Strange thought

1. Jun 29, 2006

TriTertButoxy

Ok, so we all know the following...

Elementary quantum mechanics is constructed by postulating the canonical commutation relation (CCR) between the coordinate of a particle and its conjugate momentum.

Quantum Field Theory, on the other hand, is constructed by postulating the exitence of a field and by imposing the same sort of CCR on field at every space-time point (as though each point on the field were an elementary quantum mechanical particle).

Doesn't it seem strange that the particles that emerge as a result of applying CCR to fields themselves obey the CCR imposed by ordinary quantum mechanics, (in the low energy limit)? I think its kind of spooky.

2. Jun 29, 2006

CarlB

I think the best introduction to the concept of the difference between QM and QFT is "Quantum Field Theory in a Nutshell: Books: A. Zee". Just reading the first chapter will likely reduce some of the spookiness.

Carl

3. Jul 1, 2006

nrqed

This is an excellent question, TriTertButoxy.

I have always been annoyed by the standard introduction to quantum field theory in which one imposes those commutation relations on classical fields. This always seemed very strange to me. What the heck are those classical fields that one quantizes? (no one has ever seen a classical approximation of a meson field or of a fermion field!)

As far as I know, only Weinberg presents things in a way that really makes sense, in which the starting point is not the quantization of classical fields but the need to allow the number of particles to vary, so the need to introduce annihilation and creation operators. Then , imposing Lorentz invariance, one is led to quantum fields obeying the usual commutation relations (postulated from the start in standard presentations).

Going back to your question, and as far as I understand, the commutation relations (CRs) on the quantum fields have nothing to do with the commutation relations on X and P imposed in nonrelativistic quantum mechanics! In the sense that there is no way to start from the CR of quantum fields and do some approximation to get back to the QM CRs (one way to see this clearly is that there is no analogue of the position operator X in QFT).

I have to warn you that my views on QFT are non standard (a couple of years a go I launched a thread that led to a very long discussion on this and it was clear that I think differently than almost everyone else!). The only book hat seemed to have address things the way I needed to see them addressed is Weinberg's book. Before reading his introduction to quantum fields, I had always been very confused and annoyed with the standard presentation of QFT. It just did not make sense to me. It seemed to involve such a huge leap of faith with no justification at all (we start with classical fields and we quantize them...as if this should be obvious as the thing to do!!!). And book after book after book keeps starting from exactly the same point, without justifying why this must be the way to go. Until finally Weinberg did it the "right" way (in my point of view, of course). The only problem with Weinberg is that it is so dense that it is difficult to follow for a beginner. I wish that there would be an introductory book on QFT that would present things this way!!!

Anyway, just my two cents.

Regards

Patrick

Last edited: Jul 1, 2006
4. Jul 2, 2006

Eye_in_the_Sky

My understanding of Quantum Field Theory is rather rudimentary. Nevertheless, I feel I may have some useful thoughts to offer in this thread.

... And so, was born the MISNOMER "Second Quantization".

By way of analogy to the photon, those classical fields would just be the fields which one is to envision in the (hypothetical) case where a COHERENT source produces an extremely LARGE number of the particles in question.

Of course, with regard to photons it is possible (and, moreover, quite reasonable) to expect to be able to DISCOVER the relevant field equations on the basis of classical principles alone. After all ... historically, this is precisely how it happened.

But on the contrary -- say for example, with regard to the electron -- it seems completely unreasonable to think that one could come up with the Dirac Equation by way of only classical physics principles. ... How the heck can you do that?!

This difference, however, is only a technical one. On the conceptual level,

"the Dirac Equation is to the electron" as "the free-field Maxwell Equations are to the photon".

Therefore, it makes just as much sense to take the Dirac Field and quantize it, as it does to take the Maxwell Field and quantize it. ... Wouldn't you say?
__________________________________

EDIT: This post is only part 1. The next part will appear below, some time later ... I hope.

Last edited: Jul 3, 2006
5. Jul 3, 2006

nrqed

I hear what you are saying. And I agree that this is the standard point of view.

First, this is not the way it is presented in any textbook on QFT I know (if it is done in in some textbook, I would love to hear about it). If one was serious about presenting things this way, one would have to show in depth the connection between a coherent quantum EM field and the classical field. I am still amazed (and a bit annoyed) by the fact that this is never done in details in textbooks. I could not even find the words "coherent states" in the index of Peskin and Schroeder!!
(and I must admit that I still don't feel I really understand very well this concept and the connection with definite phase and amplitude and so on)

If this is the best way to introduce quantume fields, then one should start by showing very clearly the connection between a coherent state of an EM quantum field and classical fields.

