Introduction
Weak, long-range interactions between fermions provide a broad class of probes of physics beyond the Standard Model. The classic scalar and pseudoscalar exchange potentials were developed in Ref. [1 ] , while a systematic symmetry classification of spin-dependent interactions was given by Dobrescu and Mocioiu [2 ] . Coordinate-space forms appropriate both to macroscopic and atomic-scale experiments, including contact terms, were subsequently developed in Refs. [3 , 4 ] . The latter review provides a comprehensive account of present experimental searches.
Figure 1. Elastic scattering of two fermions via the exchange of a boson
For two spin-1/2 particles, rotational invariance permits sixteen independent nonrelativistic operator structures when scalar coefficient functions of the two independent momenta are factored out [2 ] .
The first purpose of this paper is to consider a massive neutral spin-1 field with the combined interaction with a fermion
L 1 = Z μ ′ ∑ f f ˉ γ μ ( g V f + g A f γ 5 ) f + Z μ ν ′ ∑ f f ˉ σ μ ν ( g T f + i g T ~ f γ 5 ) f , (1) \begin{aligned}
\mathcal{L}_1 = Z'_\mu\sum_f \bar{f}\gamma^\mu\left(g_V^f+g_A^f\gamma_5\right)f \\
&+ Z'_{\mu\nu}\sum_f \bar{f}\sigma^{\mu\nu}\left(g_T^f+i g_{\widetilde{T}}^f\gamma_5\right)f, \tag*{(1)}
\end{aligned} L 1 = Z μ ′ f ∑ f ˉ γ μ ( g V f + g A f γ 5 ) f + Z μν ′ f ∑ f ˉ σ μν ( g T f + i g T f γ 5 ) f , (1)
where Z μ ν ′ = ∂ μ Z ν ′ − ∂ ν Z μ ′ Z'_{\mu\nu}=\partial_\mu Z'_\nu-\partial_\nu Z'_\mu Z μν ′ = ∂ μ Z ν ′ − ∂ ν Z μ ′ . The second line of Eq. (1) is analogous to the electromagnetic dipole coupling of a fermion’s anomalous magnetic and electric dipole moments to the field-strength tensor; g T g_T g T and g T ~ g_{\widetilde{T}} g T therefore have mass dimension − 1 -1 − 1 . The explicit factor i i i in the pseudotensor term makes the operator Hermitian for real diagonal couplings. This interaction produces interference terms between the vector/axial and tensor/pseudotensor vertices.
The second purpose is to extend the same potential framework to a massive spin-2 mediator. Massive spin-2 fields and their couplings to matter have been studied extensively in the Fierz–Pauli and effective-field-theory descriptions [5 , 6 , 7 , 8 , 9 , 10 ] . The standard fermion coupling is to a symmetric conserved energy–momentum tensor. Here we also consider its symmetric axial counterpart, a Lorentz structure previously used as an interpolating current for negative-parity J P C = 2 − − J^{PC}=2^{--} J P C = 2 −− tensor mesons in QCD [11, [12 ] . This operator is distinct from the antisymmetric Dirac tensor σ μ ν \sigma^{\mu\nu} σ μν in Eq. (1) .
We derive the leading nonrelativistic potentials and determine their P P P and T T T properties. Figure 2 summarizes which of the resulting boson exchange potentials are already available in the literature and which are considered here. We also map the boson exchange interactions onto the sixteen-potential Dobrescu–Mocioiu basis presented in Ref. [2 ] . We use this mapping to obtain numerical translations of existing limits for both spin-1 and spin-2 coupling products in the small- and large-mediator-mass regimes, together with representative limits at finite mediator masses. The relations derived here between the interaction constants of the boson-exchange potentials g i g_i g i and the coefficients of the basis potentials f i f_i f i allow published exclusion plots for f i f_i f i to be rescaled to g i g k g_i g_k g i g k for arbitrary mediator masses. We therefore do not reproduce these exclusion plots in the present paper.
Figure 2. Interaction between two fermions mediated by spin-1 and spin-2 bosons. Rows and columns label the vertex type on each fermion line. For the massive spin-1 mediator of Eq. (1) these are the vector (V V V ), axial vector (A A A ), tensor (T T T ) and pseudotensor (T ~ \widetilde{T} T ) vertices; for the massive spin-2 mediator of Eq. (59) they are the energy–momentum (Θ \Theta Θ ) and axial-tensor (Θ ~ \widetilde{\Theta} Θ ) currents. Each matrix is symmetric under interchange of the two fermion lines, so the spin-1 block contains ten independent coupling products and the spin-2 block three. Only g V g V g_{V}g_{V} g V g V , g A g V g_{A}g_{V} g A g V and g A g A g_{A}g_{A} g A g A are standard Proca-exchange potentials. The spin-1 channels involving at least one field-strength (dipole) vertex are considered in Sec. III; in particular, the g A g T ~ g_{A}g_{\widetilde{T}} g A g T time-reversal-odd, parity-even (TOPE) interaction considered previously [13 , 14 , 15 ] is generalized here to finite mediator mass. In the spin-2 sector, the conventional Θ Θ \Theta\Theta ΘΘ interaction has been studied previously [5 ] and is re-derived and generalized here, while the mixed Θ Θ ~ \Theta\widetilde{\Theta} Θ Θ and Θ ~ Θ ~ \widetilde{\Theta}\widetilde{\Theta} Θ Θ channels are derived in Sec. IV.
We use natural units ℏ = c = 1 \hbar= c = 1 ℏ = c = 1 and metric η μ ν = diag ( 1 , − 1 , − 1 , − 1 ) \eta_{\mu\nu} = \operatorname{diag}(1,-1,-1,-1) η μν = diag ( 1 , − 1 , − 1 , − 1 ) throughout.
