Talk:Spinor/Archive 3
![]() | This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page. |
Archive 1 | Archive 2 | Archive 3 | Archive 4 | Archive 5 | → | Archive 7 |
Non-references
I have moved the following references out of the article, since they are not actually referred to in the text, nor do they seem to have used by the original editor in writing the article. If there is a sense that either of these is a good general reference on spinors, then perhaps one or both can be added back to the article. But I don't really get the feeling that this is the case. Sławomir Biały (talk) 00:33, 27 August 2010 (UTC)
- Benn, I. M.; Tucker, R. W. (1988), An introduction to spinors and geometry with applications in physics, Institute of Physics Publishing, ISBN 0852741693.
- O'Donnell, Peter (2003), Introduction to 2-spinors in general relativity, World Scientific, ISBN 981-238-307-7.
An IP editor continues to insist that this reference be added to the article. In addition, he or she has removed references that are actually referred to in the article. Ironically, whilst the editor dismissively referred to Hitchin's classic paper "Harmonic spinors" as "unimportant", it gets three times as many citations as the Benn book, and more than 50 times more citations as the O'Donnell book. I'm beginning to suspect very strongly that there is a conflict of interests at play here in promoting these books at the exclusion of other sources already referenced in the article. Sławomir Biały (talk) 10:23, 30 August 2010 (UTC)
- The IP editor continues to edit war against consensus, and without discussion, to include the Benn and Tucker reference. I have placed a COI warning on the editor's talk page. Sławomir Biały (talk) 02:40, 31 August 2010 (UTC)
Heuristic diagram/intuitive representations of a spinor?
Please forgive my ever-persuasive additions of geometric interpretations for all things not obvious to draw... Although for this article at least something would be nice and this may be it:
![](http://upload.wikimedia.org/wikipedia/commons/thumb/9/91/Spinor_interpretation.svg/200px-Spinor_interpretation.svg.png)
Top: an ordinary vector (yellow) always remains positive under any rotation,
bottom: a spinor (green) alternates between its positive and negative for each full rotation, heuristically depicted on the surface of a Möbius strip.
This illustration should not be taken too literally.It's not obvious to a reader how a vector can flip into it's negative when rotated by 2π rads, the Möbius strip shows that nicely but it may not be the actual geometric illustration (which is analogous for the interpretation of quantum angular momentum as "precessing cones of pseudovectors").
I can't find any reference for this illustration, but have seen it here and there on websites/google images (which may be unreliable)... Anyway what do others think? I'm not forcing this in...
There are other interpretations by
- Penrose in his The Road to Reality of the "book and belt" p.204-206,
- MTW in their Gravitation give two versions:
- a "can tied to elastic bands" on p.1149 (with videos of this on youtube, for example here), similar to the book and belt,
- a really confusing one of a "laser as a flagpole shooting an oriented-entangled flag" on p.1157 ..... (no clue!)...
- here (a nice one on hand palms as spinors!),
(not saying the youtube videos would be suitable sources/refs for the article but possibly external links?) Maschen (talk) 15:45, 4 September 2012 (UTC)
- I'm afraid the Möbius strip analogy is too much of a stretch for me; it get too many things wrong. In particular, it does not generalize to rotations in three dimensions, whereas the book-and-belt does. The book-and-belt (or rotating palm) also gets it wrong, but only in the interpretation. There the spinoral object is actually the rotation, not the book (palm), and the twist of the belt (arm) simply tracks the topological aspect of the rotation, such that a 4π rotation on any axis is needed to be topologically indistinguishable from no rotation. How this ties in with "real" spinoral objects (fermions) is beyond me, but the mathematical similarity of a wavefunction sandwiching an observable to a rotation sandwiching a vector is clear. — Quondum 20:39, 4 September 2012 (UTC)
- Ok... never mind then... Maschen (talk) 21:30, 4 September 2012 (UTC)
It refers as a "Dirac spinor" to the thing described in the bispinor article. We must alter either this wording, or the article linked. Incnis Mrsi (talk) 18:06, 6 January 2013 (UTC)
- I think so too. At least after deciphering what is meant by "an element of a representation" at will.
- This quote (article); "No one fully understands spinors. Their algebra is formally understood but their general significance is mysterious. In some sense they describe the “square root” of geometry and, just as understanding the square root of −1 took centuries, the same might be true of spinors.", is good.
- There is another one by Feynman: "If we can't explain spinors to our students, it means that we don't understand them ourselves." Even better.
- I still contend that spinors don't necessarily live inside a Clifford algebra (and even less being "representations of the double cover" (whatever that means, see below)), and that they are functions of space-time (from most physicists point of view).
- I'll try to be constructive: "Elements of a vector space V equipped with a projective representation of a Lie group ..." Now differentiate clearly between spinor and spinor wave function. Well, that's a start. To proceed, one probably needs to factorize (tensor product) the spinor into "elements of the range of space-time functions" (e.g. 4-touples), and some reasonable infinite-dimensional function space of functions of space-time (e.g L2 on R4).