On the other hand, I find that it is much more natural to start with the observation that any particle is observed with a discrete nature (as "packets" of energy and momentum) and that is is natural to
build a formalism of varying number of particles (because of special relativity) and to see the concept of field as emerging from allowing the number of particles to vary *and* imposing Lorentz invariance (this is what Weinberg does). This starts from the *observation* that particles are observed and *leads* to quantum fields as a "bookkeeping" device. I find this much more intuitive than to start from postulated classical fields (which do not exist, and which have never been observed even as an approximation of a quantum field!!).

That's just my point of view, which I seem to share only with Weinberg!!

Patrick

6. Jul 3, 2006

vanesch

Staff Emeritus
Hi Pat,

I agree with most of what you write here ; especially the rather obscure voodoo at the start of most QFT books. But this is usually how many subjects are pedagogically introduced: "listen, I'm not gonna justify everything, but take my word for it now, and let's get going..." ; because otherwise 3/4 of the course time is devoted to the first chapter, with still some questions remaining, and no elementary application can be done.

I have to say Weinberg's approach was an eye-opener to me, but I didn't feel lousy about the "field" approach either. In fact, Weinberg is a bit of a party-pooper because with the equivalence of both viewpoints (variable particle count + relativity <==> quantized relativistic fields) it's now impossible to see one approach as more fundamental over the other.

The way I saw QFT before was: the quantum "prescription" applied to a specific classical model, namely the model of relativistic fields, in the same way as the quantum prescription was applied to newtonian/hamiltonian point particle mechanics, and resulted in non-relativistic quantum mechanics.
You set up a configuration space, and derive from it a hilbert space.
QFT was then just a model, where classical fields where taken as "source of inspiration", in the same way as NRQM was just a model, where newtonian mechanics was taken as a source of inspiration.
In what way this was "justified" was only to be determined by empirical success or failure.
The "miracle" for me was that this model of QFT generated the appearance of particle-like things, all by itself. Moreover, these particle-like things, in the right limits, behave as particles in NRQM. Is this a coincidence, or something deep ?

In another limit of QFT, of course, we find back our "source of inspiration", namely the classical field (in the same way as we find back something that looks like newtonian mechanics in NRQM). The only field for which this really works out are massless boson fields (EM). But this is much less of a surprise, because we put it in from the start. The true "miracle" is that particles come out of the "quantized classical field" model.

Now, Weinberg turns things around, and asks: what kind of model can spit out particle-like stuff and be in agreement with relativity ? And he finds that such a model must look like a quantized classical field. So the miracle here is that the field comes out. We put in "particles" by hand, plus relativity, and we get out "fields".
So what came first ? The field, or the particles ? Weinberg shows that the notions are equivalent.

Last edited: Jul 3, 2006
7. Jul 3, 2006

vanesch

Staff Emeritus
A book that does this much better than any "high energy physics" QFT book is "optical coherence and quantum optics" by Mandel and Wolf.

8. Jul 3, 2006

nrqed

hi Patrick... I have enjoyed very much all the exchanges we have had over this issue

I see what you are saying but what I find annoying is the following. Ok, we have learned about this mysterious theory of QM where we start from classical hamiltonian and replace X and P by operators and so on.

And now QFT is introduced. The "field theory" way of approaching things is to again impose strange commutation relations (which are *different* than the already strange CR of QM and are just justified by analogy) now defined on even weirder things than classical momenta and positions: they are defined on classical fields which are unobservable to start with (except for the EM case)!!

So it's a different type of quantization, applied to new and strange uobservable "classical" fields. :yuck:

So there are two leaps of faith here: accepting those new CRs and then accepting those weird classical fields as a starting point.

However, the "Weinberg approach" in my view is *so* much more physical and intuitive. The only thing that is needed here is to accept special relativity. That because of E=mc^2, the number of particles may vary and that equations must be Lorentz covariant, etc. *This* is much more intuitive to me than the field approach!!