Consider elastic scattering
f 1 ( p 1 ) f 2 ( p 2 ) → f 1 ( p 1 ′ ) f 2 ( p 2 ′ ) , (2) f_1(p_1)f_2(p_2) \to f_1(p_1')f_2(p_2'), \tag*{(2)} f 1 ( p 1 ) f 2 ( p 2 ) → f 1 ( p 1 ′ ) f 2 ( p 2 ′ ) , (2)
with momentum transfer
q = p 1 ′ − p 1 = p 2 − p 2 ′ , q 0 ≃ 0. (3) q = p_1' - p_1 = p_2 - p_2', \qquad q^0 \simeq0. \tag*{(3)} q = p 1 ′ − p 1 = p 2 − p 2 ′ , q 0 ≃ 0. (3)
Throughout, p a \mathbf{p}_a p a denotes the average three-momentum of fermion a a a , and v \mathbf{v} v the relative velocity,
p a = 1 2 ( p a i n + p a o u t ) , (4) \mathbf{p}_a = \frac{1}{2}\left(\mathbf{p}_a^{\mathrm{in}}+\mathbf{p}_a^{\mathrm{out}}\right), \tag*{(4)} p a = 2 1 ( p a in + p a out ) , (4)
v = p 1 m 1 − p 2 m 2 . (5) \mathbf{v} = \frac{\mathbf{p}_1}{m_1} - \frac{\mathbf{p}_2}{m_2}. \tag*{(5)} v = m 1 p 1 − m 2 p 2 . (5)
In the center-of-mass frame it is also useful to write
P ≡ p 1 = − p 2 = m r v , (6) \mathbf{P} \equiv\mathbf{p}_1 = -\mathbf{p}_2 = m_r\mathbf{v}, \tag*{(6)} P ≡ p 1 = − p 2 = m r v , (6)
where m r = m 1 m 2 / ( m 1 + m 2 ) m_r = m_1m_2/(m_1+m_2) m r = m 1 m 2 / ( m 1 + m 2 ) is the reduced mass. With relativistically normalized external spinors, u ˉ u ≃ 2 m \bar{u}u \simeq2m u ˉ u ≃ 2 m , the Born matching convention used below is
V ~ ( q ) = − M N R ( q ) 4 m 1 m 2 , (7) \widetilde{V}(\mathbf{q}) = -\frac{\mathcal{M}_{\mathrm{NR}}(\mathbf{q})}{4m_1m_2}, \tag*{(7)} V ( q ) = − 4 m 1 m 2 M NR ( q ) , (7)
V ( r ) = ∫ d 3 q ( 2 π ) 3 e i q ⋅ r V ~ ( q ) . (8) V(\mathbf{r}) = \int\frac{\mathrm{d}^3q}{(2\pi)^3}e^{i\mathbf{q}\cdot\mathbf{r}}\widetilde{V}(\mathbf{q}). \tag*{(8)} V ( r ) = ∫ ( 2 π ) 3 d 3 q e i q ⋅ r V ( q ) . (8)
Changing the overall sign convention of the interaction Lagrangian changes the overall potential sign but not the relative operator structures or symmetry classifications.
It is useful to collect the radial distributions that occur repeatedly. We define
Y M ( r ) = e − M r 4 π r , (9) Y_M(r) = \frac{e^{-Mr}}{4\pi r}, \tag*{(9)} Y M ( r ) = 4 π r e − M r , (9)
F M ( r ) = − d Y M d r = e − M r 4 π ( 1 r 2 + M r ) , (10) F_M(r) = -\frac{\mathrm{d}Y_M}{\mathrm{d}r} = \frac{e^{-Mr}}{4\pi}\left(\frac{1}{r^2}+\frac{M}{r}\right), \tag*{(10)} F M ( r ) = − d r d Y M = 4 π e − M r ( r 2 1 + r M ) , (10)
D M ( r ) = δ 3 ( r ) − M 2 Y M ( r ) , (11) D_M(r) = \delta^3(\mathbf{r}) - M^2Y_M(r), \tag*{(11)} D M ( r ) = δ 3 ( r ) − M 2 Y M ( r ) , (11)
so that D M D_M D M is the transform of q 2 / ( q 2 + M 2 ) \mathbf{q}^2/(\mathbf{q}^2+M^2) q 2 / ( q 2 + M 2 ) . We also define
H i j ( r ) = ∫ d 3 q ( 2 π ) 3 e i q ⋅ r q i q j q 2 + M 2 = e − M r 4 π [ δ i j ( 1 r 3 + M r 2 ) − r ^ i r ^ j ( 3 r 3 + 3 M r 2 + M 2 r ) ] + 1 3 δ i j δ 3 ( r ) , (12) \begin{aligned}
H_{ij}(r) &= \int\frac{\mathrm{d}^3q}{(2\pi)^3}e^{i\mathbf{q}\cdot\mathbf{r}}\frac{q_iq_j}{\mathbf{q}^2+M^2} \\
&= \frac{e^{-Mr}}{4\pi}\left[\delta_{ij}\left(\frac{1}{r^3}+\frac{M}{r^2}\right)-\hat{r}_i\hat{r}_j\left(\frac{3}{r^3}+\frac{3M}{r^2}+\frac{M^2}{r}\right)\right]+\frac{1}{3}\delta_{ij}\delta^3(\mathbf{r}), \tag*{(12)}
\end{aligned} H ij ( r ) = ∫ ( 2 π ) 3 d 3 q e i q ⋅ r q 2 + M 2 q i q j = 4 π e − M r [ δ ij ( r 3 1 + r 2 M ) − r ^ i r ^ j ( r 3 3 + r 2 3 M + r M 2 ) ] + 3 1 δ ij δ 3 ( r ) , (12)
and
Q 12 ( r ) = ( σ 1 ⋅ σ 2 ) D M ( r ) − σ 1 i σ 2 j H i j ( r ) . (13) Q_{12}(r) = (\boldsymbol{\sigma}_1\cdot\boldsymbol{\sigma}_2)D_M(r)-\sigma_{1i}\sigma_{2j}H_{ij}(r). \tag*{(13)} Q 12 ( r ) = ( σ 1 ⋅ σ 2 ) D M ( r ) − σ 1 i σ 2 j H ij ( r ) . (13)
Thus Q 12 Q_{12} Q 12 is the transform of [ ( q × σ 1 ) ⋅ ( q × σ 2 ) ] / ( q 2 + M 2 ) [(\mathbf{q} \times\boldsymbol{\sigma}_1) \cdot(\mathbf{q} \times\boldsymbol{\sigma}_2)]/(q^2 + M^2) [( q × σ 1 ) ⋅ ( q × σ 2 )] / ( q 2 + M 2 ) . For products containing momentum operators we use the Hermitian prescription
p i f ( r ) ⟶ 1 2 { p i , f ( r ) } , (14) p_i f(\mathbf{r}) \longrightarrow\frac{1}{2}\{p_i,f(\mathbf{r})\}, \tag*{(14)} p i f ( r ) ⟶ 2 1 { p i , f ( r )} , (14)
where { A , B } = A B + B A \{A,B\}=AB+BA { A , B } = A B + B A .
