Reprinted from JOURNAL OF MATHEMATICAL PHYSICS VOLUME 8, NUMBER II NOVEMBER 1967 Spin-s Spherical Harmonics and J. N. GoLparnc* AND A. J. MACTARLANEE ‘Syracuse University, Syracuse, New York E. T. Newant, University of Pittsburgh, Pittsburgh, Pennsyloania F, Rouruicu® anp B. C. G. Subansuant Syracuse University, Syracuse, New York (Received 26 September 1965) Recent work on the Bondi-Metzner-Sachs group introduced a class of functions «Yia(0, $) defined ‘onthe sphere and a related differential operator 6. In his pa Imatrices of the rotation group R, and the properties of 3a the Yn related fo the Teprezemation served fom its eaionship tan angular ‘momentum raising operator. The relationship of the 7,4, #) to the spherical harmonics of Ry is also indicated. Finally using the relationship of the Lorene group to the conformal group ofthe sphere, the Dbohavior of the Ziq under this latter group ls shown ‘0 realize a represeutation of the Lorene eroup. 1. INTRODUCTION A RECENT paper by Newman and Penrose on the Bondi-Metzner-Sachs group! features a new differential operator, symbolized by 5 (“edth,” the phonetic symbol for the hard “th”), and a related class of functions ,¥;4(0, 4), all defined on a sphere, in a central formal role, Itis the purpose of the present paper to study 9 and these generalized spherical functions and to relate them to more familiar struc tures. In Sec. 2, we review previous work and give some further geometrical interpretation of thop as well as * This research was supported ia part by the National Science Foundation "Research supported in part by the U.S. Atomic Energy Com- { Research supported in part by the U.S, Air Force Ofive of Selenite Researen, NE, Newman an R. Penrose, J. Math. Phys. 7, $53 (1966) ©The operator symbolized by'8 hat been refered to collguialy 0 ee an illustration of the suitability of 0 and the ,Yin(6, ¢). s=1, 0, —1, in the manipulation of Maxwell's equations. In Sec. 3, we introduce and develop the formalism which allows one to view 5 as a thinly disguised angular-momentum lowering operator and to relate the ,¥,4(6, 4) to the elements of the repre- sentation matrices of the rotation group Ry. This ‘work was on the one hand motivated by inspection of the results reviewed in Sec. 2 and on the other hand allows 2 simple rederivation and ready extensions of such results, Asan adjunct to this section, the relation- ship of ,¥:q(0, 4) to the spherical harmonics of Ry, i.e,, those functions which carry the representations of R, defined-on the unit sphere in four dimensions, is briefly indicated. In Sec. 4, we discuss the well- known relationship of the Lorentz group to the conformal group of the sphere and determine the behavior of the ,¥;, under the conformal group, thereby realizing’ a representation of the Lorentz group of somewhat unusual appearance. 2155 [Conriht (1968 oy te Amatcas Inuit of Pagnie2156 2. SUMMARY OF PREVIOUS WORK In this section we discuss some of the previous work! on the differential operator 8 and the spin-s spherical harmonics , Faq» In three-dimensional Euclidean space with polar coordinates r, 0, g, we introduce an orthonormal triad a, b, and e of vector fields. The vectors a and b are tangent to the sphere of radius r at each of its points while e is in the direction of the radius vector. Of course a and b are only defined up to a rotation of angle » about e. I is very convenient to introduce in place of a and b the complex vector m and its complex conjugate ii by means of V2m = a + ib; Qa) then mis defined up to a phase factor, ie., m' = em, A quantity is now said to be of (integral) spin- weight 5 if, under (2.1), it transforms according to af =e, (2.2) Examples of quantities of spin weights s = 1, 0, respectively, are Asm, Asc, Aem, where A is any vector, More generally, examples of quantities of spin-weight s are furnished by three dimensional tensors of rank contracted ky, #,, and ik, times with m, c, and mM, respectively, where ky — ka = 5, hy + Ky + ky =, We adopt the con vention that the real and imaginary parts of m point along the coordinate lines and) hence transform according to (2.2) under coordinate transformations, ‘The differential operator &, acting on a quantity 7 of spin-weight , is defined by = —(sin off Since one has + ose 8 ale ayy. 23) ny’ = en), it is seen that 0 has the important property of raising the spin weight by I. Similarly if one defines 5 by By = —(6in 9p [F-90035 [on (23a) 24) with 7 here also a quantity of spin-weight s, one can see that B lowers the spin weight by 1. Also one has Gs — By = 2, Of importance too is the effect of 3 on ordinary spherical harmonics: Yin(Os By —ESm <1, Indeed we can define spins spherical harmonies Ol GOLDBERG «7 al «Yim for integral s, 2, and me by Vins 8) = UL UT SOY (Ds Dy O?) form « complete orthonormal set for each value of 5; ie., any spin-weight s function can be expanded in a series in Yq. The spines spherical harmonies have the further? properties 6 Fin = (Yin 2.6) @) BYin = (CM ++ DNhyaYims (27a) Gil) BY = — [+ SU — 5 + A Yow» 2.70) @&) BYin=—U-N+54 DY 28) ‘Thus d and 5 act as raising and lowering operators on the “quantum number” s, and the ,¥,q. are eigen- functions of 3°. For many computations, a more convenient coordi- nate system for the sphere is the set of complex stereo- graphic coordinates (Z, £) which are introduced by [=e cot 2. es) and 5 become by = Pe *[a(Pn)/OE), By = 2PM[A(P=ny/2E), (2.10) with P = 4(1 + £2). In the (2, 2) system, the spin-s spherical harmonics take the form Ye = Goya oe +0" To) ( TAS Voy_gpmn R(T GLE) enn with Aig = (YE + om)! (UL my QE + 1/4. (2.12) Expression (2.11) applies also to “spinor harmonics" for which I, m, and $ are all half-odd integers. 8 can be related to covariant differentiation in the following manner: using coordinates on the sphere such that the metric takes the form? dst = Ped dl, we introduce two complex vectors m= v 27x, it = V282, = &, ©, From a spin-weight quantity n, we can define a totally symmetric and traco-free [The function P and the coordinates {noed not be the ones used in Eg, (2.10); a8 & muses Of fat the surface need not evea be & sphereSPIN-s SPHERICAL HARMONICS AND 6 tensor of rank 5, es) = ig Hip + Hig yy With the inverse relations Tea itt It is now easy to prove 8 = N Drang et MO 2.13) ‘As a simple example illustrating the use of 0 and the , Yin, We consider the Maxwell equations V-(e+iB)=0, VAG +B —i@iene +im)=0, ‘The quantities! G,=(E+B)-m, Go = (B+ iB) +e, 2.15) G =E+B)-m of spin weight 1, 0, and —1, respectively, cam be showin from (2.14) to satisfy the equations é 8 Ae 5 yre = 8G, =0, @.16a) (G 7 Sra — 38,6, =0, (2.166) in which the quantities G,, Gj, G_ are already un- coupled. If we assume solutions of these equations of the form 1G, = F.6D Yonl D8), PG, = Fle, Da¥ (68, 1G_ = Ft, D-aYeul Os itis seen from Eqs. (2.7) and (28) that Grail a7) Fe + (C= TE DE, a t *)r,4 Lue nF =o, ja)fot alls OF (2-2)e(2+2)e seine ae ene 8 the dependence on angular variables having canceled out. These latter equations can be solved by a variety of standard techniques, though it is not our purpose to go into this question here. © G5 Gy, @_ have been rod to elsewhere asf 8. See. eg, E'Ndiman and Re Pen Saati ‘'Sints Pays. 3,368 (1502) 2157 ‘The main point to be made is that Maxwell's equations or more generally vector equations can be simply solved in terms of the ,rq instead of the cumbersome apparatus of the vector spherical harmonics. 