- If it insisted that some other (non-)definition is kept, then please define basic terms. The term "representation" is much overused, and can be used to abbreviate what is really meant, especially when the author doesn't feel like spelling things out. In some cases, this means that the author really doesn't know himself. There are examples of this in the literature, where obscure descriptions have been copied from source to source, and even from century to century. Read for instance this; Lorentz group#The Lie algebra of the Lorentz group. (so(3;1), or so(1;3), if you want, is simple, and cannot as such be decomposed as a direct sum of other algebras). Somebody said , including in present WP articles. This sort of thing is not true in any strict sense, yet is has been, and will be forever, around, because it is in the sources (see e.g. representation theory of the Lorentz group). (A related misconception is that the finite dimensional representations of the Lorentz group are, or can be made unitary.) For those of you who are mathematically inclined, there is a lot to do. Define spinor and come to terms with the physicist. YohanN7 (talk) 19:18, 8 January 2013 (UTC)
- Spinors (or spinor fields if you prefer) are more than just functions, because to be spinors they have to possess particular transformation properties. That is essentially what makes a spinor a spinor rather than something else. Locating them in a Clifford Algebra gives a very good way to make those transformation requirements intuitive... unfortunately I've personally never quite got to that level of Zen mastery yet. Jheald (talk) 20:00, 8 January 2013 (UTC)
- YohanN7, do you copy? Dirac spinor is a bispinor-valued wave function! If we understand spinors as Weyl spinors (), then a "spinor wave function" is a Majorana fermion, not a "Dirac fermion". Incnis Mrsi (talk) 20:06, 8 January 2013 (UTC)
- It is probably best to define physics spinors using projective representations. The problem with Clifford algebras in this respect is that they house representations that are not projective. They are perfectly normal scalars, tectors, tensors,... . The gammas, for instance, are true "vectors" when it comes to transformation properties, while spinors are not, i.e. they do not transform under the rule s → Λs. A physics definition of spinor (fields) using the concept of projective representations would capture the special cases too. The connection between the gammas and the would-be-spinors is the Feynman dagger. YohanN7 (talk) 14:40, 10 January 2013 (UTC)
- The present article on the gamma matrices does in fact, in the lead, use the above mentioned for characterizing a (physics) spinor. YohanN7 (talk) 15:28, 10 January 2013 (UTC)
- It is trivial to identify the columns of the gamma matrices with subspaces of the Clifford algebra. Note that the Clifford algebra does not just house "normal scalars, vectors, tensors, ...", it also contains linear combinations of these things as first class objects. So spinors are contained within the Clifford algebra (in which both two-sided and one-sided transformations are well defined). The fact that Clifford algebra gives a single environment which included vectors, spin matrices, spinors etc as first-class objects is why it has become such a dominant setting in which to talk about such things. Spinors are curious things though, because while it's quite easy to see (simply through its multiplication properties) how a linear combination of a scalar and a bivector can take over all the rotational-operator properties of a spin matrix, it is not quite so clear (at last not to me, at least not yet) why a linear combination of a scalar and a vector, which is the CA subspace that the spinors map to, should be so fundamental an object -- can one explain in CA terms what is it that makes this the vital canary in the coalmine? Jheald (talk) 22:33, 10 January 2013 (UTC)
- Sorry I do not understand : scalar + vector CA subspace for spinors ? Where do you get that? Chessfan (talk) 10:27, 11 January 2013 (UTC)
- Sorry, you're right. The spinor subspace of the whole CA is going in fact looks something like where x is a general element of the algebra, and is an idempotent projector (with the projector getting perhaps a lot more complicated than that the bigger the space). That projector maps an x which is a scalar to a scalar plus a vector, but you are quite right, other elements of the algebra will get mapped to build up a subspace which in general will be more complicated, so e.g. Cl3,0(R) gets mapped down to Cl2,0(R) (1 + e3) I guess I'm still striving as to what in CA terms the significance of this projector is (i.e. can we talk about it in terms other than it annulling some columns of a matrix representation), and what the meaning of sandwiching by eg a scalar+vector, as opposed to an all-even scalar+bivector. Jheald (talk) 11:43, 11 January 2013 (UTC)
- Sorry I do not understand : scalar + vector CA subspace for spinors ? Where do you get that? Chessfan (talk) 10:27, 11 January 2013 (UTC)
- It is trivial to identify the columns of the gamma matrices with subspaces of the Clifford algebra. Note that the Clifford algebra does not just house "normal scalars, vectors, tensors, ...", it also contains linear combinations of these things as first class objects. So spinors are contained within the Clifford algebra (in which both two-sided and one-sided transformations are well defined). The fact that Clifford algebra gives a single environment which included vectors, spin matrices, spinors etc as first-class objects is why it has become such a dominant setting in which to talk about such things. Spinors are curious things though, because while it's quite easy to see (simply through its multiplication properties) how a linear combination of a scalar and a bivector can take over all the rotational-operator properties of a spin matrix, it is not quite so clear (at last not to me, at least not yet) why a linear combination of a scalar and a vector, which is the CA subspace that the spinors map to, should be so fundamental an object -- can one explain in CA terms what is it that makes this the vital canary in the coalmine? Jheald (talk) 22:33, 10 January 2013 (UTC)
- FYI regarding spinor / bispinor, I have now tagged the bispinor article for re-write (reason: having a self-referential or lacking definition). It would be good to consider that article too once there is consensus on the spinor as such. (Discussion about that specific topic under Talk:Bispinor#Cyclic definition.) --Chris Howard (talk) 07:26, 11 January 2013 (UTC)
- Clifford algebras are probably the primary sources for producing spinor representations. Any definition of what a physics spinor is fine, but my point is that there is probably a course at the graduate level in difference in prerequisites for understanding the various definitions. Treatments on QM at best mention Clifford algebras. Projective representations (while still being far from a trivial concept) are easier to understand. Such a description is, I dare say, the description most readers would expect to find. By contrast, requiring a spinor to live inside a Clifford algebra complicates things. "Natural" spinors (electron wave functions) are no longer spinors. YohanN7 (talk) 14:24, 11 January 2013 (UTC)
Physical and mathematical angles
I find a lot of this article pretty obscure, which is all right, as it's an advanced topic. However, I wonder if the topic of spinors as used in physics could be expressed in a more simple way. This article treats all forms of spinors, in arbitrary numbers of dimensions and relates it to mathematical topics like representation theory and algebra. Take for example this from the "Terminology in physics" section:
- "The most typical type of spinor, the Dirac spinor,[7] is an element of the fundamental representation of the complexified Clifford algebra Cℓ(p, q), into which the spin group Spin(p, q) may be embedded. On a 2^k- or 2^(k+1)-dimensional space a Dirac spinor may be represented as a vector of 2^k complex numbers."
The most common space that will be considered in physics is 2^k dimensions with k=2, i.e. 4-d space-time. I question whether it's really worth discussing spinors in physics to this degree of generality straight up, when all most people will be interested in is the 4-d case used e.g. to describe the electron field.
Later in the paragraph we've got this: "In even dimensions, this representation is reducible when taken as a representation of Spin(p, q) and may be decomposed into two: the left-handed and right-handed Weyl spinor[8] representations." To understand this, you've got to understand what it means for a representation to be reducible. But even without knowing that, you could still appreciate being told that the Dirac 4-spinor (column of 4 complex numbers) can be split meaningfully into two 2-spinors (column of 2 complex numbers). Furthermore, the 2-spinors can be simply understood as representing directions, or points on a sphere. Furthermore, you could understand these 2-spinors as being acted on by elements of SU(2) (representing rotation) without knowing the definition of Clifford algebras or spin groups.