Relativistic vertices and transversality
For fermion species a a a , the two types of spin-1 vertices in Eq. (1) are
Γ V / A , a μ = γ μ ( g V a + g A a γ 5 ) , (15) \Gamma^{\mu}_{V/A,a}=\gamma^\mu(g_V^a+g_A^a\gamma_5), \tag*{(15)} Γ V / A , a μ = γ μ ( g V a + g A a γ 5 ) , (15)
Γ T / T ~ , a μ ν = − 2 i q μ σ μ ν ( g T a + i g T ~ a γ 5 ) , (16) \Gamma^{\mu\nu}_{T/\widetilde{T},a}=-2iq_\mu\sigma^{\mu\nu}(g_T^a+ig_{\widetilde{T}}^a\gamma_5), \tag*{(16)} Γ T / T , a μν = − 2 i q μ σ μν ( g T a + i g T a γ 5 ) , (16)
where the sign of q q q is reversed at the second fermion line. The Proca propagator is
D μ ν ( q ) = − i q 2 − M 2 + i 0 ( η μ ν − q μ q ν M 2 ) . (17) D_{\mu\nu}(q)=\frac{-i}{q^2-M^2+i0}\left(\eta_{\mu\nu}-\frac{q_\mu q_\nu}{M^2}\right). \tag*{(17)} D μν ( q ) = q 2 − M 2 + i 0 − i ( η μν − M 2 q μ q ν ) . (17)
The vector current is conserved on shell. In addition, each field-strength tensor vertex is exactly transverse in its remaining vector index,
q ν q μ u ˉ ′ σ μ ν ( 1 , γ 5 ) u = 0 , (18) q_\nu q_\mu\bar{u}^{\prime}\sigma^{\mu\nu}(1,\gamma_5)u=0, \tag*{(18)} q ν q μ u ˉ ′ σ μν ( 1 , γ 5 ) u = 0 , (18)
because q μ q ν q_\mu q_\nu q μ q ν is symmetric and σ μ ν \sigma^{\mu\nu} σ μν is antisymmetric. Thus the longitudinal Proca term drops out from every amplitude containing at least one tensor or pseudotensor field-strength vertex. Notice that Eq. (18) does not mean that the rank-two Dirac tensor current itself is conserved; in fact
q μ u ˉ ′ σ μ ν u = i u ˉ ′ ( 2 m γ ν − K ν ) u , (19) q_\mu\bar{u}^{\prime}\sigma^{\mu\nu}u=i\bar{u}^{\prime}(2m\gamma^\nu-K^\nu)u, \tag*{(19)} q μ u ˉ ′ σ μν u = i u ˉ ′ ( 2 m γ ν − K ν ) u , (19)
with K = p ′ + p K=p^{\prime}+p K = p ′ + p .
The leading nonrelativistic bilinears required below are listed in Appendix B. We now give compact coordinate-space results. They include both possible orderings of inequivalent vertices, 1 ↔ 2 1\leftrightarrow2 1 ↔ 2 .
Vector–tensor interaction
The complete vector–tensor result, through the order retained here, is
V V T ( r ) = − ( g V 1 g T 2 m 2 + g T 1 g V 2 m 1 ) D M − ( g V 1 g T 2 m 1 + g T 1 g V 2 m 2 ) Q 12 + g V 1 g T 2 { σ 2 ⋅ ( v × r ^ ) , F M } + g T 1 g V 2 { σ 1 ⋅ ( v × r ^ ) , F M } . (20) \begin{aligned}
V_{VT}(\mathbf{r})={}&-\left(\frac{g_V^1g_T^2}{m_2}+\frac{g_T^1g_V^2}{m_1}\right)D_M\\
&-\left(\frac{g_V^1g_T^2}{m_1}+\frac{g_T^1g_V^2}{m_2}\right)Q_{12}\\
&+g_V^1g_T^2\{\boldsymbol{\sigma}_2\cdot(\mathbf{v}\times\hat{\mathbf{r}}),F_M\}\\
&+g_T^1g_V^2\{\boldsymbol{\sigma}_1\cdot(\mathbf{v}\times\hat{\mathbf{r}}),F_M\}. \tag*{(20)}
\end{aligned} V V T ( r ) = − ( m 2 g V 1 g T 2 + m 1 g T 1 g V 2 ) D M − ( m 1 g V 1 g T 2 + m 2 g T 1 g V 2 ) Q 12 + g V 1 g T 2 { σ 2 ⋅ ( v × r ^ ) , F M } + g T 1 g V 2 { σ 1 ⋅ ( v × r ^ ) , F M } . (20)
The four terms contain a momentum-suppressed central interaction V 1 V_1 V 1 , spin–spin/tensor terms V 2 + V 3 V_2+V_3 V 2 + V 3 , and the V 4 , 5 V_{4,5} V 4 , 5 spin–orbit structures. (The potentials V i V_i V i are defined in Ref. [4 ] ; see also Tables I and II.) All are P P P and T T T even.
Vector–pseudotensor interaction
For a vector vertex and a pseudotensor dipole vertex we obtain
V V T ~ ( r ) = 2 g V 1 g T ~ 2 ( σ 2 ⋅ r ^ ) F M − 2 g T ~ 1 g V 2 ( σ 1 ⋅ r ^ ) F M − g V 1 g T ~ 2 2 m 1 m 2 σ 2 j { ( p 2 × σ 1 ) i , H i j } − g T ~ 1 g V 2 2 m 1 m 2 σ 1 j { ( p 1 × σ 2 ) i , H i j } . (21) \begin{aligned}
V_{V\widetilde{T}}(\mathbf{r})={}&2g_V^1g_{\widetilde{T}}^2(\boldsymbol{\sigma}_2\cdot\hat{\mathbf{r}})F_M\\
&-2g_{\widetilde{T}}^1g_V^2(\boldsymbol{\sigma}_1\cdot\hat{\mathbf{r}})F_M\\
&-\frac{g_V^1g_{\widetilde{T}}^2}{2m_1m_2}\sigma_{2j}\{(\mathbf{p}_2\times\boldsymbol{\sigma}_1)_i,H_{ij}\}\\
&-\frac{g_{\widetilde{T}}^1g_V^2}{2m_1m_2}\sigma_{1j}\{(\mathbf{p}_1\times\boldsymbol{\sigma}_2)_i,H_{ij}\}. \tag*{(21)}
\end{aligned} V V T ( r ) = 2 g V 1 g T 2 ( σ 2 ⋅ r ^ ) F M − 2 g T 1 g V 2 ( σ 1 ⋅ r ^ ) F M − 2 m 1 m 2 g V 1 g T 2 σ 2 j {( p 2 × σ 1 ) i , H ij } − 2 m 1 m 2 g T 1 g V 2 σ 1 j {( p 1 × σ 2 ) i , H ij } . (21)
The first line is the familiar monopole–dipole V 9 , 10 V_{9,10} V 9 , 10 structure, while the second contains the q 2 V 14 ± V 15 q^2V_{14}\pm V_{15} q 2 V 14 ± V 15 combination. The interaction is odd under both P P P and T T T .