There are additional bonuses from this approach, too. First, the field operator Phi (for a scalar case, let's say) is clearly secondary in Weinberg's approach. It is just a way to combine creation and annihilation operators with a certain Lorentz symmetry. It is never thought of as an observable. In the standard field theory approach, because one *starts* with that field, it feels like it is a key physical operator or something. It's annoying that it is the central quantity in the field approach and yet nobody says at the get go that this is not an observable, that we will never care about its eigenvalues and so on. In the Weinberg approach, it clearly has a secondary role. For example, Weinberg applies that the *Hamiltonian* at spacelike points commute and derive conditions from there. The standard approach is to impose this on the field Phi even though it's not an observable (and it is left unsaid that this is a sufficient condition because physical observables will be built out of fields Phi).

So to me it is annoying and confusing to put the field Phi at the forefront when in fact it is not even an observable.

Another point is that the usual field approach gives the impression that the CRs of the fields is similar to the CRs of X and P in QM. But actually, the CRs of the fields have nothing whatsoever to do with position vs momentum. They arise because of the varying number of particles and the CRs of the annihilation/creation operators. I do know that one can turn the tables around and see the annihilation/creation operators arising from the CRs of the field (and the conjugate momentum) but these CRs are totally unrelated to a Heisenberg-like uncertainty in position vs momentum of a quantum particle. So it seems to me that the analogy used to get the CRs of quantum fields from the CRs of QM obfuscates their meaning.

I have to admit that I, on the other hand, *did* feel lousy about the field approach .
I can see that they are equivalent, but the field approach to me has much more the feel of a "trick" to organize things in a very efficient way, whereas the Weinberg approach makes much more sense physically. There might be something deeper, physically, about the field approach but I haven't seen it.

The only situation in which I *would* start with the field approach would be in Condensed Matter type of situations, to explain phonons, say. Then it makes sense to me to do it that way.

For me it was an annoyance to wonder about this question (miracle or not)!!! It always felt like a miracle to me (before I read Weinberg), and I was wondering what was new in the CRs of the classical fields as opposed to the CRs of QM. What annoyed me was that nobody was stating cleraly :"This is a miracle, there is something deep here"!! OR "There is nothing deep here, and here's the reason why". I have been so annoyed for years because I wanted to find out which it was!!
Very true. But pedagogically speaking, I find that it makes much more sense to accept that there are quantized packets of energy called mesons and electrons, etc and that because of relativity their number may vary and to work from that than to start from inexistent classical fields and to quantize them using new CRs that are inspired by analogy with QM.

I guess that if I was teaching QFT now, I would never be able to get myself to try to get the students to swallow the field theory approach (I mean, I am pretty sure that most students *would* be willing to accept it and by the time we would be doing applications, they would have forgotten about their initial misgivings, which is I think what happens in almost every QFT class...But as the instructor I would feel very uneasy about that approach.

But hey, you know that this is a pet peeve of mine:tongue2:

Last edited: Jul 3, 2006
9. Jul 3, 2006

Eye_in_the_Sky

WOW !!!

Initially, I felt I may have some insights to share in this thread. But now it seems that any of my current thoughts on these matters can at best be not much more than trivial to the two of you Pats.

I do have some questions, though.

First of all, I would like to verify whether or not I have sufficiently comprehended Pat's (that is, nrqed's) grievances against the "field" approach and also whether or not we see eye to eye on some of the basic physics and math. Please bear with me.

Consider a world which respects Galilean invariance and in which the Schrödinger equation is exact. Then, Pat (nrqed), am I correct in the following assertions?

a) You would have no complaints against construing the Schrödinger field as a 'classical' field, since it does have physical meaning in terms of probability amplitudes.

b) Nevertheless, you would harbor complaints against its direct quantization as a field, because a quantized probability-amplitude makes no physical sense.

c) You would not assert that the CRs of the fields have nothing to do with CRs of X and P, because X and P can be 'fashioned' from the field operator.

10. Jul 3, 2006

nrqed

Just a few thoughts that crossed my mind while driving today. I am basically repeating myself so everryone is welcome to ignore this post but it will help *me* clarify my point of view.

If I were to compare side by side the "field" approach vs Weinberg's approach, I would say the following:

In the field approach, the starting point is the introduction of mysterious classical fields that have never been observed (except the EM field but again, I am talking about the usual presentation where the EM field is discussed much later because of all its complications) and the introduction of mysterious CRs which are only justified by analogy with QM and are applied to fields which are not observables in the first place!

(And again, the CRs of the fields and corresponding conjugate momenta have nothing to do, in the end, with commutation relations with any sort of position and momentum operators, they arise from the CRs of the annihilation and destruction operators! So the fact that they are "justified" using an analogy with the CRs of QM, is a huge cheat, IMNSHO:tongue2: )

So in the field approach, in the foreground are those mysterious fields and those mysterious CRs.