So I'd suggest that there could be ways of presenting this information without requiring the reader to understand more mathematical definitions than necessary. Count Truthstein (talk) 01:11, 26 December 2012 (UTC)
- "The most typical type of spinor, the Dirac spinor,[7] is an element of the fundamental representation of the complexified Clifford algebra Cℓ(p, q), into which the spin group Spin(p, q) may be embedded. On a 2^k- or 2^(k+1)-dimensional space a Dirac spinor may be represented as a vector of 2^k complex numbers."
- The really most typical spinor in the physics lingo is a thingie in a vector space on which the elements of the Clifford algebra act. This may well be an element of the Clifford algebra itself, that is true, but follow the link Dirac spinor for the typical intended meaning when a physicist says "spinor". Dirac spinors (spin 1/2) in this sense are foremost 4-valued functions of space-time (wave functions). Only when evaluated at a point, they become 4-tuples of numbers that will change under coordinate transformations (i.e. Lorentz transformations) in a peculiar way. In QFT, the thought of a spinor (now an operator on the space of the former thingies) living in the Clifford algebra comes closer, but the space-time dependence is still there. YohanN7 (talk) 03:29, 4 January 2013 (UTC)
- 2-spinors can be explained by a flagpole and flag plane (I think that's how they were originally conceived) which is quite geometrical and something which could be visualised. I am not sure to what extent that generalises up to other spinors. Francis Davey (talk) 08:45, 4 January 2013 (UTC)
The given examples (in geometric algebra) are completely misleading ; they should be removed or completely rewritten. It makes no sense to compare double-sided rotation operators sandwiching vectors , with single-sided geometric products between two spinors . Look for instance at the so-called spinor rotation of only 90° : that is absurd ! Products like psy phy appear in physics simultaneously with the conjugate quantities and have the significance of the composition of two successive doubl-sided rotation operators. Chessfan (talk) 15:56, 9 January 2013 (UTC) By the way I understand that a reverational bow in memory of Sir Hamilton is necessary, but to reduce geometric algebra to quaternions is a deep error and a very bad manner, like relegating it in a 19th century museum. Chessfan (talk) 16:16, 9 January 2013 (UTC)
- The real trouble with the 2D and 3d examples is that they are wrong. There are spinors which exist in Clifford algebras of 2 and 3 base dimensions, but they're more subtle creatures -- what is currently described in the article is not them. For which I apologise, because it was me that originally added this material to the article. In my defence, I would say that this is how some of the geometric algebra people use the word "spinor" (see e.g. quotes/refs in the second half in the second half of this 2010 discussion in the archives of this talk page), but as far as I can see their doing so is a misuse of the word, or certainly a different use to what "spinor" has traditionally meant.
- I think having a section introducing the actual (traditional) spinors for 2D & 3D base spaces would be a good thing, plus a discussion of what sort of objects they are in the relevant Clifford algebras. I did mean to eventually do that, once I understood enough about what actual spinors in these dimensionalities to have more of an idea what should replace it. But other things have since come to the fore, and I've never got to that point.
- See eg Spinors in three dimensions for actual 3D spinors, though I suspect this is probably not the way we should be wanting to present them. I think we should be looking to present what they mean as objects in a Clifford algebra, rather than simply as a column of complex numbers. I started some scratch notes on the 2D case and 3D case for my own benefit in 2011, but they only go so far, and I never got to the light at the end of the tunnel. Jheald (talk) 23:51, 9 January 2013 (UTC)
- Hi Jheald. I suspected you had something to do with the text under review ! Now we are back to our february 2010 discussion. I think one must make a clear distinction between the traditional, very abstract use of Clifford algebra (SO3, SU2 ,etc...) and the geometric algebra as developped by Hestenes and followers in quantum mechanics. Maybe spin and spinors are more subtle beasts in the main route, but even if that hurts the opinion of 98% of the physicists, one must say that the spinors defined in geometric algebra do the job. And last but not least they are much easier to understand and to explain. I have myself put a tentative vulgarisation article on that subject on my French Internet site, and amazingly that article appears since three years ahead of all others, better ones. Why ? Because it is understandable perhaps ... Chessfan (talk) 11:30, 10 January 2013 (UTC)Chessfan (talk) 21:56, 10 January 2013 (UTC)
- I will try to be a bit more explicit on the relationship between classical Clifford algebra spinors and spinors as defined in geometric algebra by the work of Hestenes and the Cambridge people (Doran & al). As those spinors operate in very different spaces and at different levels of abstraction , the ones with complex numbers and the others with real numbers, it seems indeed very difficult to explain the classical ones with the GA ones. But what we can try to show is what is the main source of the difficulties. Let us take a look at 3D spinors ; they can be written :
- where is the GA pseudoscalar.
- Call them quaternions if you want, but please dont reintroduce the "false" quaternion unit vectors vectors i,j,k ! What happens now when you transform the in Pauli matrices and the pseudoscalar into the pure imaginary number i ? By summing it up all and multiplying the (2,2) matrix by a column spinor (1,0) you get the classical Clifford spinor ! THat is you have transformed a perfectly well defined geometric object, that is a quaternion, into a dismembered corpus. No wonder nobody can recognize it and describe it in simple words. Of course the same is true, and worse with Dirac's biquaternion.
- What would have happened if Elie Cartan and later Paul Dirac had been aware of that fact ? Chessfan (talk) 17:32, 13 January 2013 (UTC)
- Yes, but. There's a key step you're glossing over in the above. What the Pauli complex column vectors most directly translate over to in GA is the subspace made up by ψ ∈ Cl2,0(R) (1 + e3). (Lounesto, p. 60). In 3D we can map this isomophically to a spin-matrix / rotor / quaternion (in 3D they are equivalent), if we introduce Ψ = (Lounesto, pp 63 and 64), which is okay because the equation has invariance in . So it works for the Pauli equation in 3D.
- But nevertheless ψ and Ψ are different kinds of things. It is not (yet) clear to me that one can always make a move like this; and ψ, the more direct equivalent of the traditional spinor, has a particular meaning, being a general element of a subspace which is (I think) swept out by an irrep of the rotation group.