Axial-vector–tensor interaction
The axial-vector–tensor potential is
V A T ( r ) = − g A 1 g T 2 2 m 1 m 2 { σ 1 ⋅ p 1 , D M } − g T 1 g A 2 2 m 1 m 2 { σ 2 ⋅ p 2 , D M } − 2 ( g A 1 g T 2 + g T 1 g A 2 ) ( σ 1 × σ 2 ) ⋅ r ^ F M . (22) \begin{aligned}
V_{AT}(\mathbf{r})={}&-\frac{g_A^1g_T^2}{2m_1m_2}\{\boldsymbol{\sigma}_1\cdot\mathbf{p}_1,D_M\}\\
&-\frac{g_T^1g_A^2}{2m_1m_2}\{\boldsymbol{\sigma}_2\cdot\mathbf{p}_2,D_M\}\\
&-2(g_A^1g_T^2+g_T^1g_A^2)(\boldsymbol{\sigma}_1\times\boldsymbol{\sigma}_2)\cdot\hat{\mathbf{r}}F_M. \tag*{(22)}
\end{aligned} V A T ( r ) = − 2 m 1 m 2 g A 1 g T 2 { σ 1 ⋅ p 1 , D M } − 2 m 1 m 2 g T 1 g A 2 { σ 2 ⋅ p 2 , D M } − 2 ( g A 1 g T 2 + g T 1 g A 2 ) ( σ 1 × σ 2 ) ⋅ r ^ F M . (22)
It contains momentum-suppressed V 12 , 13 V_{12,13} V 12 , 13 terms and the V 11 V_{11} V 11 spin–spin structure. Both are P P P odd and T T T even.
Axial-vector–pseudotensor interaction and the TOPE potential
The axial-vector–pseudotensor cross term has an especially simple form,
V A T ~ ( r ) = g A 1 g T ~ 2 { σ 1 ⋅ v , ( σ 2 ⋅ r ^ ) F M } + g T ~ 1 g A 2 { σ 2 ⋅ v , ( σ 1 ⋅ r ^ ) F M } . (23) V_{A\widetilde{T}}(\mathbf{r})=g_A^1g_{\widetilde{T}}^2\{\boldsymbol{\sigma}_1\cdot\mathbf{v},(\boldsymbol{\sigma}_2\cdot\hat{\mathbf{r}})F_M\}+g_{\widetilde{T}}^1g_A^2\{\boldsymbol{\sigma}_2\cdot\mathbf{v},(\boldsymbol{\sigma}_1\cdot\hat{\mathbf{r}})F_M\}. \tag*{(23)} V A T ( r ) = g A 1 g T 2 { σ 1 ⋅ v , ( σ 2 ⋅ r ^ ) F M } + g T 1 g A 2 { σ 2 ⋅ v , ( σ 1 ⋅ r ^ ) F M } . (23)
It is a linear combination of the two V 6 , 7 V_{6,7} V 6 , 7 operators,
( σ 1 ⋅ v ) ( σ 2 ⋅ r ^ ) ± ( σ 1 ⋅ r ^ ) ( σ 2 ⋅ v ) . (24) (\boldsymbol{\sigma}_1\cdot\mathbf{v})(\boldsymbol{\sigma}_2\cdot\hat{\mathbf{r}})\pm(\boldsymbol{\sigma}_1\cdot\hat{\mathbf{r}})(\boldsymbol{\sigma}_2\cdot\mathbf{v}). \tag*{(24)} ( σ 1 ⋅ v ) ( σ 2 ⋅ r ^ ) ± ( σ 1 ⋅ r ^ ) ( σ 2 ⋅ v ) . (24)
Under parity, v → − v \mathbf{v}\to-\mathbf{v} v → − v and r ^ → − r ^ \hat{\mathbf{r}}\to-\hat{\mathbf{r}} r ^ → − r ^ , whereas spin is an axial vector, so Eq. (23) is P P P even. Under time reversal, v → − v \mathbf{v}\to-\mathbf{v} v → − v , σ a → − σ a \boldsymbol{\sigma}_a\to-\boldsymbol{\sigma}_a σ a → − σ a , and r ^ → r ^ \hat{\mathbf{r}}\to\hat{\mathbf{r}} r ^ → r ^ ; consequently Eq. (23) is T T T odd. Thus a single massive spin-1 boson with simultaneous axial-vector and pseudotensor couplings produces the TOPE interaction at tree level.
The g A g T ~ g_{A}g_{\tilde{T}} g A g T ~ mechanism for a P P P -even, T T T -odd (TOPE) interaction has been considered previously [13 , 14 , 15 ] . In particular, Conti and Khriplovich derived the long-range electron–nucleon interaction containing both independent TOPE structures. We show below that their result is recovered from Eq. (23) in the long-range limit, while Eq. (23) provides its finite-range generalization.
For an electron–nucleon system, take particle 1 = e e e and particle 2 = N N N , and define
A = g A e g T ~ N , B = g T ~ e g A N , (25) A = g_{A}^{e}g_{\tilde{T}}^{N}, \qquad B = g_{\tilde{T}}^{e}g_{A}^{N}, \tag*{(25)} A = g A e g T ~ N , B = g T ~ e g A N , (25)
together with
X = { σ e ⋅ v , ( σ N ⋅ r ^ ) F M } , Y = { σ N ⋅ v , ( σ e ⋅ r ^ ) F M } . (26) X = \left\{\boldsymbol{\sigma}_{e}\cdot\mathbf{v},\left(\boldsymbol{\sigma}_{N}\cdot\hat{\mathbf{r}}\right)F_{M}\right\}, \qquad
Y = \left\{\boldsymbol{\sigma}_{N}\cdot\mathbf{v},\left(\boldsymbol{\sigma}_{e}\cdot\hat{\mathbf{r}}\right)F_{M}\right\}. \tag*{(26)} X = { σ e ⋅ v , ( σ N ⋅ r ^ ) F M } , Y = { σ N ⋅ v , ( σ e ⋅ r ^ ) F M } . (26)
Equation (23) can then be written as
V A T ~ = A X + B Y = A + B 2 ( X + Y ) + B − A 2 ( Y − X ) . (27) V_{A\tilde{T}}=AX+BY=\frac{A+B}{2}(X+Y)+\frac{B-A}{2}(Y-X). \tag*{(27)} V A T ~ = A X + B Y = 2 A + B ( X + Y ) + 2 B − A ( Y − X ) . (27)
We compare this with the long-range TOPE interaction of Conti and Khriplovich given by
V = 1 4 m p m r { ( β e N + β N e ) i [ ( σ 1 ⋅ p ) ( σ 2 ⋅ p ) 1 r − 1 r ( σ 2 ⋅ p ) ( σ 1 ⋅ p ) ] − ( β e N − β N e ) ( σ 1 × σ 2 ) ⋅ L r 3 } . (28) V=\frac{1}{4m_{p}m_{r}}\left\{(\beta_{eN}+\beta_{Ne})i\left[(\boldsymbol{\sigma}_{1}\cdot\mathbf{p})(\boldsymbol{\sigma}_{2}\cdot\mathbf{p})\frac{1}{r}-\frac{1}{r}(\boldsymbol{\sigma}_{2}\cdot\mathbf{p})(\boldsymbol{\sigma}_{1}\cdot\mathbf{p})\right]-(\beta_{eN}-\beta_{Ne})\frac{(\boldsymbol{\sigma}_{1}\times\boldsymbol{\sigma}_{2})\cdot\mathbf{L}}{r^{3}}\right\}. \tag*{(28)} V = 4 m p m r 1 { ( β e N + β N e ) i [ ( σ 1 ⋅ p ) ( σ 2 ⋅ p ) r 1 − r 1 ( σ 2 ⋅ p ) ( σ 1 ⋅ p ) ] − ( β e N − β N e ) r 3 ( σ 1 × σ 2 ) ⋅ L } . (28)
where β e N \beta_{eN} β e N and β N e \beta_{Ne} β N e are the TOPE electron-nucleon dimensionless constants. m p m_{p} m p is not a dynamical fermion mass but the reference mass introduced in Ref. [14 ] through the factor 1 / ( 2 m ) 1/(2m) 1/ ( 2 m ) , with m m m conventionally fixed to the proton mass so that β \beta β is dimensionless.
