WorksPhysics and Astronomy
Geometric Formulation of Covariant Spin Operator
Abstract
In relativistic quantum theory, Lorentz boosts induce momentum-dependent Wigner rotations, complicating the coupling of particle spins in arbitrary frames. We describe particle spin spaces as fibers of an associate bundle over the velocity manifold, and different particle's spins residing at distinct fibers. Analogy to general relativity, we introduce parallel transport to bring spins at different velocities into a common frame, where they can be coupled. Using spinor-helicity variables, we construct the parallel transport operator along geodesics and show how it is uniquely determined by little group covariance. This construction gives explicit form of the covariant spin and orbital operators, and yields a covariant 3-point amplitude basis with definite spin and orbital angular momentum. Higher-point amplitudes can be constructively obtained and applied to the covariant partial wave expansion for amplitudes, form factors, and correlation functions, etc.
Introduction
Introduction. Spin is a fundamental, quantized property of an elementary particle. In quantum mechanics, spin transforms as intrinsic angular momentum under spatial rotations, but its behavior under boosts differs between nonrelativistic and relativistic theories. In nonrelativistic quantum theory, the spin operator, , is invariant under Galilean boosts, which therefore do not mix spin and orbital degrees of freedom. In relativistic quantum theory, by contrast, spin operators such as the Newton–Wigner spin operator [1, 2] generally undergo momentum-dependent Wigner rotations under Lorentz boosts. Consequently, a Lorentz boost generally mixes spin and orbital degrees of freedom.
In multiparticle systems, coupling individual spins to form states of definite total spin is essential for describing scattering, decays, and bound states. In the nonrelativistic setting, Galilean invariance allows spins to be coupled using Clebsch–Gordan coefficients independently of the system’s overall motion. In the relativistic setting, however, spin transformations follow induced representations of the Poincaré group and involve momentum-dependent Wigner rotations. Although spins can be coupled in the center-of-mass (COM) frame, extending the spin coupling to arbitrary frames while keeping its covariant form has remained a longstanding challenge.
We realize the relativistic spin is defined locally in the momentum space, treated as the bundle structure over the velocity manifold. Each velocity carries a spin space, represented by a fiber of an associated bundle. Spins at different velocities must be referred to a common frame before they can be coupled covariantly. Analogy to parallel transport in general relativity, we define the transport to carry spins in different fibers into a common point along the geodesic path. This construction defines an additive covariant spin operator for a multiparticle system and a corresponding orbital operator. It also provides a geometric route from Wigner states to a covariant orbital–spin amplitude basis.
Wigner state, spin operator and fiber bundle
In relativistic quantum theory a massive elementary particle is defined as a unitary irreducible representation of the Poincaré group, and Wigner’s method of induced representations [3, 4] builds it from two ingredients. The first is the little group , the Lorentz subgroup that stabilizes a chosen reference velocity : its unitary irreducible representations label the rest-frame states . The second is the coset space , whose representatives label the velocities up to a rest-frame rotation , the momentum-dependent states being obtained by the boost action, with . The set of all four-velocities , that is, the coset space , forms the velocity manifold
Its Riemannian metric is on tangent vectors, and it has constant curvature. A fixed-mass momentum shell is a rescaled copy of ; particles of different masses are therefore compared on the same normalized manifold, not on one fixed-mass shell. The little group is the stabilizer of a chosen reference point of . At any other point of , with , the corresponding stabilizer is the conjugate subgroup . Every admits the coset decomposition with , and two choices of the boost representative differ by a rest-frame rotation ; that rotation is the freedom of choosing the spin axis at the velocity , and geometrically it is a gauge transformation [5].
Lifting this coset picture to a bundle, forms a principal -bundle over the coset base manifold , with each velocity point equipped with a little-group fiber. A point of the bundle can be written ; choosing the standard boosts and using the coset decomposition identifies two equivalent bundle structures [6, 7] , which is a point . A right multiplication with defines the right action: it changes the section,
, and the fiber coordinate, , since ; is the gauge freedom that rotates the spin axis at fixed velocity and does not move . Left action on is defined by a Lorentz multiplication from the left, . Using the coset decomposition with the Wigner rotation , one has : the left action moves the velocity, , and rotates the fiber by as shown in Figure 1.