- And yet it does seem to make sense that one ought tobe able to map between such a thing and something like Ψ which more obviously corresponds to a rotation.
- So is something like this isomorphic identification always possible, in dimensionalities other than three? And how does the count of degrees of freedom work out, because an n dimensional complex column vector would in general appear not to have the same number of degrees of freedom as the even part of a Clifford algebra represented as an n × n matrix ?
- The answer no doubt is out there somewhere, but at the moment I simply haven't got the time to go looking for it. But perhaps someone here has already worked through this and can explain? Jheald (talk) 21:59, 13 January 2013 (UTC)
- Well, perhaps the simplified Hestenes definitions , without ideals and idempotents, are not the most rigorous ones from the pure mathematical point of view, but in R3 and Minkovski space they work. Alas I could not find the interesting pages from Lounesto on Internet. Chessfan (talk) 10:49, 14 January 2013 (UTC)
- See [1] and [2], but the whole book is worth getting hold of. Jheald (talk) 10:53, 14 January 2013 (UTC)
- Also useful is pp. 143-4, where in a not-too-dissimilar way Lounesto shows an analog from Dirac spinors to a set of real even multivectors (generalised rotors) in Cl1,3(R).
- However, note also p. 146, where Lounesto writes "Higher dimensional analogies for spin operators: ... We conclude that there is no analogy for spinor operators in higher dimensions." Jheald (talk) 12:00, 14 January 2013 (UTC)
Head section is not satisfactory
The first paragraph of an article is supposed to define the title term, in the most accessible language possible, even at the cost of some vagueness or "dangling ends". Then, in the following paragraphs, one may give some justification for the concept and summarize ramifications, variants, etc; but always in a language that makes sense to those who are not familiar with the subject. The current version of the head section fails to do so: instead of defining the term, it digresses on peripheral issues, and seems to make a point of "showing off" jargon that may well be understandable only by readers who already know what a spinor is. --Jorge Stolfi (talk) 22:40, 23 February 2013 (UTC)
- You are right here that a two-sentence spot on obvious definition of spinor is lacking, but it is extremely hard to come up with a reasonably accurate description of spinors that is easily and quickly understood by people new to the subject. There exists only bad and worse non-mathematical descriptions of them. Even the only moderately bad descriptions of spinors require an effort and a vivid geometrical fantasy as well as knowledge of a few mathematical concepts on part of the reader.
- Do you have any concrete examples of what what you mean by digression and peripheral issues and jargon? Critique of this sort is pretty much jargon too. You can't reasonably expect the full article on a difficult subject to use a language that makes sense to all readers new to the subject. It simply must use some definitions and terminology from the subject of matter. Then there is at least a chance that somebody willing to put in an effort will pick up something useful - or at least get started. The blue links aren't there to show off.
- But, the article does probably start out at one notch too high level of abstraction and rigor. The first order of business is perhaps to describe what a transformation is in this context. It is, at a basic level, simply change of our point of view that does not change the object we are viewing. An infinitesimal transformation is then only a slight change in our point of view. Well, that is a start! YohanN7 (talk) 13:47, 24 February 2013 (UTC)
Inconsistent notation for Clifford algebras
In the space of about two paragraphs Clifford algebras are variously labelled Cl(V, g), Cln(C) and Cl(p, q). Needless to say, this is confusing. Count Truthstein (talk) 13:47, 10 March 2013 (UTC)
- I have changed Cl(p, q) to Clp,q(R), to remove one source of inconsistency.
- This notation indicates a Clifford algebra over the reals R, that has p base vectors for which eiei = +1 and q base vectors for which ejej = -1. (Caution: this ordering of p and q is standard in the geometric algebra community, and in GA-inspired articles on WP; but some authors use the opposite ordering for p and q).
- The notation Cl(V, g) is more general. Here V is the space of base vectors, and g is the metric for an inner product over it. This is used in more abstract works on Clifford algebra, for example our Clifford algebra article, because it is more agnostic as to what fields the vectors are built over, for example fields of characteristic 2. Jheald (talk) 14:43, 10 March 2013 (UTC)
Proposal for new Transformation article
See my proposal at Wikipedia_talk:WikiProject_Physics#General_article_on_Transformation_may_be_needed. I think that such an article would be a useful prerequisite for reading the spinor article. Count Truthstein (talk) 16:41, 16 March 2013 (UTC)
More on topological analogies through illustrations/animations
![](http://upload.wikimedia.org/wikipedia/commons/a/ad/Spinor_orientation_entanglement.gif)
Editors of this article may like this link Teply found if they haven't encountered it already:
As a possible addition later (busy with exams for now), I drew a while back this crinkly and horrible animation for the orientation entanglement of a spinor, and intended to add it to this article but it was never finished to happen... It's based on the "book and belt" analogy in Penrose's The Road to Reality, corresponding to the orientation entanglement for a spin-1/2 particle (no clue for higher spins...).
If editors think this would be a helpful addition, I'll redraw it in a few weeks (if not then there is no point in redrawing, obviously). M∧Ŝc2ħεИτlk 17:30, 17 May 2013 (UTC)
Wikipedia:Make technical articles understandable
this article does not meet the wiki policy of making a general audience encyclopedia. the first paragraph is full of jargon that not only someone with a math undergrad degree would understand. engineers, biologists, etc, would have to look up the definition of several of these words. Decora (talk) 23:43, 4 August 2013 (UTC)
Spinors, spin-tensors, spin tensor, tensor products, and irreducibility
I don't object to this revert, but it would help to explicitly state in the lead somewhere (if you don't mind me copying/pasting your edit summary, Sławomir Biały):
- " "Spin tensors" refers to tensor products of the spin representations. This article treats only the irreducible case."
so my mistake will not be made again. Any thoughts? M∧Ŝc2ħεИτlk 17:21, 3 November 2013 (UTC)
" Rotation " of spinors ?
Since a long time I wonder if some physicists don't introduce an artificial scent of mystery in the spinor question by insisting heavily on vague general definitions like : Spinors are rotated in a different manner than vectors. They change sign when rotated by a 360 degrees angle ; to restore the original spinor one must rotate it by a 720 degrees angle ! That is particularly the case in the present Wikipedia article, first in the introduction, second in the 2-d and 3-d examples, which as they are now written are deeply misleading.