For M → 0 M\to0 M → 0 ,
F 0 ( r ) = 1 4 π r 2 . (29) F_{0}(r)=\frac{1}{4\pi r^{2}}. \tag*{(29)} F 0 ( r ) = 4 π r 2 1 . (29)
Defining
D ( r ) = 1 4 π r , (30) D(r)=\frac{1}{4\pi r}, \tag*{(30)} D ( r ) = 4 π r 1 , (30)
we have
− ∇ D = r ^ F 0 , r ^ i F 0 = − i [ p i , D ] . (31) -\nabla D=\hat{\mathbf{r}}F_{0},\qquad\hat{r}_{i}F_{0}=-i[p_{i},D]. \tag*{(31)} − ∇ D = r ^ F 0 , r ^ i F 0 = − i [ p i , D ] . (31)
In the center-of-mass frame,
v = p m r , m r = m e m N m e + m N , (32) \mathbf{v}=\frac{\mathbf{p}}{m_{r}},\qquad m_{r}=\frac{m_{e}m_{N}}{m_{e}+m_{N}}, \tag*{(32)} v = m r p , m r = m e + m N m e m N , (32)
and, defining
P e = σ e ⋅ p , P N = σ N ⋅ p , (33) P_{e}=\boldsymbol{\sigma}_{e}\cdot\mathbf{p},\qquad P_{N}=\boldsymbol{\sigma}_{N}\cdot\mathbf{p}, \tag*{(33)} P e = σ e ⋅ p , P N = σ N ⋅ p , (33)
X = − i m r { P e , [ P N , D ] } , Y = − i m r { P N , [ P e , D ] } . (34) X=-\frac{i}{m_{r}}\{P_{e},[P_{N},D]\},\qquad Y=-\frac{i}{m_{r}}\{P_{N},[P_{e},D]\}. \tag*{(34)} X = − m r i { P e , [ P N , D ]} , Y = − m r i { P N , [ P e , D ]} . (34)
Since [ P e , P N ] = 0 [P_{e},P_{N}]=0 [ P e , P N ] = 0 , the symmetric combination becomes
X + Y = − 2 i m r [ P e P N , D ] = − i 2 π m r [ ( σ e ⋅ p ) ( σ N ⋅ p ) , 1 r ] . (35) \begin{aligned}
X+Y=-\frac{2i}{m_{r}}[P_{e}P_{N},D] \\
&=-\frac{i}{2\pi m_{r}}\left[(\boldsymbol{\sigma}_{e}\cdot\mathbf{p})(\boldsymbol{\sigma}_{N}\cdot\mathbf{p}),\frac{1}{r}\right]. \tag*{(35)}
\end{aligned} X + Y = − m r 2 i [ P e P N , D ] = − 2 π m r i [ ( σ e ⋅ p ) ( σ N ⋅ p ) , r 1 ] . (35)
⇒ i [ ( σ e ⋅ p ) ( σ N ⋅ p ) , 1 r ] = − 2 π m r ( X + Y ) (36) \Rightarrow i\left[(\boldsymbol{\sigma}_{e}\cdot\mathbf{p})(\boldsymbol{\sigma}_{N}\cdot\mathbf{p}),\frac{1}{r}\right]=-2\pi m_{r}(X+Y) \tag*{(36)} ⇒ i [ ( σ e ⋅ p ) ( σ N ⋅ p ) , r 1 ] = − 2 π m r ( X + Y ) (36)
For the antisymmetric combination, define
A j ( r ) = r ^ j F 0 ( r ) . (37) A_{j}(r)=\hat{r}_{j}F_{0}(r). \tag*{(37)} A j ( r ) = r ^ j F 0 ( r ) . (37)
Then we can write
Y − X = 1 m r [ σ N i σ e j { p i , A j } − σ e i σ N j { p i , A j } ] . (38) Y-X=\frac{1}{m_{r}}\left[\sigma_{N i}\sigma_{e j}\{p_{i},A_{j}\}-\sigma_{e i}\sigma_{N j}\{p_{i},A_{j}\}\right]. \tag*{(38)} Y − X = m r 1 [ σ N i σ e j { p i , A j } − σ e i σ N j { p i , A j } ] . (38)
Since spin matrices belonging to different particles commute,
Y − X = 1 m r ( σ N i σ e j − σ e i σ N j ) { p i , A j } . (39) Y-X=\frac{1}{m_{r}}\left(\sigma_{N i}\sigma_{e j}-\sigma_{e i}\sigma_{N j}\right)\{p_{i},A_{j}\}. \tag*{(39)} Y − X = m r 1 ( σ N i σ e j − σ e i σ N j ) { p i , A j } . (39)
Using
{ p i , A j } = 2 A j p i − i ∂ i A j , (40) \{p_{i},A_{j}\}=2A_{j}p_{i}-i\partial_{i}A_{j}, \tag*{(40)} { p i , A j } = 2 A j p i − i ∂ i A j , (40)
the derivative is
∂ i A j = F 0 ( r ) r δ i j + [ F 0 ′ ( r ) − F 0 ( r ) r ] r ^ i r ^ j . (41) \partial_{i}A_{j}=\frac{F_{0}(r)}{r}\delta_{ij}+\left[F_{0}'(r)-\frac{F_{0}(r)}{r}\right]\hat{r}_{i}\hat{r}_{j}. \tag*{(41)} ∂ i A j = r F 0 ( r ) δ ij + [ F 0 ′ ( r ) − r F 0 ( r ) ] r ^ i r ^ j . (41)
This tensor is symmetric under i ↔ j i\leftrightarrow j i ↔ j , whereas
C i j ≡ σ N i σ e j − σ e i σ N j (42) C_{ij}\equiv\sigma_{N i}\sigma_{e j}-\sigma_{e i}\sigma_{N j} \tag*{(42)} C ij ≡ σ N i σ e j − σ e i σ N j (42)
is antisymmetric. Consequently,
C i j ∂ i A j = 0. (43) C_{ij}\partial_{i}A_{j}=0. \tag*{(43)} C ij ∂ i A j = 0. (43)
Thus all terms in which the momentum differentiates the radial function cancel, and Eq. (39) reduces to