Figure 1. The coset decomposition on principal bundle. The left action of carries the frame into fiber over , and the Wigner rotation is the right multiplication with .
The above tells that carries a left -action and a right -action, which act on the two indices of separately; the group element is therefore conveniently labelled by the spinor [8, 9]. In the spinor-helicity formalism [10, 11, 12, 13] it reads
with the little-group index and the Lorentz indices. Once we fix to be the standard boost , the spinor transforms covariantly under the -action as , and under the -action as .
The spin operator is the infinitesimal generator of the right action of , that is, of the transformation that rotates the spin axis at fixed velocity. Using the coordinate on the principal bundle, its differential realization in terms of vector field is:
where parentheses denote symmetrization in , and c.c. denotes the complex conjugation and labels the spherical basis of Pauli matrices.
Once a section is chosen, the spin operator can also be written in terms of the Poincaré generators. With the Lorentz generators , the Pauli–Lubanski vector is , and the polarization vectors of the section are ; expanding Eq. (3) in the spinor coordinates then gives
Taking to be the canonical boost gives the Newton-Wigner spin operator .
The principal bundle has an associated bundle to describe massive spin- particles, with the spin space carrying spin- irreducible representation of . The fiber at velocity is then the representation space of the spin operator . A spin- particle state is built from Wigner states together with a wave function over ,
with the Lorentz-invariant measure on . These states span the Hilbert space , that is, the space of square-integrable functions on .
Since is associated to through , the vertical vector fields of Eq. (3) act on the fiber through the differential of , yielding the fiber-wise operators , where are the spin- angular-momentum matrices in the spherical basis. The unitary isomorphism Eq. (5), which identifies the component of the wave function with the coefficient of , gives the quantum operator action .
From Eq. (3) and Eq. (4), the spin operator is defined relative to the chosen , and therefore depends on the choice of spin axes. This is what obstructs a direct coupling of two particle spins. Under , the spin operator rotates according to , with the adjoint (spin-one) representation: it is gauge covariant rather than gauge invariant, carrying one adjoint little-group index. A sum of two such operators is therefore not covariant: under a change of frame, becomes
because the two frames can be redefined independently, by and by ; the sum acquires two different rotations and is no longer of the same form. No natural unified choice of the two frames exists that would repair this, so a direct coupling of the two spins is impossible; they can be coupled only after being referred to a common frame.
Spin transport along geodesic. For the spin sum to transform covariantly under the gauge transformation, with , we introduce a transport that refers the spin at to the frame at . The transported sum is a little-group tensor at only if the transport compensates the two independent redefinitions of the frames, that is, only if
under and $L_v\to L_v h(v): with this gauge law the two little-group factors in the transported sum combine into the single rotation of an operator at , and the sum keeps its form. To construct , an element of obeying the gauge law of Eq. (7), we need a connection on the principal bundle. There a single element of encodes the transport for every representation, the action on then following by applying fiberwise. The canonical connection is fixed by the Cartan decomposition , with the little-group algebra: its form is the -component of the Maurer–Cartan form [7, 14, 15, 16]: , with the vertical component , while the horizontal component dropped. To compare the frames at two different velocities one integrates along a curve in joining them, and the transport is the path-ordered exponential of ,
which belongs to the little group fiber, and thus the gauge law under the right action follows.
In addition to their behavior under changes of section, both and must respect the left action of the Lorentz transformation. For the spin operator, Lorentz transformation induces the Wigner rotation of the frame at , and pushing this to through the isomorphism Eq. (5) gives
where the Wigner rotation depends on and the chosen section. Therefore, the same gives different Wigner rotations at different velocities,
so the direct sum does not generally transform by a common rotation. On the other hand, the transported sum, written schematically as , obeys , only if the transport itself compensates the two rotations,
so that the two little-group factors combine into the single rotation of an operator at .