*Then*, one obtains as a side result that there are quantized packest of energy and momenta. And that their number is not conserved.

In Weinberg's approach, however, what is placed in the foreground is that a) there are things which act like discrete bundles of energy and momentum (like electrons ) and that b) from E=mc^2 one expects their number not to be conserved.

Now, by analogy with the harmonic oscillator formalism one introduces annihilation and creation operators, one builds operators, one imposes causality, Lorentz covariance, etc etc...And everything comes out working nicely, with quantum fields now a *byproduct* of the whole thing. Two huge bonuses (IMHO): first, no need to even *ever* introduce the weird classical fields that are the starting point of the field approach! They are gone! Second bonus: no need to postulate the weird CRs between those weird quantum fields and their conjugate momentum!
They *follow* from the CRs of the annihilation/creation operators and *their* CRs are obvious to justify since they must raise or lower the energy by an amount equal to the energy of a particle.

Now, I do understand that the field approach is an efficient way to organize things and to build in Lorentz covariance, etc. But as way to *teach* the subject, I find that they are incredibly obscure and confusing. Well, maybe it's because I am not too smart but they sure confused me for years and years until I finally came to peace with them because of Weinberg (whom I could never thank enough for writing this book!)

Anyway, sorry for the babbling.... Had to get this off my chest

Pat

11. Jul 3, 2006

nrqed

Don't feel that way!!!
It is very useful to me to hear comments from others (especially if they are starting to learn QFT) because it gives me much needed perspective. I like to know how other people think, what they find natural, etc. I learn a lot from that.

I think I would have a problem already at this level. Because a probability wave is already in itself completely different from any classical wave (for example the E or B field in classical E&M).

But I see your point so let's say that we are ok with treating this as a "classical" wave.

Yes, that would definitely bother the heck out of me!

But not only because of the quantization of a *probability* amplitude. The quantization of a classical field in itself bothers me because it is an additional "leap of faith", on top of the usual things we have to make when doing QM (the existence and role of the wavefunction, the replacement of classical quantities by operators, measurements, etc). Now we have to accept a new type of commutation relations, on top of those from NQRM.

( ASIDE: I know that everybody says that "second quantization" is a misleading term, but I personally think that it's a perfectly appropriate expression in the context of the usual presentations on QFT.
(in one type of presentation, people say that one is rewriting the wavefunction as an operator..then it *is* litterally a second quantization (i.e. a quantization on *top* of another quantization). On the other hand, when one quantizes classical fields, the expression "second quantization" is still appropriate because the quantization used there is totally different in nature from the QM quantization, so it's a second quantization in the sense now of "a different type of quantization").)

I am not sure what you mean. I think we would have to be more explicit to discuss this. I don't know what you have in mind by saying that they can be "fashioned: from the field operator. In NRQM, one cannot write X or P in terms of $\psi$.

In NRQM, one promotes P_x and X to operators. So in the case of this "wave", one would expect to impose commutation relations on $p_x \psi$ and $x \psi$. That's the only way in which I could see a "natural" generalization to impose commutation relations on. However, if you write the Lagrangian leading to Schrodinger equation, the quantities unto which one imposes the QFT commutation relations are obviously not the above expressions! Far from it!

Thanks for the interesting discussion btw!!

Regards
Pat

Last edited: Jul 3, 2006
12. Jul 3, 2006

Eye_in_the_Sky

From your response to item a), I see that for you a necessary feature of a "classical" field is:

The field amplitude can, under appropriate conditions, be measured.

This condition is too strong for my tastes. I prefer the much weaker condition:

The field amplitude exists.

"What? There are things which exist but cannot be measured?"

Well, not exactly ... but maybe.

It could be that I have never been able to create the appropriate conditions. Or it could be that I have not yet found the right device which couples to it. ... Or maybe it just cannot be measured. Or maybe it exist only as a representation within space of the influence of something which resides outside of space.

Pardon my voodoo ... but I'm serious!

Perhaps it exists as an agent which contributes to phenomena that I can measure, yet it itself cannot be measured directly.
__________________

Next ... my item b), it turns out, wasn't quite the right probe I was looking for. I'll have to think about that one some more and see if I can hone it down.