Y − X = 2 F 0 ( r ) m r ( σ N i σ e j − σ e i σ N j ) r ^ j p i . (44) Y-X=\frac{2F_{0}(r)}{m_{r}}\left(\sigma_{N i}\sigma_{e j}-\sigma_{e i}\sigma_{N j}\right)\hat{r}_{j}p_{i}. \tag*{(44)} Y − X = m r 2 F 0 ( r ) ( σ N i σ e j − σ e i σ N j ) r ^ j p i . (44)
Using
σ N i σ e j − σ e i σ N j = − ϵ i j k ( σ e × σ N ) k , r ^ × p = r × p r = L r , (45) \sigma_{N i}\sigma_{e j}-\sigma_{e i}\sigma_{N j}=-\epsilon_{ijk}(\boldsymbol{\sigma}_{e}\times\boldsymbol{\sigma}_{N})_{k},\qquad\hat{\mathbf{r}}\times\mathbf{p}=\frac{\mathbf{r}\times\mathbf{p}}{r}=\frac{\mathbf{L}}{r}, \tag*{(45)} σ N i σ e j − σ e i σ N j = − ϵ ij k ( σ e × σ N ) k , r ^ × p = r r × p = r L , (45)
we obtain
( σ N i σ e j − σ e i σ N j ) r ^ j p i = − ϵ i j k ( σ e × σ N ) k r ^ j p i = ( σ e × σ N ) ⋅ L r . (46) \begin{aligned}
\left(\sigma_{N i}\sigma_{e j}-\sigma_{e i}\sigma_{N j}\right)\hat{r}_{j}p_{i}=-\epsilon_{ijk}(\boldsymbol{\sigma}_{e}\times\boldsymbol{\sigma}_{N})_{k}\hat{r}_{j}p_{i} \\
&=(\boldsymbol{\sigma}_{e}\times\boldsymbol{\sigma}_{N})\cdot\frac{\mathbf{L}}{r}. \tag*{(46)}
\end{aligned} ( σ N i σ e j − σ e i σ N j ) r ^ j p i = − ϵ ij k ( σ e × σ N ) k r ^ j p i = ( σ e × σ N ) ⋅ r L . (46)
Therefore,
Y − X = 2 F 0 ( r ) m r r ( σ e × σ N ) ⋅ L = 1 2 π m r ( σ e × σ N ) ⋅ L r 3 . (47) Y-X=\frac{2F_0(r)}{m_r r}(\boldsymbol{\sigma}_e\times\boldsymbol{\sigma}_N)\cdot\mathbf{L}=\frac{1}{2\pi m_r}\frac{(\boldsymbol{\sigma}_e\times\boldsymbol{\sigma}_N)\cdot\mathbf{L}}{r^3}. \tag*{(47)} Y − X = m r r 2 F 0 ( r ) ( σ e × σ N ) ⋅ L = 2 π m r 1 r 3 ( σ e × σ N ) ⋅ L . (47)
Substituting Eqs. (36) and (47) into Eq. (28) gives
V = 1 4 m p m r { − ( β e N + β N e ) 2 π m r ( X + Y ) − ( β e N − β N e ) 4 π m r F 0 m r r ( σ 1 × σ 2 ) ⋅ L } . (48) V=\frac{1}{4m_p m_r}\left\{-(\beta_{eN}+\beta_{Ne})2\pi m_r(X+Y)-(\beta_{eN}-\beta_{Ne})4\pi m_r\frac{F_0}{m_r r}(\boldsymbol{\sigma}_1\times\boldsymbol{\sigma}_2)\cdot\mathbf{L}\right\}. \tag*{(48)} V = 4 m p m r 1 { − ( β e N + β N e ) 2 π m r ( X + Y ) − ( β e N − β N e ) 4 π m r m r r F 0 ( σ 1 × σ 2 ) ⋅ L } . (48)
V = − π ( β e N + β N e ) 2 m p ( X + Y ) − π ( β e N − β N e ) m p F 0 m r r ( σ 1 × σ 2 ) ⋅ L . (49) V=-\frac{\pi(\beta_{eN}+\beta_{Ne})}{2m_p}(X+Y)-\frac{\pi(\beta_{eN}-\beta_{Ne})}{m_p}\frac{F_0}{m_r r}(\boldsymbol{\sigma}_1\times\boldsymbol{\sigma}_2)\cdot\mathbf{L}. \tag*{(49)} V = − 2 m p π ( β e N + β N e ) ( X + Y ) − m p π ( β e N − β N e ) m r r F 0 ( σ 1 × σ 2 ) ⋅ L . (49)
Using Eq. (47) , we can write our axial-pseudotensor TOPE equation Eq. (27) as
V A ~ T = A + B 2 ( X + Y ) + ( B − A ) F 0 m r r ( σ 1 × σ 2 ) ⋅ L . (50) V_{\widetilde{A}T}=\frac{A+B}{2}(X+Y)+(B-A)\frac{F_0}{m_r r}(\boldsymbol{\sigma}_1\times\boldsymbol{\sigma}_2)\cdot\mathbf{L}. \tag*{(50)} V A T = 2 A + B ( X + Y ) + ( B − A ) m r r F 0 ( σ 1 × σ 2 ) ⋅ L . (50)
Comparing Eq. (50) with Eq. (49) , we get the relation
A + B 2 = − π ( β e N + β N e ) 2 m p , B − A = − π ( β e N − β N e ) m p . (51) \frac{A+B}{2}=-\frac{\pi(\beta_{eN}+\beta_{Ne})}{2m_p},\qquad B-A=-\frac{\pi(\beta_{eN}-\beta_{Ne})}{m_p}. \tag*{(51)} 2 A + B = − 2 m p π ( β e N + β N e ) , B − A = − m p π ( β e N − β N e ) . (51)
Solving this, we obtain
B = g T ~ 1 g A N = − π m p β e N , A = g A e g T ~ N = − π m p β N e . (52) B=g_{\widetilde{T}}^{1}g_{A}^{N}=-\frac{\pi}{m_p}\beta_{eN},\qquad A=g_{A}^{e}g_{\widetilde{T}}^{N}=-\frac{\pi}{m_p}\beta_{Ne}. \tag*{(52)} B = g T 1 g A N = − m p π β e N , A = g A e g T N = − m p π β N e . (52)
β e N \beta_{eN} β e N corresponds to the derivative vertex on the electron, whereas β N e \beta_{Ne} β N e corresponds to the derivative vertex on the nucleon.