The transport itself depends on the curve chosen for it. The transport relation in Eq. (11) would determine the path to be geodesic. Let us start from each side of the Eq. (11), there are two unrelated curves, from to and from to , and nothing connects them. To make it hold we must therefore require the curves themselves to be related, point by point (Figure 2)

Figure 2. The principal bundle over the velocity manifold : is the shaded patch, the dotted lines are the fibers . The path joins the velocities and , and a Lorentz transformation carries the whole configuration into , joining and . The lifts of the two paths induce the transports (blue), while the frames at and , and at and , are related by the Wigner rotations (red). In both diagrams the arrows commute: the two ways from to compose to the same map, Eq. (12) on the base and Eq. (11) on the fibers.
This condition fixes the curve uniquely: it is the geodesic path
which is built from the two velocities alone and therefore satisfies Eq. (12) by construction; its parameter is in general not affine. The corresponding geodesic boost is B(v,u)=\frac{1+\slashed{v}\slashed{u}}{\sqrt{2(1+u\cdot v)}}, and the transport along the geodesic could be evaluated although quite complicated. In this work, however, we would adapt more elegant way to construct the covariant spinor form of the transport in the principal bundle from symmetry argument, irrelevant of the explicit form of the geodesic and Lorentz boost, provided in supplementary material.
Transport and on-shell amplitudes in spinor formalism. The transport is constrained by Eq. (11), which indicates that carries one little-group index at and one at , so that can be written as a function of the spinors and . For constituents of spin the transport is a matrix ; like the form of scattering amplitudes below, it is a little-group tensor, carrying one index at and one at .
Just like the three-point amplitudes are uniquely determined by the Poincare symmetry [8, 17, 18], we note two requirements determine the amplitude form of the transport completely, although only two momenta are involved in . (i) Symmetry: a Lorentz-invariant object carrying one index at each leg can only be built from the two spinors and , and the only invariant contractions with one index at each leg are and , a mixed contraction carries a Lorentz index and is excluded, while every other contraction is a scalar that belongs to the coefficient,
(ii) Kinematics: choosing the canonical section and , we note in the COM frame the spin coupling is trivial, , which fixes and the normalization,
The normalization comes from the in Eq. (2): for the physical momenta and , the above carries dimension . Eq. (15) is the unique matrix carrying one index at and one at and reducing to in the COM frame, and the horizontal lift along a geodesic has precisely these properties by construction. The symmetric construction above is consistent with the earlier geometric construction using as in Eq. (11). For spin it acts on the symmetric index set as the symmetrized product
so that a single spin- object generates every spin.
With at hand the spins of several particles can be brought into a single frame. Transporting each little-group index separately gives the transported spin operator,
the general case being the symmetric product of Eq. (16), the little-group indices at being contracted in favour of those at , the transported spin operators all carry the same indices and can be added,
and coupled by ordinary SU(2) Clebsch–Gordan coefficients.
The amplitude as a linear functional. In the spinor formalism, the basic ingredient is the 3-point on-shell amplitudes, and higher-point amplitudes can be constructed using the recursive relation or the bootstrap method [19, 20].
Let us start with three-point amplitudes in the principal bundle. In the particle-state description the 3-point amplitude is , with three velocities in the manifold , and . Using the unitary equivalence Eq. (5) between particle states and sections of , and multiplying the (distributional) matrix element by wave functions and integrating, this becomes the probability amplitude, with and ,
which is linear in and , antilinear in .
The amplitude is built in two steps. First, the and transport the constituents little group indices of the spin parts in and , to the ones in , respectively,
where is an tensor at , and it maps the spins and carried by and onto the spin through the ordinary Clebsch–Gordan series
The space corresponds to the irreducible component of . Second, every index carried by belongs to indices at , corresponding to the amplitude , whose unique form is determined to be
with totally symmetric indices; in the COM frame it is the spherical harmonic . That it carries orbital angular momentum follows from
Massive spin-spinor amplitudes. Coupling the orbital part to the spin part as in Eq. (21), the master three-point covariant orbital-spin (CLS) amplitude is given by
This expression organizes three-point amplitudes by their orbital-spin content, complementing the spinor structures of Arkani-Hamed, Huang, and Huang (AHH) [8]. In the COM frame, these CLS amplitudes reproduce the orbital-spin couplings used in the Jacob–Wick formalism [21, 22, 23]. This CLS form is completely covariant under the covariant spin and orbital operator action in Eq. (23), without any boost dependence shown in literature [24, 25, 26, 27].