It should be clear to everyone, that expressions like :
represent the composition of two successive rotations of a vector u first by an angle , followed by an angle , but that interpreting as the rotation of the rotor by the rotor with an angle is a double mathematical interpretation error, because :
1/ The acting rotation parameters in are not and but that angles halved.
2/ Interpreting the unitary transformation of by as a rotation has no pure geometrical signification, neither in geometric algebra nor of course in traditional Clifford algebra. (In GA such a significative rotation would be expressed as , where the scalar part and the bivector part would be separately rotated in a very simple manner). Of course what remains true is the change of sign of under the unitary transformation with parameter , but that has nothing mysterious ! Call it if you want.
Those facts are not small details ; they justify in my opinion a rewriting of the introduction and the examples.
By the way the editors of the examples have not made a clear choice between a traditional Clifford algebra presentation and GA ; that is also obscuring the question, as the fact that in Clifford algebra the so-called rotations operate in a complex SU(2) group is not even mentioned.
Chessfan (talk) 15:44, 1 September 2013 (UTC)
- I agree that this needs to be redone with care. The physicist's concept of "rotation" confuses this picture: a fermion is considered to have rotated a certain angle if its spin axis (or spin phase, if this is aligned with the axis of the "rotation") moves by that amount. It is not correct to identify this "rotation" with the same rotation we speak of when the orientation of a vector is changed; it is a different operation. Accordingly, it should be given a different name. This is even further confused by the nonsense "analogy" used by even eminent physicists to give a feel for spinoral objects: the belt-and-book analogy. :Incidentally, a further confounding factor is that rotors (representing rotations), and spinors (such as those that represent quantum fields) are mathematically fundamentally different: the action of the first uses the inverse, whereas the action of the latter uses the reversal. This distinction cannot be ignored for an indefinite metric (in that the sign of the result for the same GA object change when treated as rotors and as spinors, even if one ignores the "normalization" distinction as a rotor).
- Unfortunately, to source this "more correct" perspective might be tricky, since the confusion is probably widespread. — Quondum 04:27, 2 September 2013 (UTC)
- Thank you for your support. I felt lonely ! As for the belt trick and other comments on GA and Clifford Algebra see http://phymatheco.pagesperso-orange.fr/ . Chessfan (talk) 07:01, 2 September 2013 (UTC)
- Right, the article spreads popular misconceptions. In fact, the numbers 360 and 720 of above should properly be regarded as parameters to (coordinates of) the Lie algebra of the group in question, not really as angles. They become angles in the standard interpretation of the standard representation of the group, but not in the spin representations. So, you aren't necessarily rotating anything by 360 or 720 degrees when these numbers appear in formulas. Also, analogies and illustrations should probably be avoided unless they are 100% faithful about what's really going on, which I doubt is even possible in a 2d illustration. YohanN7 (talk) 17:18, 2 December 2013 (UTC)
- All "spinor analogies", in whatever form (books, belts, palms, etc.) are absolutely useless. They don't provide anything helpful except indirect and trivial topology.
- Also, when the term "rotate" is used in the context of spin matrices, e.g.
- and spinors, e.g.
- as far as I can tell, this usually means the spinor components are simply mixed up according to the transformation matrix
- rather than any physical rotation.
- Not sure if what I said was already obvious or helpful, just chipping in... M∧Ŝc2ħεИτlk 18:07, 2 December 2013 (UTC)
- Very illuminating of the discussion. Unfortunately, one cannot get away from "rotations" all together, since if we subject 3-d space to (what we all would call) a rotation (active or passive, doesn't matter), then the spinors of various dimension (the (2s + 1) of above), being functions of 3-d space, will undergo the component mixing. The latter transformations aren't rotations, but they are still related to an ordinary non-mysterious rotation. This relation is not one-to-one. The clue to a good description lies, I believe, in using a path in the Lie algebra, but don't press me on this.
- If the article can make this clear, it would be very nice. Getting it right verbally and formally, and sourced, will not be easy (people have tried for a hundred years by now to make non-mathematical sense of this). YohanN7 (talk) 19:36, 2 December 2013 (UTC)
- Yes. The point on component mixing is also not too helpful (actually trivial) because that happens under any linear transformation anyway. The correspondence between a "physical rotation" and component mixing is really vague... It's not like the simple 2d rotation
- but something else "related". M∧Ŝc2ħεИτlk 21:14, 4 December 2013 (UTC)
- Yes. The point on component mixing is also not too helpful (actually trivial) because that happens under any linear transformation anyway. The correspondence between a "physical rotation" and component mixing is really vague... It's not like the simple 2d rotation
History
Cartan (Elie Cartan: The Theory of Spinors, Hermann, Paris, 1966), the first sentence of the Introduction section of the beginning of the book (before the page numbers start) reads:
- Spinors were first used under that name, by physicists, in the field of Quantum Mechanics. In their most general form, spinors were discovered in 1913 by the author of this work, in his investigations on the linear representations of simple groups*; they provide a simple(P.S.: a typo: this should read "linear" not "simple". C.H. 22:11, 26 February 2014 (UTC)) representation of the group of rotations in a space with any number n of dimensions, each spinor having components where or .
and the star (*) refers in the footnote to Cartan's work of 1913 (E. Cartan "Les groupes projectifs qui ne laissent invariante aucune multiplicité plane", Bull. Soc. Math. France, 41, 1913, 53–96).