Therefore we obtain the limits on axial–vector-pseudotensor coupling constants with the relation
∣ g T ~ e g A N ∣ = π m p ∣ β e N ∣ , ∣ g A e g T ~ N ∣ = π m p ∣ β N e ∣ . (53) \left|g_{\widetilde{T}}^{e}g_{A}^{N}\right|=\frac{\pi}{m_p}\left|\beta_{eN}\right|,\qquad\left|g_{A}^{e}g_{\widetilde{T}}^{N}\right|=\frac{\pi}{m_p}\left|\beta_{Ne}\right|. \tag*{(53)} g T e g A N = m p π ∣ β e N ∣ , g A e g T N = m p π ∣ β N e ∣ . (53)
Thus Eq. (23) reproduces the complete Conti–Khriplovich electron–nucleon TOPE interaction in the long-range limit and extends it to finite mediator mass.
Tensor–tensor, tensor–pseudotensor, and pseudotensor–pseudotensor terms
The two diagonal dipole combinations are compactly
V T T ( r ) = − 4 g T 1 g T 2 Q 12 ( r ) , (54) V_{TT}(r)=-4g_{T}^{1}g_{T}^{2}Q_{12}(r), \tag*{(54)} V T T ( r ) = − 4 g T 1 g T 2 Q 12 ( r ) , (54)
V T ~ T ~ ( r ) = 4 g T ~ 1 g T ~ 2 σ 1 i σ 2 j H i j ( r ) . (55) V_{\widetilde{T}\widetilde{T}}(r)=4g_{\widetilde{T}}^{1}g_{\widetilde{T}}^{2}\sigma_{1i}\sigma_{2j}H_{ij}(r). \tag*{(55)} V T T ( r ) = 4 g T 1 g T 2 σ 1 i σ 2 j H ij ( r ) . (55)
Both are P , T P,T P , T even. The mixed pseudotensor–tensor potential can be written
V T T ~ ( r ) = − 2 g T ~ 1 g T 2 m 2 σ 1 ⋅ ∇ D M + 2 g T ~ 1 g T 2 σ 1 i { H i j , ( σ 2 × v ) j } + 2 g T 1 g T ~ 2 m 1 σ 2 ⋅ ∇ D M − 2 g T 1 g T ~ 2 σ 2 i { H i j , ( σ 1 × v ) j } . (56) V_{T\widetilde{T}}(r)=-\frac{2g_{\widetilde{T}}^{1}g_{T}^{2}}{m_2}\boldsymbol{\sigma}_1\cdot\nabla D_M+2g_{\widetilde{T}}^{1}g_{T}^{2}\sigma_{1i}\{H_{ij},(\boldsymbol{\sigma}_2\times\mathbf{v})_j\}+\frac{2g_{T}^{1}g_{\widetilde{T}}^{2}}{m_1}\boldsymbol{\sigma}_2\cdot\nabla D_M-2g_{T}^{1}g_{\widetilde{T}}^{2}\sigma_{2i}\{H_{ij},(\boldsymbol{\sigma}_1\times\mathbf{v})_j\}. \tag*{(56)} V T T ( r ) = − m 2 2 g T 1 g T 2 σ 1 ⋅ ∇ D M + 2 g T 1 g T 2 σ 1 i { H ij , ( σ 2 × v ) j } + m 1 2 g T 1 g T 2 σ 2 ⋅ ∇ D M − 2 g T 1 g T 2 σ 2 i { H ij , ( σ 1 × v ) j } . (56)
This interaction is P , T P,T P , T odd and corresponds to the momentum-suppressed q 2 V 9 , 10 q^2V_{9,10} q 2 V 9 , 10 and q 2 V 14 ± V 15 q^2V_{14}\pm V_{15} q 2 V 14 ± V 15 structures.
A spin-2 particle is described by a symmetric rank-two field h μ ν = h ν μ h_{\mu\nu}=h_{\nu\mu} h μν = h ν μ . It therefore cannot couple linearly to the antisymmetric current ψ ˉ σ μ ν ψ \bar{\psi}\sigma^{\mu\nu}\psi ψ ˉ σ μν ψ without additional derivatives or fields. The natural lowest-derivative fermion current is instead the symmetric energy–momentum tensor. To explore parity violation we also introduce a symmetric axial-tensor current.
Relativistic interaction and currents
For fermion species a a a define
Θ a μ ν = i 4 ψ ˉ a ( γ μ ∂ ↔ ν + γ ν ∂ ↔ μ ) ψ a − η μ ν L a , (57) \Theta_a^{\mu\nu}=\frac{i}{4}\bar{\psi}_a\left(\gamma^\mu\overleftrightarrow{\partial}^{\nu}+\gamma^\nu\overleftrightarrow{\partial}^{\mu}\right)\psi_a-\eta^{\mu\nu}\mathcal{L}_a, \tag*{(57)} Θ a μν = 4 i ψ ˉ a ( γ μ ∂ ν + γ ν ∂ μ ) ψ a − η μν L a , (57)
Θ ~ a μ ν = i 4 ψ ˉ a ( γ μ γ 5 ∂ ↔ ν + γ ν γ 5 ∂ ↔ μ ) ψ a . (58) \widetilde{\Theta}_a^{\mu\nu}=\frac{i}{4}\bar{\psi}_a\left(\gamma^\mu\gamma_5\overleftrightarrow{\partial}^{\nu}+\gamma^\nu\gamma_5\overleftrightarrow{\partial}^{\mu}\right)\psi_a. \tag*{(58)} Θ a μν = 4 i ψ ˉ a ( γ μ γ 5 ∂ ν + γ ν γ 5 ∂ μ ) ψ a . (58)
We consider
L 2 = − h μ ν ∑ a ( g 2 a Θ a μ ν + g ~ 2 a Θ ~ a μ ν ) , (59) \mathcal{L}_2=-h_{\mu\nu}\sum_a\left(g_{2a}\Theta_a^{\mu\nu}+\widetilde{g}_{2a}\widetilde{\Theta}_a^{\mu\nu}\right), \tag*{(59)} L 2 = − h μν a ∑ ( g 2 a Θ a μν + g 2 a Θ a μν ) , (59)
where g 2 a g_{2a} g 2 a and g ~ 2 a \widetilde{g}_{2a} g 2 a have mass dimension − 1 -1 − 1 . For a graviton-like interaction g 2 a g_{2a} g 2 a is usually written as a dimensionless coefficient divided by a common scale [5 , 6 ] .