In the AHH formalism, the little group space indices are converted into the Lorentz space ones via the spinors of external particles. While in this work, built in little-group space rather than from Lorentz invariants, the CLS amplitudes carry well-defined spin and orbital quantum numbers and form an independent, complete basis. For example, for basic three massive-gauge-boson amplitudes, one could write down totally eight amplitudes , , , etc. However, CLS amplitudes only contain seven independent elements with $(L,S)=(1,0),(0,1),(1,1),(2,1),(1,2),(2,2),… (3, 2), which are the combination of the above eight amplitudes. Similarly for all other basic three-point amplitudes.
Fermion–antifermion bilinears illustrate the spin–orbital classification. For an equal-mass pair, such as , in a charge-conjugation eigenstate, and . The spatial vector current has and admits and , corresponding to the and channels. The pseudoscalar channel has and . Near threshold, these include the leading ortho- and parapositronium configurations, respectively. In literature [21, 22, 23], the structure is only hold in the COM frame, however, in this work both and hold for any momenta. The supplementary material gives the bilinear decomposition. This can be extended to covariant multipole expansion for general form factors [28].
Given 3-point amplitude basis, higher-point amplitudes can be constructively obtained by recursive relation and bootstrap method [19, 20]. The advantage of this amplitudes is that it is the partial wave eigen-states with covariant orbital and spin eigen-values. Therefore, partial wave analysis can be performed covariantly, without the need of boosting back to the successive COM frames in the decay-chains. This covariant partial-wave analysis has been applied [29] to , to help identify all the immediate particles: , , , , , , and [30, 31].
The above amplitudes can be extended [32] to the ones with massless constituents. When particle 2 becomes massless, both and are absent; instead the helicity of particle 2 couples to orbital momentum, or depending on the sign of . The full amplitude couples this orbital factor to the spin sector of particle 1, described by . M = C^{\{K\}\{I\}\{J\}}_{\{K\}\{J\}\{I\}}Y^L_L\tau^{2s_{\{I\}}_{\{J\}\{I'\}}}. If both constituents become massless, the amplitude is dictated by the spin of the massive final state: .
Summary and Outlook
By identifying the spin fiber and parallel transport over the velocity manifold, we have constructed a new covariant spin operator, which couples different particle’s spins in a Lorentz covariant way. The resulting three-point structures organize amplitudes by orbital and spin angular momentum while retaining an explicit prescription for comparing spin frames. Therefore, by introducing the covariant spin and orbital operators in relativistic quantum theory, there is no mixture between the covariant spin and orbital structures under the Lorentz boost. As application, higher-point amplitude, as the partial wave basis, can be constructed from the three-point amplitude basis, which can be applied to high energy scattering, hadronic physics, nucleon structure, and gravitational wave calculation, etc.
Supplementary Materials
The associated bundle, its sections and frames. Let carry the irreducible unitary representation of the little group , and let be the principal -bundle over the velocity manifold , with projection . For and the associated bundle is the set of pairs modulo the identification
where an element of is written , the pair and its class being identified in the notation; it lies over and carries the spin state . The spin- field is a section of this bundle, that is, a smooth map with . Once a section of is chosen, any element over reads with , so every class over has the representative and the fiber is the spin space at velocity ; a section is then the same thing as a map , through , and replacing by the equally admissible lift , , changes it according to
since must equal ; this is (26), the little-group redefinition of the spin axis.
Fixing an orthonormal basis of (the representation being taken in the standard basis with respect to it), the same trivialization provides a global frame of ,
with corresponding to the Wigner state , , which is the distributional section localized at ; the choice of is precisely the choice of its rest-frame axes.
The one-particle Hilbert space is the space of square-integrable sections,
where is the Lorentz-invariant measure on and the -invariant Hermitian form on . The integrand is independent of the chosen : by (26) both representatives acquire , and
by unitarity of . In a fixed trivialization, (28) gives the unitary isomorphism of (5). The particle-state description is the more basic of the two: the states are distributional, -function normalized, and need not be normalizable, whereas contains only normalizable states; the two are nevertheless physically equivalent, agreeing on all matrix elements of bounded observables, and we pass freely between them.