Unless this statement is explicitly refuted by a reliable source, this statement that was recently deleted should certainly be reinstated. (And by the way it is still stated thus in the "History" section of the same Spinor article.) I will notify the person who made the change, wait a day or so for reactions, and otherwise revert that deletion. --Chris Howard (talk) 07:58, 26 February 2014 (UTC)
- There is clearly something wrong with your reading of that passage. The emphasis on the first sentence should be on the phrase under that name. It is true that Cartan did not come up with the term "spinor", but his 1913 paper obviously predates anything to do with their application to quantum mechanics. The spin of the electron was not even discovered experimentally until 1925, and it was not explained mathematically (using spinors) until a few years later. Incidentally, the second sentence of the cited passage actually supports the article as written. Sławomir Biały (talk) 12:04, 26 February 2014 (UTC)
- Seems you got my message in a way that is exactly the contrary of what I mean. I actually criticize the deletion. That is, I fully agree wish you that "The emphasis on the first sentence should be on the phrase under that name. It is true that Cartan did not come up with the term "spinor", but his 1913 paper obviously predates anything to do with their application to quantum mechanics. The spin of the electron was not even discovered experimentally until 1925, and it was not explained mathematically (using spinors) until a few years later." That is, to all apparences we agree that we have to conclude that it was Cartan who discovered spinors first? That is, you agree that the deletion should be reverted? --Chris Howard (talk) 13:16, 26 February 2014 (UTC)
- Yes, I misunderstood you. We are in agreement that the original was correct. Sławomir Biały (talk) 13:24, 26 February 2014 (UTC)
- I agree with you two now on the emphasis of Cartan's statement -- I was too hasty in making the edit. A.entropy (talk) 16:14, 26 February 2014 (UTC)
- Fine, it's solved then. And someone has already made the change to the article text. --Chris Howard (talk) 21:52, 26 February 2014 (UTC)
The new lead
The new lead has taken a quantum leap towards becoming decent in terms of describing spinors in layman terms. By "decent" I mean better(?) than is to be found elsewhere, because most (all?) existing descriptions are all more or less bad. That is to say, spinors are notoriously (and inherently) difficult to describe in words. One slight problem with the current lead: What is a continuous rotation? Mathematically, a rotation is a rotation period, but I think I can see what Slawomir is getting at. I can also envisage the layman reader understanding the gist of "continuous" here. My point now is that if we manage to explain the "continuous rotation" versus just a "rotation" already in the lead, then we'll really get somewhere. Not all "continuous rotations" are equal, even if they are equal (in one particular sense) as "rotations". I believe the key is to (as the current lead does more or less point out) is to follow a path in the Lie algebra and contrast what is happening in the "ordinary group representation" compared to a "spinor group representation". Can stuff like this be put into layman terms? YohanN7 (talk) 20:35, 15 August 2014 (UTC)
- By continuous rotation, I just mean a rotation that depends continuously on a parameter. I struggled somewhat with how to fit this in the lead, but it may be time to bite the bullet and just bluntly say this early on somewhere in the first paragraph. Sławomir Biały (talk) 21:00, 15 August 2014 (UTC)
- Yes, depending continuously on a parameter certainly will define "continuous rotation". But it it will not suffice for spin. For this we need another "continuous rotation", using another continuous rotation depending on some other parameter, resulting in the same final "rotation". Just thinking out loud. YohanN7 (talk) 22:20, 15 August 2014 (UTC)
- That's if you need to think of the "spin" as a concrete sign. If you just think of it as the cocycle that labels which sheet of the 2:1 cover the rotation lives on, then it is a homotopy invariant of the path. Sławomir Biały (talk) 23:24, 15 August 2014 (UTC)
- Sławomir, just a mention (I have no intention of getting involved in a debate): your use of the word "spin" seems unusual (as a binary disambiguator, rather than something that is measured as angular momentum). This article is about spinors, and not spin, so would it not be less confusing to leave it out? Also, the connection with quantization (spin states) in quantum mechanics seems to me relatively incidental and overemphasized in the lead, and risks adding unnecessary quantum mystique to a fundamentally geometric concept. —Quondum 00:09, 16 August 2014 (UTC)
- It is not my intention here to emphasize the angular momentum operator from physics. Here by "spin", I mean merely the discrete invariant indicating the sheet of the 2:1 cover of the orthogonal group by the spin group. This is obviously essential to any proper treatment of spinors. I'm open to a better term, but there does not seem to be an entirely standard term for it (although this use is arguably consistent with its use in connection with spin structures). Sławomir Biały (talk) 00:43, 16 August 2014 (UTC)
- Does this edit address your reservations without generating any new objections or making the content less clear? I have used the proper term for what I called the "spin" earlier (it is a homotopy class), but fear that perhaps the term itself might be unduly intimidating. Sławomir Biały (talk) 01:13, 16 August 2014 (UTC)
- In response to Slawomir: Putting it more abstractly isn't going to help the reader. YohanN7 (talk) 01:25, 16 August 2014 (UTC)
- Thanks. How does the present version sit? (Signing off, will look tomorrow for replies.) Sławomir Biały (talk) 01:34, 16 August 2014 (UTC)
- Sits nicely, which is why I started this thread to begin with! YohanN7 (talk) 05:51, 16 August 2014 (UTC)
- To be more precise, now we have a chance to get a decent description of spinors in words. We (you that is) are close, but not quite there
YohanN7 (talk) 06:01, 16 August 2014 (UTC)
- Ooops, things have happened since I read the lead last time. Something to think about here. Maybe close to a good decent description? YohanN7 (talk) 06:06, 16 August 2014 (UTC)
- From my perspective, this change works very well. In an article like this, a few unfamiliar terms would be expected by the reader. I've had this before, where use of a semi-familiar term in an unfamiliar way can be an impediment to understanding. Incidentally, this is the first time I've seen the plate trick example used in a way that is really apt – it is nicely done. —Quondum 06:53, 16 August 2014 (UTC)
- Thanks. How does the present version sit? (Signing off, will look tomorrow for replies.) Sławomir Biały (talk) 01:34, 16 August 2014 (UTC)
- Sławomir, just a mention (I have no intention of getting involved in a debate): your use of the word "spin" seems unusual (as a binary disambiguator, rather than something that is measured as angular momentum). This article is about spinors, and not spin, so would it not be less confusing to leave it out? Also, the connection with quantization (spin states) in quantum mechanics seems to me relatively incidental and overemphasized in the lead, and risks adding unnecessary quantum mystique to a fundamentally geometric concept. —Quondum 00:09, 16 August 2014 (UTC)
- That's if you need to think of the "spin" as a concrete sign. If you just think of it as the cocycle that labels which sheet of the 2:1 cover the rotation lives on, then it is a homotopy invariant of the path. Sławomir Biały (talk) 23:24, 15 August 2014 (UTC)
- Yes, depending continuously on a parameter certainly will define "continuous rotation". But it it will not suffice for spin. For this we need another "continuous rotation", using another continuous rotation depending on some other parameter, resulting in the same final "rotation". Just thinking out loud. YohanN7 (talk) 22:20, 15 August 2014 (UTC)
- Can we get in how the "shape" (the actual path in the standard representation of the group) of the continuous rotation is important? It is there now, but not verbatim. YohanN7 (talk) 06:39, 16 August 2014 (UTC)
- I'll have to take that back. But still, even more emphasis on the path that leads to the "final" rotation? YohanN7 (talk) 06:49, 16 August 2014 (UTC)
"like vectors"
The first thing we now say about spinors is they are "like vectors".