The larger part of your reply, though, concerning that additional "leap of faith" over and above a heap of such leaps (and again in the ASIDE that that additional leap renders appropriate an expression which by your estimation was already fine, but not by mine) ... all of that, I would like to put aside until after we part with Galileo for something relativity more intricate.
__________________

Regarding c), now I'll be more explicit. Quantization of the Schrödinger field leads to a field operator Ψ(x,t) such that the object

Ψ(x,0)|0>

is in fact the ket |x> on the single-particle subspace, so that

∫xΨ(x,0)|0><0|Ψ(x,0)dx

is the single-particle position operator X. Since we have |x>, we can get |p> by means of a Fourier-type transformation, and with that construct P, conjugate to X.

This is what I meant when I wrote X and P can be 'fashioned' from the field operator.

So, item c) amounts to: "Do you accept the above to be true? And if so, would this be this be an instance in which you are inclined to retract your statement that the CRs of the fields have nothing to do with the CRs of X and P?"

13. Jul 4, 2006

masudr

um, I'm like a 2nd year undergraduate, so I probably don't have a clue what I'm going on about. But here I go anyway

nrqed said that the CRs of NRQM were strange enough, let alone the CRs of QFT. But aren't they just they standard "quantization" procedure (which is admittedly mysterious, but one can accept) applying to the Poisson brackets of Hamiltonian mechanics? And couldn't one show the same Poisson brackets for fields and conjugate momenta of classical fields? And then apply the same (admittedly mysterious) quantization procedure?

My point is that probably the CCRs for NRQM are just as mysterious as those for QFT, not any more mysterious or any less.

14. Jul 4, 2006

nrqed

Hi Masudr. Thanks for your input!

I will try to clarify my position (of course this is all an issue of perosnal taste and is very subjective. I can say that I don't find something nature=al or pedagogically sound and if someone else disagrees, that's the end of the discussion. So I am not trying to prove anything or to convince anyone of anything, I am just expressing my opinions. But I am learning a lot by seeing how others think of those things and that's why I like to have those discussions).

Two points.

A) In QM we apply the quantization rules to position and momenta of point particles, quantities which have a clear classical meaning and which we are used to work with classically. However in QFT, we apply quantization to "classical fields" that we have to postulate from the start and which have never been observed at all (like the Dirac field and so on). The only exception is the EM field, but there is no such thing as observations of classical scalar of spin 1/2 fields. So we have to "ivent" those "classical" fields before quantizing. And that's one of the two things that "bothers" me. In the "Weinberg approach", the quantum fields are a *byproduct* of the derivation and there is no need to even introduce classical fields which is great from my point of view since they have never been observed (again, except for the EM field).

B) I agree with you that one may take the position that the key idea about quantization is to replace the classical Poisson brackets by quantum CRs. However, I still see this step applied to fields as being distinct to this step applied to point particles. One can argue by "analogy" to "justify" the CR of fields from the CRs of point particles, but in the end, there is *still* a leap involved there. The CRs for the fields do not mathematically follow from the CRs for point particles. They are *analogue* but one cannot *mathematically* go from the CRs of point particles to the CRs of the fields (the only case where this can be done is if we seriously think of the field as being made of a large number of particles, i.e. a physical string. But the waves that we quantize in particle physics are not of that nature) . I agree that the analogy is strong but it is still a *new* quantization procedure that must be taken as an extra axiom on top of the axiom for the CRs of point particles. So in that sense, this is *really* a "second quantization". It's the quantization applied to fields as opposed to applied to point particles of NRQM.

There is nothing wrong with adding a new axiom to QM, this is not what I am saying. But my point is first that it sohould be recognized as such: a new axiom. Despite analogies with point particles, it is still a new axiom.
My second point is that in Weinberg's approach, those CRs on the *fields* are not postulated at all! They are a byproduct! Where do they come from? The staring point is the assumption that the number of particles is not constant (because of mc^2). Then it is natural to introduce annihilation and creation operators which then have an obvious set of CRs. So, in the field approach, there is an axiom giving the CRs of "classical fields" (which again, are not even observed) whereas in Weinberg's approach this axiom is replaced by the "axiom" that particle number is not conserved. I personally much prefer the second approach!:tongue2:

Does this make sense to you? I would appreciate yoru feedback!

Thanks again for you comment.

Regards

Patrick

15. Jul 4, 2006

masudr

I must disagree. What is the classical notion of spin? You could tell me that infinitesimal rotations don't commute, but that's just as bad (for me, at least) as saying that we will quantize a spin-1/2 field.