The left action of on itself descends to the associated bundle,
thus Lorentz transformation carries into a new wave function , whose value at the velocity is ; its value at is therefore obtained with ,
On the particle states the same operator reads
which is the standard Wigner formula; the two forms are identified by the isomorphism of Eq. (5).
Spin as the generator of the right action. For the derived representation on the fiber is
On the principal bundle the same element generates the vertical vector fields of Eq. (3) of the main text, and , which close on and move along its fiber, since stabilizes . Carried to every fiber by the trivialization of Eq. (25), the same element acts through , yielding the fiber-wise generator
so that in the same trivialization for every . Its components along the spherical basis , normalized by , are the fiber-wise operators of the main text. Since the trivialization is built from , the operator depends on the choice of standard boosts and is not canonically defined on .
The right action on does not descend to : the prescription is not well defined, because applied to the equivalent representative it gives
which differs from unless is conjugation invariant, i.e. unless is Abelian. What the right action induces instead is an action on the trivialization itself, , whose infinitesimal form is Eq. (33). Consequently is gauge covariant rather than gauge invariant: under one has . The corresponding Hilbert-space operator acts pointwise:
As a cross-check that is indeed the spin operator, note that it has a covariant realization in terms of the Pauli–Lubanski vector. The polarization vectors are those of the section ,
built on the spherical basis , . At the rest velocity one has , so Eq. (36) gives , while the Pauli–Lubanski vector there is by Eq. (4). The covariant operator Writing the covariant form of the operator as , one therefore has
the spherical component of the spin: this is thus indeed the spin operator.
The transport operator, Wigner rotation and holonomy. In above the spin is defined at a single velocity. To compare two such spins one must carry the frame from one velocity to another via transport.
The horizontal lift and the transport. The connection specifies which curves in are horizontal, lifting a path in to a path in , and the frame that is carried along this lift is what we call the transport . Its horizontal lift is the unique curve with whose tangent is horizontal, . Starting at the frame , it ends at a lift of , that is, at times an element of ; that element is .
and of Eq. (8) is the integrated form of the horizontality condition. The horizontal directions are the left translates of ; the lift is unique once its starting point is fixed, and a different starting point changes by the right multiplication that relates the two frames.
The Wilson line and the Wilson loop. For a general path the transport of Eq. (8) is the Wilson line of the connection,
formally identical to the Wilson line of a gauge field, the connection playing the role of the gauge potential. Applied to the motion of an accelerated particle it reproduces the Thomas precession [33, 34]. When the open path is closed into a loop [35] by adjoining the canonical boost paths and , which are the canonical gauge trivializes, the Wilson line becomes the holonomy of a closed loop, i.e. a Wilson loop, and the Wigner rotation is the geometric phase [36, 37] of the triangle . The construction holds for any section. The transport is therefore a Wigner rotation, the rotation that induces at and that Eq. (9) of the main text carries: the trivialised form of the transport, not an independent ingredient. An arbitrary path gives a well-defined transport, but it depends on the whole path rather than on its two endpoints.
Why the path is a geodesic. Not every curve is admissible: a path cannot connect two different mass shells, so it is defined on the common velocity manifold . The covariance condition Eq. (12) is what singles out the geodesic. Take a Lorentz transformation that leaves both and invariant; it lies in their common stabiliser, and Eq. (12) then reduces to , so every point of the curve is invariant under that stabiliser. In the rest frame of the stabiliser of the pair is the of rotations about the direction of , and a curve from to left pointwise invariant by it must be the straight line joining them.
Parameterise that line by . In the rest frame of , where and , its points are
the denominator normalising the four-velocity; in a general frame this reads with , which is Eq. (13). Since , the covariance condition Eq. (12) holds automatically, so this is the geodesic of .
The transport along a geodesic and the boost. With the connection , the horizontal lift of the geodesic Eq. (13) through the frame is , indeed , so that
with the boost along the geodesic. Taking to be a pure boost, with , gives ; therefore satisfies Lorentz covariance, , and is fixed as
Substituting it into Eq. (40) gives the transport in closed form,
which is the spinor expression of Eq. (14) of the main text, written there with one index at each leg. For this boost the Wigner rotation is defined by the standard relation , that is with ; since , the two coincide, , so the transport is the Wigner rotation of the boost and not an independent ingredient.