I think this is a collossal mistake. Especially people (eg physicists) introduced for the first time to spinors as a column of complex numbers are likely to think that this is some exotic form of Euclidean vector. But this is not a good way to think about them. Instead, it is important to understand that they carry around with them a lot more information than just an arrow in space.
Yes, what YohanN7 has written is of course perfectly true. A rotation causes each of a vector, a spinor and a tensor to transform in their own particular way.
But nevertheless, to start out by saying to a novice reader than a spinor is "like" a vector is not helpful. The associating of them with tensors gives a much better picture. Jheald (talk) 23:25, 17 August 2014 (UTC)
- I neutral on the matter of whether it is helpful to think of spinors as like vectors, or if it rather does more harm than good. It is certainly true that they are like vectors in the sense that they transform under rotations, which is all the current line of the article now says. But whether this leads to a misleading view of spinors or not, I cannot say. Sławomir Biały (talk) 11:04, 18 August 2014 (UTC)
- And the perfect thing, of course, is to say they are like tensors (like your edit proposes)? Vectors are tensors, just more familiar.
- You are just bitter because you were reverted. YohanN7 (talk) 16:01, 18 August 2014 (UTC)
- I don't know. I can see the point that Jheald is making. The description we have written here already makes more than a small nod to the tensor concept, and it is best I think to assume that most prospective readers have at least a passing familiarity with tensors. But I also don't think it is excessive for the lead sentence to contain a small mention of vectors, intended primarily for readers who would not make it past the first few sentences of the article. Sławomir Biały (talk) 20:03, 18 August 2014 (UTC)
- And, by the way, spinors as being "some exotic form of Euclidean vector" isn't that bad a description. Spinors are vectors, which might surprise you. But they aren't really tensors. YohanN7 (talk) 16:14, 18 August 2014 (UTC)
- Citation needed? I know 2-component spinors can be consistently represented by a little vector with a flag attached, but is that generally true for general spinors in n dimensions? Jheald (talk) 21:57, 18 August 2014 (UTC)
- I took Yohan to mean that spinors are elements of a vector space, and so "vectors" more in the mathematician's sense of the word than the physicist's. I agree it would be very misleading for the article to suggest that spinors are vectors. But comparing spinors to vectors seems like something possibly worthwhile. Sławomir Biały (talk) 22:32, 18 August 2014 (UTC)
- If we compare spinors to tensors, it would be pretty ridiculous not to compare them to vectors. They are vectors, but are not, in the very sense of the article, and in every other (semi-mathematical) sense, NOT tensors. Who is right in this discussion? You or me? YohanN7 (talk) 23:45, 18 August 2014
Just to try to get consensus, does this edit, emphasizing also the way that spinors are unlike vectors, satisfy the original objection? Sławomir Biały (talk) 11:46, 19 August 2014 (UTC)
- It is an improvement. Also, when rereading what I wrote previously in this thread, I don't like it much. The tone is bad, and for that I apologize. YohanN7 (talk) 12:37, 21 August 2014 (UTC)
Overly focused lead
This is an "are we not missing the wood for the trees?"-type observation: the lead presently focuses on a particularly unusual aspect of spinors (essentially the topological property under coordinate transformations), at the expense of leaving some fundamental questions unanswered. For example:
- It unclear whether spinors form part of a vector space (addition, scalar multiplication, zero element); e.g. unitary matrices can change sign but do not form a vector space. This is sort of covered by the second paragraph, but it is difficult to interpret.
- The lead seems to describe spin-1/2 objects only. Are spin-0, spin-1, spin-3/2 and spin-2 objects also called spinors? If not, what are these called?
- What does a spinor do? Does it act on other objects? Does it have a magnitude?
—Quondum 14:42, 21 August 2014 (UTC)
- In regard to the first question, from the article "spinors are elements of a vector space, often over the complex numbers, together with a linear representation of the special orthogonal Lie algebra." For the second question, spinors are spin representations of the orthogonal groups. These can either be identified by the weights, or by some topological feature. This includes objects of any spin. But there is a slight difference in how mathematicians and physicists would describe objects of spin other than 1/2. For integral spin, the corresponding representations are actually all tensorial; "spinors" in the mathematician's sense are not even strictly required. For half-integral spin, the representation splits into irreducible half-spin representations (with some multiplicities).