3. RELATIONSHIP TOR, AND Ry In this section, we identify the functions , Yig with the elements of the matrices of the representation Dé of the ordinary rotation group Ry, and relate & to an ordinary angular-momentum raising operator We thereby obtain the principal properties of the Yim and 0 as transcriptions of results familiar in the theory of angular momentum. We proceed first to the above mentioned identi- fication of the ,¥iq. For our purpose it is convenient to have an explicit definition of ,¥),(5, 4) rather than the expression in terms of stereographic coordinates given in Eq. (2.11). By direct substitution of (2.9) we obtain’ re, 9) = [deem = at NT no 3079 AE Joram Bn) ‘Now we give’ careful definitions of and appropriate explicit formulas for the elements of the matrix D', of the representation of Ry associated with total angular momentum J. Ia spatial rotation R of angle wv about a unit vector m is given by oto = RE, Re = & 605 w + ntnt(L — 608) — sino, G2) thea the matrix D‘ may be defined by its action on spherical harmonics &| lm) > VinlB, Gin 9 008 g,sin 6 sin g,c08 9), (3.3) Yip) Dhol RY. FT hav rossitly been pointed out co us that the Functions Fig have aieady beh introdoed, though by very diferent techniques ieallcnate methods and its detailed application to Mayall ate, Mc Galiand, R.A. Minios, and 2. Ya Shapiro, Repre Sturn of the Rotation aed Loren: Groups ead thew Aippitctions Gtk Macminn Company, New York, 190). Sethe pasage fom £q. C21) debaing »Yo(E.)c0 Eq GO.) fone ssa not only insect the cetntion (2.9) but abo Introd! ditional phaee factor «#0 ta account Tor the Sssoeietod withthe change of enordinates(,€)t0 6 8 fhe nesessiy forthe deial of the Ucission hese stems from tn far that we Cou oL simply refer to ore ofthe few completely Conasctenttrestments of tne theory of the roUtion group avaiasle inthe Itersture, without extensive mocifeaton of the notation aplyed in Ref 4 and elated paper.2158 If we define a rotation R(x, 8,7) of Euler angles a, By 7X8 being composed of 7 sbout OZ followed ‘by Babout OY and then 2 about OZ we have Doar 2fty) = Drural REaByY Ga) din BJO. Following Wigner,” in principle it not in detail, we employ the relationship of Ry to SU, in order to give an explicit formula for D\.,.(xgy). If the element A of SU, acts on a two-component spinor w = (;), where G5) so that ufo = ¢, according to w—r w= An then the correspondence of ASU, to RER, can be given in the form'® RY =} Tr (otAata!), A= £(1 + oo RA + Te RIE, w= e% cos 19, v= el? sin 16, BO which allows us to obtain the image A(@jy) of R(@py) ithe form Now defining Bratt = [V+ mtg — mF as usual we can, in agreement with Eq, (3.5), write Prnltle > Pyal's OD, Poult’, 0) = YF jpets WD pr ACP Ds Die whaBy) = Di (Alay) ; a fUtmi— mip (itm) j—int (sa 23") 2) = [Gm = OT payer - [Gav wil » w3(")( Ja Jom G8) x em cot 1pyromn 6) "This procedure is clearly quivatent to the more usbal one of & station 4 around OZ, followed by f acound OF" and Winall > STE. P, Wigner, Croup Theory (Academic Press Inc, New York 1359) 18 Ie follows now that under we m= the (ransormation Ini "= Sourse rquires & has GOLDBERG « al We may now insert a=, B= 0, j=h m' = —s into Eq, (3.9) and, by comparison with Eq, (3.3) obtain Men = [2+ N/A] Dl 657), B10) we can make the promised identi- so that for y fication Yen) = [2E-+ Dlx] DE, ($60), Note that for s oYin(68) = (Al + D/4m] D5 (490) = Yo,(08), so that the spin-s spherical harmonies with weight s=0 are exactly the ordinary spherical harmonics. It may also be noted that our procedure extends the definition of spin-s spherical harmonics to the case of s halintegral Now the functions D",(¢f)) provide!” « complete orthonormal basis for Tunetions defined on Ry, so that orthogonality and completeness relations for «Yin(9, $) follow easily, The orthogonality relations [jaa]! dc08 0 ty Dian) Dil 209) = BQ + DbeSamdse (3.12) translate, by use of Eq. (3.10) and relabeling, into ffi £05 6, Fin( 08). Yew (OB) = It is noteworthy that we obtain in this way only an orthogonality relation involving spin-s spherical harmonics of the same spin weight. Orthogonatity of the D!,,, with respect to s in Eq, (3.12) is of course associated with the variable y which is absent in Eq, 3.13), Also from the completeness relation © Brena) Dien 2 22) a8/(21 + 1)8(a — 2')8(cos B — cos PHY — 7'), 14) @an ), we have Smee B13) we can prove, by evaluating [faeres (where s is any integer) on both sides, that we have a completeness relation YE Fin(O),Yin(6'#') = a4 — o')8(c0s 6 — c05 6°, - G.15) for each integral value of s. Thus for each integral s the function ,Yrq(0f) Form according to Eqs. (3.13) and (3.15) x complete orthonormal set of functions on the unit sphere with respect to which any function of spin-welght + defined on the unit sphere can be expanded. 