Interesting... the reason it doesn't bother me is because I see it like this. We have, the rules of QM. Then we have different Hamiltonians. So, we have a classical field we know: EM. We apply QM to it, and we get all these results. Then we can have different Lagrangians (e.g. the Dirac field Lagrangian etc.)

In all fairness, I haven't seen the Weinberg approach, and it does sound quite cool. I shall remember to give his text a look when I study QFT in my 4th year.

There's no analogy involved here. We have, say, the simple harmonic oscillator. We apply the rules of QM to it, i.e. represent states by those that satisfy the TDSE, observables by linear operators that obey the CRs etc. We now have a field with Lagrangian given by some expression. We apply QFT to it, so we have these position valued operators, etc. etc.

So we have Lagrangians and Hamiltonians for fields, and we quantize the fields just like we quantize particles. Your objection is similar to objecting that the rules that apply for classical particles shouldn't apply to classical fields, unless I have missed something.

I would agree with this.

But even the classical EM field isn't particles oscillating on a string. The EM field is something quite mysterious as it is. We can get the field equations for a string by saying classical field theory is the study of continuous/infinite classical particles. But we can't really say that for the EM field. Again, what we do is set up the formalism, and then apply the rules to a brand new Lagrangian, and find that it works.

Is it really applied on top of the rest of the axioms? We impose the CCRs for the field operators, but dont the ones for the particles of the field emerge from that? Or do we separately impose that too?

As you have said, it is all a matter of taste and opinion, and of course, that is fair enough. I don't happen to have that much of a problem with the "standard" approach, although I must say, I haven't looked at the standard approach in much detail, nor the Weinberg approach, so I'm not really in a position to comment.

No worries.

Masud.

16. Jul 5, 2006

Eye_in_the_Sky

Whoops !!!

Nowhere in the above have I invoked the CRs of the fields to say what I say about the particle. According what is written there, the single-particle X operator will mathematically follow from no more than the existence of Ψ(x,0). And then, constructing P from X, well again that too has nothing to do with the CRs of the fields.

Besides, the CRs of the fields are connected to spin, whereas those of particle have no such connection.

Yes, yes. I see it now: an ADDITIONAL LEAP OF FAITH, of course(!), is involved when one invokes the CRs of the fields.

In NRQM, invoking the CRs (or anti-CRs) of the fields is the same as POSTULATING symmetric (or antisymmetric) subspaces for the many-Boson (or many-Fermion) system.

In hindsight, this is painfully obvious.

... Whoops !!!

17. Jul 5, 2006

nrqed

Hi Masudr.

Again, this is all quite subjective and a matter of taste, to a large extent. My arguments are more about pedagogy in teaching the subject (and writing textbooks). What is more "logical" as a way to introduce the subject. As a student I was bewildered by the standard presentation and did not know what was "profound", what was supposed to be analogies vs derivations, what were new axioms vs derived results, etc. And I was bewildered by the starting point of it all. To be honest, I could never get myself to teach the subject the standard way if I were to teach this class tomorrow.

Again, here's briefly my objection:

In the standard field approach, one starts with those unobserved classical field theories (like the KG field or Dirac field). One then quantizes them by postulating the CRs for fields. The presence of particle-like excitations comes out as a byproduct. (By the way, I might feel much better about this approach if they would also apply this approach to *noncovariant* field theories and show what happens then. And show clearly what is the difference and why there is a difference with applying this technique to invariant (scalar) vs non-invariant Lagrangian densities.)

(Of course, the EM field case is different because we do observe E and B fields classically. But then textbooks should start with EM fields, quantize them, show clearly the connection with classical fields through coherent states, the connection with the classical amplitude and phase and on and on. Then, *after* a solid intuition has been built on the connection between classical field theories, quantum fields and particle excitations, books should discuss why coherent states of massive particles are not observed, and so on. And *then* it would make sense to go on to quantizing KG and Dirac fields. But things are never presented this way On the other hand, Weinberg's approach bypasses all the pedagogical difficulties of the field approach, IMHO)

In Weinberg's approach, one starts with the idea that particles can be created or annihilated. Then one has many-body theories. If in addition one imposes Lorentz invariance, one is forced to introduce quantum fields.
The fields and their CRs are a *byproduct*. No need to postulate strange classical fields to quantize. No need to postulate CRs on the fields. The difference between Lorentz invariant and non-invariant theories is clear.