The properties usually imposed on now follow rather than being assumed: , and because , and also from the spinor form Eq. (14) of the main text; from Eq. (42); the gauge law under ; and , which is Eq. (11) of the main text, from the -invariance of the connection, the same invariance that makes the transport depend on the endpoints only.
Three-point CLS amplitudes. We consider the processes with each at most one, writing , and for a scalar, a fermion and a vector boson. Their amplitudes are constructed in Ref. [8] and collected in Table I. That construction does not distinguish incoming from outgoing particles, whereas for our CLS amplitudes exchanging the incoming and outgoing pairs gives different amplitudes: for , for instance, the CLS amplitudes can be organised either as or as , because the LS quantum numbers are defined with respect to a chosen pair. We take LS to be the orbital and spin quantum numbers of the outgoing pair.
With Eq. (24) the three-point CLS amplitudes are constructed directly, as follows. For three scalars, , there is the single combination ,
For , that is with and , the two combinations and give
and for , with , the two combinations and give
for , with and , the four combinations , , and give
for , with , the four combinations , , and give
for , with , the single combination ,
for , with and , the single combination ,
for , with and , the three combinations , and ,
and for , with , the three combinations , and ,
For there are seven partial waves in total, as follows
where , each line being a polynomial in and in . These seven structures are equivalent to the seven collected in Table 1.
Table 1. Independent three-point amplitude structures for the indicated external particles.
From to in bilinear operators. The analysis above is not only a way of constructing amplitudes: the same structure serves to analyse form factors, since the matrix element of an operator can be resolved in the same basis.
Form factor of fermion bilinear operators. Consider, for instance, the annihilation of two spinning particles,
where is a bilinear local operator, in general not a Lorentz scalar. For instance, it may be one of
which we use as the examples below. This matrix element can be decomposed by analysis; we use the spinor formalism to assign definite quantum numbers. For a two-particle state the discrete quantum numbers are
with and the charge-conjugation and parity eigenvalues. The quantum numbers of a given operator are collected in the triple , with the total angular momentum obtained by coupling and , and , the eigenvalues above.
The form factor in Eq. (59) corresponds, in the COM frame, to a three-point amplitude in which the operator insertion plays the role of the third particle, and in the covariant spinor language this amplitude can be written in an arbitrary frame. Writing the operators above in terms of their time and space components corresponds to the decomposition at rest, : in that frame the zeroth component of a Lorentz tensor is obtained by contracting it with , and its spatial components by contracting it with the polarization vectors . The operators and , the third and fourth in Eq. (54), are components of the bilinear current $j^\mu\equiv\bar{\psi}\gamma^\mu\psi: the former is the contraction and the latter the contraction . Placed in the three-point amplitude, the operator insertions therefore read
;
;
;
. The first three are matched directly; for the fourth the matching is less obvious. Expanded with Eq. (47), it splits into two channels: combining the rearrangement
with the identity
obtained by using the Schouten identity on the first term of in Eq. (47), one immediately obtains the two channels and .
Each line is obtained in the same steps: the operator is contracted with or with , the fields and are replaced by the wavefunctions in the spinor-helicity form, and the result is reduced with the spinor techniques. This procedure applies to the remaining fermion bilinears below, where .
For the first line the contraction reduces, after integration by parts and use of the equations of motion, to . The second line holds for , the third uses the same technique as for , and the fourth follows from and .
From spinor-helicity form to non-relativistic multipole expansion
By the same procedure each of these fermion bilinears can be expanded in the amplitudes; in the COM frame the amplitude reads
where , , denote the little-group indices. The Clebsch–Gordan coefficients are fixed by the little-group algebra: for the spin-coupling ones are
and the ones are
with all index sets , , totally symmetric. The orbital part is carried by the harmonic tensors
the same objects may also be written with their little-group indices; in the COM frame they read
Here are the Pauli matrices acting on the little-group indices, to be distinguished from the spin label . For instance, for and this gives, respectively,
For the part with we use a vector index: writing the amplitude as , we represent it by
As a check, the structure reproduces the vector form,
where the last step uses the Pauli completeness relation in the trace form
written with , which turns each contracted in the second line into a factor of . With this notation the four structures read
The arguments label the orbital and spin angular momenta of each structure.
