's a lemma and what's a theorem

@dmckee Hm, a quick search turns up nothing. Would it even be desirable to remove it from the queue?

I always think of a lemma as being a mini-theorem that the author doesn't feel (or expect) has use outside of the current context. But of course, you find exceptions.

@ACuriousMind When you learned about index theorems and whatnot, did you learn the proofs too? I looked at Zeidler's references for the proofs, and they are way over my head. (Not that that really means anything.)

Or should I just not worry about the formal proofs?

@0celo7 With index theorems, you mean the stuff like "Euler characteristic = sum of indices" and such? Yeah, I learnt the proofs because it was a dedicated course on algebraic topology (so I don't know the proofs you are seeing, but I know some proofs). I couldn't reproduce any of them though - it's nice to have seen them and have a rough idea what's going on, but not essential.

@ChrisWhite It's an eternal fight among the mathematicians here whether the correct plural is lemmas or lemmata

@ACuriousMind I looked at Zeidler's reference for the Atiyah-Singer index theorem and the first chapter was a review of pseudo-differential operators and functors. I have no idea what either one of those are :/

@0celo7: If you mean the Atiyah-Singer index theorem, I have no idea how the proof works - I'll be learning Riemann-Roch as a special case of it next semester

@ChrisWhite It's the correct Greek plural, and German loan words generally tend to keep the original plural

Lemmas is also valid in English.

@0celo7 Did you see this question ? I think that you are our champion in GR.

@ACuriousMind In learning topics in QFT such as CFT, we Wick rotate. Why can't we Wick rotate in general?

@Sofia you know @ChrisWhite is a cosmologist/astronerd, right?

@0celo7 No, I didn't know, and I don't know him.

@0celo7 Well, you can Wick rotate in a general QFT. If you do that, it's called Euclidean QFT and looks suspiciously like statistical mechanics ;)

@ACuriousMind I know, but I mean in general.

@0celo7 But you are with the GR, aren't you?

@Sofia Sometimes. I'm trying to learn QFT/CFT and the associated mathematics right now so I can learn String Theory.

@0celo7 Whaddaya mean, in general? The purpose of the Wick rotation is to make the exponential in the path integral well-behaved - what would we gain Wick-rotating a theory that has no such exponential?

@0celo7 what you try to pass the dead cat to the courtyard of someone else? Can't you give a hand of help about that question?

(It's also annoying that, if you want to be rigorous, you have to check all sorts of stuff to make sure the analytic continuation back into the Minkowski space works)

@ACuriousMind We all learned in SR that $dt^2-dx^2$ is invariant. What's stopping us from simply defining time as imaginary and making life Euclidean?

@0celo7 Ah, but then your time is not the phyiscal time. The physical time obeys the Minkowski metric.

@Sofia The question has already been answered. I don't disagree with the answer.

@ACuriousMind What is this physical nonsense?

@0celo7 you'd better leave the Curious Mind, or otherwise you may dream of him, and you'll wake up frightened.

Also, you can work with imaginary time in SR

That was actually done until the '50s or so

People realized it really didn't make things much easier

And it doesn't help you in GR

@ACuriousMind Yeah, causal structure would be quite boring.

How are null curves manifest in Wick rotated space?

@0celo7 Nah, the problem is, you don't even have a special "time coordinate" globally in curved spaces in the sense that Wick requires. Wick rotation only works nicely in flat space/SR

I think, I'm not that sure about it

But I've never seen anyone do it in GR, and some of the QFT/GR crossover people should've thought about it.

@Sofia How could anyone be frightened by a dream in which I appear? ;)

@ACuriousMind I've never seen Wick rotation mentioned in a GR text, and I've downed a few of them.

@0celo7 It probably doesn't play nice with the structure. In flat space, the Lorentz group is simply turned into the rotation group, but I imagine the effect on the diffeomorphism group is quite different. I'm just guessing here, though

@ACuriousMind Semantical question: Let $\gamma(t):[0,1]\rightarrow M$ be a curve on a Lorentz manifold. For $t\in[0,\tfrac{1}{2})$, let the tangent vector be timelike. For $t\in(\tfrac{1}{2},1]$, let the tangent vector be spacelike. It is correct to say that the curve changes from timelike to spacelike at $t=\frac{1}{2}$?

@0celo7 Uhhhhhh. I'd say the curve is neither timelike nor spacelike, and leave it at that, I think. I don't see anything wrong about that statement, though

@ACuriousMind Look at the picture that you chose for our site. I insisted that you change it, but you refused.

@ACuriousMind I am joking.

@ACuriousMind I know the other picture of you, but I am sorry that you painted your face.

@ACuriousMind Danu took serious offense with the word "change". He said it doesn't make sense that a curve, which is a fixed object, should change. I argued that one part of the curve is timelike, the other spacelike, and hence a change must occur somewhere.

@ChrisWhite Make it $-i\,dt\,d\phi$ of course. (Not that that helps.)

@Sofia I know you're joking, I'm not that serious, either. ( The smiley " ;) " at the end of a message is intended to look like a wink, and show that the message is not completely serious)

@ACuriousMind my dear one, can you give a hand of help? There is someone here that has a problem with GR, in which I am tabulla rassa. Here is the fellow.

@ACuriousMind @ChrisWhite If you haven't, you should check out Sachs and Wu -- GR for Mathematicians. They call cosmology, like the rest of physics, circular reasoning, because we use GR to interpret observational data and then claim GR is verified.

@ACuriousMind He expect an object rotating around the Earth to spin.

I'm not sure if I agree with the "rest of physics" part.

@0celo7 If they take issue with us using the theories we want to validate to interpret the data we get, then they haven't understood what science is about at all.

@ACuriousMind In the Newtonian mechanics there is no spin. Would you take a look?

@ACuriousMind They literally have a paragraph titled "circular reasoning"

@Sofia I've looked, and the question makes sense to me. I don't know the answer, though. My understanding of GR in actual, physical situations is limited, but since it reduces to Newtonian physics in limits, I think the thing should not turn, because gravity around the earth is very well approximated as Newtonian for most things.

@0celo7 Yeah, as I said, they haven't understood how one goes and tests theories at all. If you don't use any theory to interpret the data, all you got is a meaningless bunch of numbers and observations.

@ACuriousMind can you leave him a comment? Please!

@ACuriousMind I don't know why they wrote the book. They listed in the preface the people who would be disappointed by the book. It includes: people not willing to do horrendous calculations in cosmology, people seeking an axiomatic treatment (because axioms don't exist in physics) and those seeking an encyclopedic treatment. Furthermore, they cite Hawking & Ellis as a source for physical intuition, because their book is devoid of it :)