- The last question seems to concern specific spin representations. These may or may not admit complex, quaternionic, hermitian, or symplectic structures depending on the signature (and whether the ground field is real or complex). I'm not sure I understand the question of "What does a spinor do?" A spinor is not a verb. It does not "do" anything. There are decompositions of the Clebsch-Gordan variety that tell us how to combine different spinors or spinors and vectors, etc. Is that what you mean? Sławomir Biały (talk) 17:34, 21 August 2014 (UTC)
- With regard to the first point, the phrasing "In this view, spinors are elements of a vector space", the introducing phrase creates a qualification, equivalent to "As it were, ...", which complicates interpretation. A matrix group gives an algebraic representation of a group inside a vector space, e.g. of an orthogonal group. That does not turn orthogonal groups into vector space. The spin group is a group, not a vector space. The Clifford group supports scaling by a nonzero real. It is my impression that spinors are one-to-one with the Clifford group, plus the zero spinor, so they'd support multiplication by a scalar (from the underlying field). The group operation in this example suggests that spinors can be composed. Continue on this line, and we might end up with addition being supported, but I'm doubtful here (the result would not generally be a spinor). Am I on the right track? Either way, I can see wording changes needed. I'll leave the other questions for now. Let's say I'm trying to formalize the object that is the set of spinors. Is it a group? A monoid? —Quondum 03:04, 22 August 2014 (UTC)
- The by far easiest way to view spinors is as elements of Cn for some n. (This is what they become anyway under any description when given a basis.) They transform by the usual matrix multiplication from the left, and can certainly be added and multiplied by scalars (and are hence vectors in this sense). Inside the Clifford algebra sits a representation of the Lie algebra in question. When this representation is exponentiated, it becomes a spin representation (projective representation) of the group in question. But don't confuse the representation with the representation space. YohanN7 (talk) 06:49, 22 August 2014 (UTC
- Both bispinor and representation theory of the Lorentz group give fairly explicit examples. YohanN7 (talk) 06:49, 22 August 2014 (UTC)
- The set of spinors is a vector space. It is a representation of the spin group (or Clifford group), but it is not in one-to-one correspondence with either of these. The article itself should be clearer about this, I think, but there is limited scope for such matters in the lead. As Yohan says, probably the easiest concrete way to view the spinors is as elements of Cn. The Clifford algebra can be concretely represented, basically using Pauli spin matrices and their tensor products. These act naturally on a complex vector space. Sławomir Biały (talk) 10:51, 22 August 2014 (UTC)
Can we remove the words "In this view", then? And(thanks for your edit) What do you mean by Cn? I understand this notation as the direct sum of n copies if C. I presume you really mean square matrices Mn(C)? And I understand the spin group as being 2-to-1 to the orthogonal group, and the Clifford group as being ∞-to-1 to the orthogonal group. —Quondum 13:26, 22 August 2014 (UTC)
- Already done. Here by Cn I also mean the direct sum on n copies of C. The Clifford algebra is Mn(C). It acts on Cn in the usual way. In this picture, the vectors of Cn are the spinors. The spin group is a subgroup of (the group of invertible elements in) Mn(C), and so it also acts on the spinors. Sławomir Biały (talk) 13:33, 22 August 2014 (UTC)
- In the case of the Lorentz group, the Clifford algebra plays two roles. In addition to containing representations, it is a representation space. It decomposes (direct sum of vector spaces), in physics lingo, as a scalar, a vector, a tensor, a pseudo-vector, and a pseudo-scalar (but no spinors in this sense). The part being the representation space of the tensor representation (six-dimensional) is itself the Lie algebra representation. I don't know to what extent this is a general feature. YohanN7 (talk) 12:23, 22 August 2014 (UTC)
- I'm afraid the term "representation" gets used a lot, but it seems to have multiple senses, and it does not help if these are not defined form the purposes of the discussion. I can understand matrix an GA representations of spinors, because these contain a subset that is isomorphic under certain operations; scalar multiplication, spinor addition, and spinor composition (multiplication in the representations) seem to be part of this structure. Your "In addition" suggests that there is a second meaning that I'm missing. —Quondum 13:26, 22 August 2014 (UTC)
- A representation of a group G on a vector space V is group homomorphism of G into GL(V). For spinors, G is the spin group and V is an appropriate vector space (the construction of which is deliberately left vague at this point). The main feature of V that distinguishes it as a "spin representation" is that this representation be faithful (no kernel). Elements of V are called spinors. (This all can also be recast in terms of Lie algebras, but then identifying the spin representations becomes trickier.) Sławomir Biały (talk) 13:38, 22 August 2014 (UTC)
Cool. This is my understanding of a representation of a group. So we have spinors as a representation of a group. The elements of G would then map onto a subset of spinors in V, those which (for want of a better term) have norm 1, and the group operation on G maps onto an operation between spinors, which in the Clifford algebra and matrix forms would be multiplication. Spinors, by virtue of a representation being a vector space brings in fresh operations of addition and scalar multiplication. The even part of the Clifford group is presumably represented by the spinors that represent the spin group G, plus all elements of V obtained through scalar multiplication. Addition of spinors presumably fits in somewhere – it seems to in physics. (I'm guessing a bit.) If you see what I'm driving at, this could possibly be developed into a way of constructing spinors that could be understood by physicists. The spin group could start as the double cover of the orthogonal group. —Quondum 16:01, 22 August 2014 (UTC)- Sorry, I need to rethink, I'm probably confusing things like V and GL(V). The representation maps G → GL(V), not V, which makes nonsense of my statements. —Quondum 16:10, 22 August 2014 (UTC)
- What Sławomir writes is of course true, but one should be aware that terminological abuse is common, both in math and physics. A representation can refer both to GL(V) and V. It may sometimes get even worse in physics when the distinction between a group and its Lie algebra is blurred. The only way to handle this is to stay awake when reading. It has caused me some pain too. Your last comment indicates that you are at least coming close. YohanN7 (talk) 16:13, 22 August 2014 (UTC)
- Thanks, I think you've managed to capture why I keep getting confused whenever I see the word representation. It'll be nice if we can disambiguate where needed. —Quondum 16:22, 22 August 2014 (UTC)
- @YohanN7: Did you mean "A representation can refer both to GL(V) and V"? I can make sense of a different statement "A representation can refer both to the group homomorphism G → GL(V) and to its image in GL(V)". I also posted a question with the associated article. —Quondum 06:03, 23 August 2014 (UTC)
- I think it is worth it to waste a sentence on what a representation is in the article. It takes at least three ingredients. A group G, a vector space V, and homomorphism G → GL(V). Probably the spin group (the double cover of our indefinite orthogonal group) deserves a place too. YohanN7 (talk) 16:37, 22 August 2014 (UTC)
- Sławomir's description just above gives me the first solid feel that a sharp definition can be grasped. So yes, I absolutely agree that a definition in terms of these concepts should be in the article, along with clarification of the concepts as you mention here. We could even use it to define in what sense the word representation is meant in this article. —Quondum 18:25, 22 August 2014 (UTC)
- I took a wild stab here. Have at it. Sławomir Biały (talk) 19:20, 22 August 2014 (UTC)
- This is giving about as much introduction from the group perspective as we can expect. The Group representation article fills in a bit about what such a representation is, so I'm not sure how much we want to repeat in the article, though I imagine a little repetition here would help. In particular, in § Overview or § Explicit constructions, using the symbolic notation G → GL(V) in the description might help to make it more concrete. —Quondum 06:03, 23 August 2014 (UTC)