General form factors and correlation functions
The construction of the previous section generalises to a general form factor structure and to a general frame. Form factor construction. For the general form factor
with , arbitrary external states and of spins and , and the operator carrying the spinor indices of the Lorentz representation , all independent and complete Lorentz structures can be built from spinors without returning to the COM frame. It suffices to define the object that carries little-group rather than Lorentz indices, the index sets and being of rank and ,
Its inverse follows by contracting with the dual tensors,
the structures of being in one-to-one correspondence with those of . Decomposing into irreducible parts,
with , the structures of are exactly those of the amplitude for , given by Eq. (24) and expanded as
Substituting Eq. (88) into the inverse relations (87) and (85) yields the complete set of Lorentz structures of ,
The structures are therefore built from spinor variables, and their number follows from the same spin-coupling technique as that of the amplitudes [28].
Parton correlation functions. These are the special case of a form factor structure in which the two external momenta coincide and the relative momentum is carried by the internal operator,
with the density matrix. Contracting with the spinors reduces the problem to a little-group tensor, , whose independent forms follow from decomposing the direct product of all little-group indices into a scalar structure. The density matrix then serves as a general coefficient, and the approach reproduces the known quark–quark and gluon–gluon correlations.
Covariant partial wave analysis on cascade decay. Traditionally the partial-wave decomposition is performed through the coupling of the orbital and spin parts in the two-body COM frame; when the process involves cascade decay chains the momenta must be boosted and the resulting amplitudes are not covariant, the orbital and spin parts being mixed [38]. With canonical particle states the decomposition can be performed directly among particles in an arbitrary frame, without boost transformations; this partial-wave analysis [22, 23, 24] has been applied to the process [29].
This cascade decay contains several possible intermediate resonant states, with each resonance corresponding to a distinct decay chain. We consider the intermediate states that Ref. [30] reports above , namely the , , , , , and states together with the nonresonant . These components fall into three topological classes:
Topology I (-type): , , , with meson resonance decaying in ; the two leading terms are the and the nonresonant one, written . The amplitudes corresponding to these channel are defined as:
where each particle symbol stands for the particle together with its little-group indices, , , , ; the scalars carry the trivial little-group index, that is none, and appear only as labels.
Topology II (-type): , , , with an excited baryon; the candidates are the , and , and
Topology III (-type): , , , with another excited baryon; the candidates are the , and , and
In each topology every three-point amplitude can be analysed in the basis; bootstrapping them gives the total amplitude of that topology, and the amplitude of the whole decay is the sum of the amplitudes of the topologies. Using this amplitude basis together with the open-source package TF-PWA [31], events are generated and fitted; the comparison with the conventional treatments carried out in Ref. [29] shows that the results agree very well.
References
- [1]T. D. Newton and E. P. Wigner, Rev. Mod. Phys. 21, 400 (1949).
- [2]W. N. Polyzou, W. Glöckle, and H. Witala, Few Body Syst. 54, 1667 (2013), arXiv:1208.5840 [nucl-th].
- [3]E. P. Wigner, Annals Math. 40, 149 (1939).
- [4]S. Weinberg, The Quantum theory of fields. Vol. 1: Foundations (Cambridge University Press, 2005).
- [5]A Lorentz transformation L ∈ SL(2, C) acts on a four-velocity through its Hermitian matrix representative: with u ≡ uμσμ one has u ↦ L u L†, which is the double cover SL(2, C) → SO+(1, 3). The velocity reached from u* is written u = Lu*, so that u = L u* L†.
- [6]B. A. Douvnin, A. T. Fomenko, and S. P. Novikov, Modern Geometry — Methods and Applications: Part II, Graduate Texts in Mathematics, Vol. 104 (Springer, 1985).
- [7]S. Kobayashi and K. Nomizu, Foundations of Differential Geometry. Volume II, Interscience Tracts in Pure and Applied Mathematics No. 15 (Interscience Publishers, New York, 1969).
- [8]N. Arkani-Hamed, T.-C. Huang, and Y.-t. Huang, JHEP 11, 070 (2021), arXiv:1709.04891 [hep-th].