1, Penrose, Prov. Cambridge Phil So, $8,137 (1958).SPIN-s SPHERICAL HARMONICS AND 6 We turn now to the use of Eqs. (3.10) and (3.11) to motivate the association of 0 with an angolar-momen- tum raising operator. We set out from the observation, familiar from the theory of the symmetric top, that if one defines operators 1, 1 a -g, ico a ea tose 2) G19) ap which obey the commutation relations We tel the, (yb) = 2b, of angular momentum, then for each allowed value of s, Dé.,.(ay) behaves like an eigenvector |in}, Le, LDL = + DDE LDi jg = MD on, Ly Dhom = (QF my ra + DY] Diyas Gl) We do not relate L, to 9, of eourse, but instead define a second angular-momentum operator K, which commutes with L, and with respect to which DE ,, behaves like an eigenvector |/s) for each allowed value of m. The way to define K follows easily from the symmetry of D',,,(2, 8, 7) with respect to m, « on the one hand ands —"y, on the other. Thus, we define x=id, ay K.= sen(Se ira + ioe?) G18) and deduce Ik. Ik,,K1=2K,, 6, and ia 1) DF, K,Dh ig = Dh imy K Dion = (EF EES + DED scrim B19) We are now in a position to make explicit the relation- ship of K, to 0. When acting on D!,,, the operator K, can be written in the form (5 - iscot 8 + rese8 2) (sin oss + tesep 2)csin By. (20) so that [KDE sp lang,p-t,y0 = ODE sy 1800) (3.21) follows in accordance with Eq, (3.1). Thus K_ is the ‘differential operator to which the operator 3 is more closely related. The reason that 0 is not defined as a differential operator by Newman and Penrose stems 2159 from the fuet that they work only with ,Yin(06) ~ Dr,,(@60) rather than DE,(66>), ie, from the nonappearance of the variable ». Of course this in turn results from the fact that such a variable is not needed by thom on physical grounds, However, the properties of 5 follow very easily from its relation to K,.. For example, from (19), (21), and (10) we get lirectly BYou(O$) = [= + 5 + Ra (68), (3.22) which is Eq. (.22) of the paper by Newman and Pen- rose, OF course, it was results like this one which initially suggested the relationship of s to a magnetic quantum number and motivated the identifications of 8 with an angular-momentum operator Finally, it may be worthwhile here to point out the relationship of & to representations of R, defined on the unit 4-sphere a?+ 23422 + xd = 1. Tt i well known that the generators of infinitesimal rota- tions of Ry can be defined according to My = —i42)— 18), 1k, 154, and replaced by a pair of commuting angular momens tum operators £, XK: E.Mey +t Migs Go May + Macy {= Mt Ma, Ky = My — My Ky = Ma — Maas Hy = Mis — Mua in view of the consequence £* = .K* of these definitions, only the subset (J, k) of representations of Ry with = k= 0, }, 1,+7+ ean be defined on the unit 4-sphere with these “‘standard” definitions of the six infinitesimal generators, However, only. these representations arise in the previous discussion. We can explicitly make contact with the formalism of the previous paragraph by introducing polar coordinates according to x = sin $8 cos B(x — y), Ha = cos BB sin x + 7), $B sin d(x — 7), X= C08 £6 cos Ha + 7), G23) for then £=L, X= K follow. Alternatively we could remark that the Ds,,,(4-!) form a complete orthonormal basis for functions defined ‘on SUQ).” Explicitly this latter term refers to functions of @, 6 such that a 5) (5 belongs to SU(2), or simply such that lal? + Jol? Now from (7) and (23), we have benim,2160 so that funetions of a, b such that lal? + |b? = xf+Ptdtx2=1 can be read as functions defined on the unit 4-sphere. This remark is of course what underlies the identi- fication of the Ds,,, with the basis of the representa: tion (j,/) or Ry It is perhaps worth emphasizing