I find this second approach much more satisfying. Seems to be so much more logical than the field approach. And a *much* better way to *teach* the subject. Of course, after all is absorbed and the field connection is made, then the field approach should be shown too.

To be honest, I am still not totally sure whether there is something "deep" about the field approach. I still have to understand exactly how the CR between the fields and their conjugate momenta arise out of Weinberg's approach vs how they arise from the usual Lagrange/Hamiltonian approach.

I did not mean to say that everything we quantize in NQRM has a clear classical analogue. I agree with you that it's not the case. I was focusing on the fundamental CR, between X and P. My point is that we impose CRs on these quantities. And then when we get to a KG field or a Dirac field, we impose a *new* set of CRs on the field and conjugate field momenta, which are themselves quantity unobservable classically (and this is a key difference even with the spin case. Spin has no classical analogue, but you get a Stern-Gerlach apparatus and you can measure it. If books want to start with a classical KG or Dirac field, theyshould explain how one would go about measuring their phase or amplitude )

On the other hand, Weinberg's approach does not require to introduce those unobservables classical fields and does not require to postulate their CRs. What *is* postulated is the need for a formalism with a vraying number of particles. The CRs are the CRs of annihilation/creation operators which are quite natural from the example of the harmonic oscillator where the energy levels different by integer multiple of an energy quantum.

I understand. But QFT books never show clearly the correspondence between the quantum EM fields and the classical fields we all know and love (not just showing that photons quanta arise..but showing what's the connection with classical phase and amplitude, talking about measurements of E and B fields, etc). If the field approach is to be taken as a starting point, it seems to me that the first chapter of any QFT book should be an in-depth discussion of EM fields. But QFt books sometimes don't even mention coherent states:surprised Or why we don't observe classical limits of the KG or Dirac field, etc.

I highly recommend doing so. Unfortunately, its is very dense so it takes a lot of work to see the basic ideas through all the notation. If I would teach QFT tomorrow, I would unfortunately not feel that I could use that as a textbook. I wish there was a lower level book on QFT that would introduce the ideas this way.

You agreed that the CRs of the quantum fields cannot be derived from the CRs of point particles. My point is that the CRs of fields must be seen as a new axiom. Sure, the basic principle is similar to what we do for point particles so it's easy to accept/ But it is still a new axiom. And on top of that it is applied to fields which are not even observed classically.

So the field approach says : look, we have all those particles around (electrons, mesons, etc). To do calculations, we will introduce those "classical fields" (that we have never observed) and we will quantize them, based on an analogy with point particles NRQM and assuming that it makes sense to take the continuum limit for these unobservable fields. To *me* this seems like voodoo. And my definition of voodoo is this: I would personally would have never thought about doing this if someone had not told me that this was the way to go.

Sure, you will say that it's a natural way to go given EM fields. But then why not work out in details the EM fields and the quantum field/classical field correspondence limit in great depth before jumping to those abstract KG and Dirac fields?

On the other hand, Weinberg's approach says: look, we have those particles around. Because of relativity, we expect particle number to change. Let's go from there. That's it

I agree!!! The point that I had made is that strictly speaking, the correspondence between the CRs of point particles and of QFTs can only be made explicit (i.e. one can derive one from the other) in the case of a vibrating string. I agree completely that even in the EM case, there is a "leap of faith" involved in defining the CRs. In the sense that it's a new axiom. And the correspondence quantum field/classical field is far from obvious and trivial in the EM case.

So my point is that if one is serious about presenting the field apporach, one shoudl devote some time discussing the EM field in depth.

On the other hand, Weinberg's approach does not require to postulate those CRs for quantum fields as a starting point. One only need to introduce creation/annihilation operators. And one can build any field theory (KG, Dirac, EM) from this principle. I personally find this much more satisfying.

As far as I can tell, they are completely separate (although another poster suggested an idea that I still have to look at).
If you (or anyone) could show me how to start from the field CRs and recover the position-momentum CRs for a single particle, I would be very interested. The CRs of the field arise from the annihilation/destruction operators CRs, which don't say anything about the X,P CRs of a single particle state.

This is one of my main arguments. One talks about the field and its conjugate momentum and use an analogy with the position and momentum of a single particle in order to "justify" the CRs for the fields. But it seems to me to be very decieving, because in the end, the CRs of the field and conjugate field momentum have nothing to do with position/momentum CRs. They are connected instead to the varying number of particles. So it feels like a cheat to me. It uses an analogy to get the right CR but later one realizes that the analogy has no basis. This is one of the main things that bothers me about the usual approach!
(again, in the case of an actual string, then there really is a connection between the CRs of the position and momenta of each particle and the CRs of the field, but as you pointed out, even EM is not a quantized string!)