- [9]S. Dittmaier, Phys. Rev. D 59, 016007 (1998), arXiv:hep-ph/9805445.DOI
- [10]L. J. Dixon, in TASI in Elementary Particle Physics (2014) pp. 31–67, arXiv:1310.5353 [hep-ph].
- [11]H. Elvang and Y.-t. Huang, (2015).
- [12]C. Cheung, in TASI in Elementary Particle Physics (2018) pp. 571–623, arXiv:1708.03872 [hep-ph].
- [13]S. Badger, J. Henn, J. C. Plefka, and S. Zoia, Lect. Notes Phys. 1021, pp. (2024), arXiv:2306.05976 [hep-th].
- [14]S. R. Coleman, J. Wess, and B. Zumino, Phys. Rev. 177, 2239 (1969).
- [15]C. G. Callan, Jr., S. R. Coleman, J. Wess, and B. Zumino, Phys. Rev. 177, 2247 (1969).
- [16]T. Eguchi, P. B. Gilkey, and A. J. Hanson, Phys. Rept. 66, 213 (1980).
- [17]P. Benincasa and F. Cachazo, (2007), arXiv:0705.4305 [hep-th].arxiv.org/abs/0705.4305
- [18]E. Witten, Commun. Math. Phys. 252, 189 (2004), arXiv:hep-th/0312171.DOI
- [19]R. Britto, F. Cachazo, and B. Feng, Nucl. Phys. B 715, 499 (2005), arXiv:hep-th/0412308.DOI
- [20]R. Britto, F. Cachazo, B. Feng, and E. Witten, Phys. Rev. Lett. 94, 181602 (2005), arXiv:hep-th/0501052.DOI
- [21]M. Jacob and G. C. Wick, Annals Phys. 7, 404 (1959).
- [22]S. U. Chung, Phys. Rev. D 48, 1225 (1993), [Erratum: Phys.Rev.D 56, 4419 (1997)].
- [23]S. U. Chung, Phys. Rev. D 57, 431 (1998).
- [24]B. S. Zou and D. V. Bugg, Eur. Phys. J. A 16, 537 (2003), arXiv:hep-ph/0211457.DOI
- [25]B. S. Zou and F. Hussain, Phys. Rev. C 67, 015204 (2003), arXiv:nucl-th/0210164.
- [26]H.-J. Jing, D. Ben, S.-M. Wu, J.-J. Wu, and B.-S. Zou, JHEP 06, 039 (2023), arXiv:2301.01575 [hep-ph].DOI
- [27]H.-J. Jing, S.-M. Wu, and J.-J. Wu, Phys. Rev. D 110, 016014 (2024), arXiv:2405.05676 [hep-ph].
- [28]H. Huang, T. Tan, Y.-N. Wang, and J.-H. Yu, (2026), arXiv:2605.04030 [hep-ph].arxiv.org/abs/2605.04030
- [29]H. Huang, Y.-N. Wang, and J.-H. Yu, (2026), arXiv:2603.04487 [hep-ph].
- [30]M. Ablikim et al. (BESIII), JHEP 12, 033 (2022), arXiv:2209.08464 [hep-ex].DOI
- [31]Y. Jiang et al., GitHub link: https://github.com/jiangyi15/tf-pwa (2020).
- [32]In principle, massless particles have a different momentum orbit and little group, so their bundle construction must be treated separately. However, in the spinor formalism, we find the massive partial wave decomposition can be extended to amplitudes with massless particles.
- [33]L. H. Thomas, Nature 117, 514 (1926).
- [34]J. Oblak, J. Geom. Phys. 129, 168 (2018), arXiv:1710.06883 [hep-th].DOI
- [35]V. Palge and C. Pfeifer, Phys. Rev. A 109, 032206 (2024), arXiv:2310.08121 [quant-ph].DOI
- [36]M. V. Berry, Proc. Roy. Soc. Lond. A 392, 45 (1984).
- [37]Y. Aharonov and J. Anandan, Phys. Rev. Lett. 58, 1593 (1987).
- [38]K. Habermann and M. Mikhasenko, Phys. Rev. D 111, 056015 (2025), arXiv:2409.06913 [hep-ph].DOI