that the ,Yia(08) or the Dt,,, play two very different roles being on the one hand’ closely related to matrix elements of the representation matrices of O, and on the other hand closely related to bases functions of certain representa- tions of Oy. 4. THE. LORENTZ, ‘TRANSFORMATION AND SPIN=¢ SPHERICAL HARMONICS A. Conformal Mappings Up to this time the discussion of the pherical harmonics has been based on their relationship to the rotation group. The rigid rotations are a three- parameter group of isometric mappings of the unit sphere onto itself, Thus dst = dO? + sin? 9 dg* = di" + sin? O'ds? (4.1) if the mapping {9, 6) > (0', 4°} is a rigid rotation, In order to relate the spin-s spherical functions to the Lorentz group it is necessary to enlarge this group of homeomorphic mappings of the 2sphere. ‘The mapping (0, 6} > (@' $' is conformal if dst = df? + sin® 0 dpe KUO, $¥d0"® + sin? 0" dg), (42) Clearly the rigid rotations form that subgroup of the ‘conformal transformations for which the conformal factor A? = 1. The conformal proup, which preserves, the angle between two curves and its direction, can be shown to be a sixcparameter Lie group which is isomorphic to the proper homogeneous Lorentz group? The result can be easily derived and as it introduces the notation we wish to use in our discussion of spin-s spherical functions, we give the proof here In terms of the stereographic coordinates {= e cot 9/2 which were introduced in Sec. 2, the metric on the unit sphere has the form dst = 41 + C8)* a dl. aD ‘The complex coordinate { defines @ point in the complex plane. Therefore, the conformal trans- formations of the complex plane will induce the conformal transformations of the unit sphere onto itself, The only transformations with a simple pole and a simple zero at the new north and south poles, ERK Sachs, Phys, Rev, 128, 385 (1982) GOLDBERG ef al, respectively, are given by the Mobius transformation V=aGL+ DIGL+S: 2 — fy = 1. 44) Applying this transformation to Eq. (4:3) we find ds = KP = KML VEY aE, (4.5) ce tO t+ N+ E+ OL4I gy 1+ ‘The complex constants a, By y, and 6 together with the restriction indicated in Eq. (4.4) represent six real purameters To show the isomorphism of Eq. (4.3) with the proper homogeneous Lorentz group, we introduce 4 two-dimensional complex linear vector space. Let 14, and st be the components of a vector inthis space To each transformation (4) there corresponds a transformation of SLQ) as follows: Wy, = aay + Buy, uy = yy + buy, (4.7) as can be seen by the identification £ = wy/uy. SL(2) furnishes a double covering of the conformal trans- formation exactly as it furnishes « double covering of the proper homogeneous Lorentz group. Thus the required isomorphism is established. B. The Irreducle Representations ‘Ds! If € and 7 are two independent basis vectors in the two-dimensional spinor space [the space of vectors (uy, tq) which satisfy the transformation law (4.7)], then a basis for the Iinear vector space defining the irreducible representation of the Lorentz group denoted by!® 1)! is given by (Gin Brnpn), O [s| they form a redundant set of functions for definite spin-weight s, However, the spin-s spherical tens SARE HARMONICS AND 6 2161 harmonies ,¥1,. do form an orthonormal set for fixed 5. It is easy to show that for I< Z the ,¥,,, are given uniquely by the 2%) win ES az, 403) seleL, |m| < ts Em Sim &, prema ft — 3° ae i “ woosop ace "(5') las \/ bat “(ogee ml (osm =a) % Somasmytetens (4.14) Pasi (hn sm, 1— 514m) da) and the dj, are the constants defined in Eq. (2.11) For fixed s and Z the coefficients ,Bf,""* form nonsingular “ast VLtst x G-stVLts4 matrix [(l,m), (m,,m,)] connecting the ,Z%,,, to the Yim Since the ,Z2,,, transform under the plicondeant representation of the Lorentz group up to the factor K-e4, it follows that the .Yia(ls| <7< LE and |m| <1} transform under an equivalent representation up to the same factor The above results hold both for Land s integral, or half-integral.