Sometimes I feel like an alien:tongue2:
Because from the moment I first read about QFT I was bothered about what appeared to me to be huge leaps of faith with no logical basis. I do know that some leaps of faith are required in QM but at least see how experiments and observations led to those leaps fo faith. Then I would read about QFT and the very first thing they would say is "relativistic wavefunction euqations have problems. So "obviously" what we will do is to consider classical fields and quantize them:surprised :surprised It's the biggest non-sequitur that I have seen in physics . I mean, special relativity involves a weird leap of faith, but at least I could see why it made sense to make this leap. I surely would never had been able to do it myself, but after I learned the theory, it did make sense to me. Same thing for GR. But the way textbooks present QFT, it did not make any sense to me.

I understand that historically, quantization of the EM field played a major role. But then why don't textbooks work this out in details first?

But again, now that I have read Weinberg, it makes even more sense to me to start this way (with the idea of varying number of particles) and to build from there. No non-sequitur involved there.

Again, just a question of taste.

egards

patrick

Last edited: Jul 5, 2006
18. Jul 5, 2006

nrqed

Right!!
well, I see what you mean. Whether we have CR or anti-CR depend on the spin, yes. But I would say that the Crs of the fields are connected to the *varying number of particles*. In weinberg's approach , one *starts* for the need for annihilation and creation operators and then one *derives* the CRs for the fields. I find it so much more satisfying to start from the need to have varying number of particles (because of relativity) than to have to postulate those weird unobserved classical fields and to postuale field CRs. But apparently, I am alone feeling this way (with Weinberg)

Right! In Weinberg's approach, there *is* a leap of faith, but that's the need for varying number of particles. I would have much rather learned QFT this way than to have to accept as a starting point those weird classical fields and then to accept the field CRs as a new axiom.

Right. But again, I think you are focusing on the distinction between CRs vs anti-CRs. But even if we consider only bosonic excitations, the question can still be asked: what is the principle behind the need to invoke CRs of the fields and their conjugate momenta? The answer is the need to have a *many-body* system. It has nothing to do with position/momentum CR of a point particle.

And this is one of my pet peeves. One invokes the analogy with the position/momentum CR of NRQM to justify the axiom of the CRs of quantum fields. But later one realizes that the CRs of the quantum fields have nothing to do with position/momentum CRs. They have to do with varying number of particles. So it feels like the analogy used to postulate the CRs in the first place is a cheat.

(to be honest, I am still wondering if there is something deep going on here that I am missing. On one hand the CRs can be derived simply by considering varying number of particles. On the other hand, they can be derived by assuming a continuum analogue of position/momentum CRs on fields. Is the fact that the two approaches lead to the same result "deep" or not? I am still unclear about this. And I really want to understand this because I feel that once I can answer this to my satisfaction, no matter what the answer is, I will finally understand QFT at more than a very shallow level)

I hope this helps make my point of view sound a bit less crazy :tongue2:

Regards

Patrick

19. Jul 7, 2006

nrqed

Thank you for the suggestion, Patrick. I will try to get my hands on a copy. I always found it so strange that QFT books don't cover the crucial topic of the correspondence between the quanized EM fields and classical EM. This seems to me to be one way to make the "quantization of classical fields" less of a huge non sequitur.

Thanks again.

Pat

20. Jul 7, 2006

masudr

Hello again (after a bit of a break)

Well, we don't impose a new set of CCRs: it's really the old set. But this time we are applying them to a new system.

That is:
• In NRQM we quantize hypothetical (or not) classical particle systems with any Hamiltonian. So, we upgrade observables to operators, and upgrade the Poisson brackets of position and conjugate momenta (from Hamiltonian particle mechanics) to commutators on those operators.
• In QFT, we quantize hypothetical (or not) classical fields with any Hamiltonian. So, we upgrade observables to operators, and upgrade the Poisson brackets of position and conjugate momenta (from Hamiltonian field mechanics) to commutators on those operators.

So it's not an additional imposition: it's just the standard quantization procedure (which I assume you are happy with for particles). Luckily for us, the Hamiltonian for some fields ends up looking like the Hamiltonian for an oscillator...

Masud