Between on-shell spinors, with K a = p a ′ + p a K_a=p'_a+p_a K a = p a ′ + p a ,
T a μ ν = 1 4 u ˉ a ( γ μ K a ν + γ ν K a μ ) u a , (60) T_a^{\mu\nu}=\frac{1}{4}\bar{u}_a\left(\gamma^\mu K_a^\nu+\gamma^\nu K_a^\mu\right)u_a, \tag*{(60)} T a μν = 4 1 u ˉ a ( γ μ K a ν + γ ν K a μ ) u a , (60)
T ~ a μ ν = 1 4 u ˉ a ( γ μ γ 5 K a ν + γ ν γ 5 K a μ ) u a . (61) \widetilde{T}_a^{\mu\nu}=\frac{1}{4}\bar{u}_a\left(\gamma^\mu\gamma_5K_a^\nu+\gamma^\nu\gamma_5K_a^\mu\right)u_a. \tag*{(61)} T a μν = 4 1 u ˉ a ( γ μ γ 5 K a ν + γ ν γ 5 K a μ ) u a . (61)
For equal-mass free on-shell external fermions, the matrix element of the Dirac Lagrangian vanishes by the Dirac equation, ⟨ p a ′ ∣ L a ∣ p a ⟩ = 0 \langle p'_a|\mathcal{L}_a|p_a\rangle=0 ⟨ p a ′ ∣ L a ∣ p a ⟩ = 0 , so the − η μ ν L a -\eta^{\mu\nu}\mathcal{L}_a − η μν L a term in Eq. (57) does not contribute to T a μ ν T_a^{\mu\nu} T a μν . The ordinary current is conserved,
q μ T a μ ν = 0 , (62) q_\mu T_a^{\mu\nu}=0, \tag*{(62)} q μ T a μν = 0 , (62)
whereas the axial-tensor current is traceless but nonconserved for m a ≠ 0 m_a \ne0 m a = 0 :
T ~ a μ μ = 0 , (63) \widetilde{T}_{a\mu}^{\mu}=0, \tag*{(63)} T a μ μ = 0 , (63)
q μ T ~ 1 μ ν = m 1 2 K 1 ν u ˉ 1 γ 5 u 1 , (64) q_\mu\widetilde{T}_1^{\mu\nu}=\frac{m_1}{2}K_1^\nu\bar{u}_1\gamma_5u_1, \tag*{(64)} q μ T 1 μν = 2 m 1 K 1 ν u ˉ 1 γ 5 u 1 , (64)
q μ T ~ 2 μ ν = − m 2 2 K 2 ν u ˉ 2 γ 5 u 2 , (65) q_\mu\widetilde{T}_2^{\mu\nu}=-\frac{m_2}{2}K_2^\nu\bar{u}_2\gamma_5u_2, \tag*{(65)} q μ T 2 μν = − 2 m 2 K 2 ν u ˉ 2 γ 5 u 2 , (65)
q μ q ν T ~ a μ ν = 0. (66) q_\mu q_\nu\widetilde{T}_a^{\mu\nu}=0. \tag*{(66)} q μ q ν T a μν = 0. (66)
The last identity follows from q ⋅ K a = 0 q\cdot K_a=0 q ⋅ K a = 0 in elastic scattering.
Fierz–Pauli propagator and longitudinal helicities
The massive spin-2 propagator can be written [7 , 8 ]
D μ ν , α β ( q ) = i q 2 − M 2 + i 0 P μ ν , α β ( q ) , (67) D_{\mu\nu,\alpha\beta}(q)=\frac{i}{q^2-M^2+i0}\mathcal{P}_{\mu\nu,\alpha\beta}(q), \tag*{(67)} D μν , α β ( q ) = q 2 − M 2 + i 0 i P μν , α β ( q ) , (67)
with
P μ ν , α β = 1 2 ( θ μ α θ ν β + θ μ β θ ν α ) − 1 3 θ μ ν θ α β , (68) \mathcal{P}_{\mu\nu,\alpha\beta}=\frac{1}{2}\left(\theta_{\mu\alpha}\theta_{\nu\beta}+\theta_{\mu\beta}\theta_{\nu\alpha}\right)-\frac{1}{3}\theta_{\mu\nu}\theta_{\alpha\beta}, \tag*{(68)} P μν , α β = 2 1 ( θ μα θ ν β + θ μ β θ ν α ) − 3 1 θ μν θ α β , (68)
θ μ ν = η μ ν − q μ q ν M 2 . (69) \theta_{\mu\nu}=\eta_{\mu\nu}-\frac{q_\mu q_\nu}{M^2}. \tag*{(69)} θ μν = η μν − M 2 q μ q ν . (69)
For two symmetric sources J μ ν J^{\mu\nu} J μν and K μ ν K^{\mu\nu} K μν ,
J P K = J μ ν K μ ν − 1 3 J K − 2 M 2 ( q μ J μ ν ) ( q α K α ν ) + 1 3 M 2 [ J ( q K q ) + K ( q J q ) ] + 2 3 M 4 ( q J q ) ( q K q ) , (70) \begin{aligned}
J\mathcal{P}K={}&J^{\mu\nu}K_{\mu\nu}-\frac{1}{3}JK\\
&-\frac{2}{M^2}(q_\mu J^{\mu\nu})(q^\alpha K_{\alpha\nu})\\
&+\frac{1}{3M^2}\left[J(qKq)+K(qJq)\right]\\
&+\frac{2}{3M^4}(qJq)(qKq), \tag*{(70)}
\end{aligned} J P K = J μν K μν − 3 1 J K − M 2 2 ( q μ J μν ) ( q α K α ν ) + 3 M 2 1 [ J ( q K q ) + K ( q J q ) ] + 3 M 4 2 ( q J q ) ( q K q ) , (70)
where J = J μ μ J=J^\mu_{\ \mu} J = J μ μ and q J q = q μ J μ ν q ν qJq=q_\mu J^{\mu\nu}q_\nu q J q = q μ J μν q ν . Equation (70) shows explicitly why all longitudinal pieces vanish for two conserved energy–momentum currents. They also vanish in the mixed T T ~ T\widetilde{T} T T amplitude because T T T is conserved and T ~ \widetilde{T} T is traceless with q T ~ q = 0 q\widetilde{T}q=0 q T q = 0 . In contrast, for two axial-tensor currents,
T ~ 1 P T ~ 2 = T ~ 1 μ ν T ~ 2 μ ν − 2 M 2 ( q μ T ~ 1 μ ν ) ( q α T ~ 2 α ν ) , (71) \widetilde{T}_1\mathcal{P}\widetilde{T}_2=\widetilde{T}_1^{\mu\nu}\widetilde{T}_{2\mu\nu}-\frac{2}{M^2}(q_\mu\widetilde{T}_1^{\mu\nu})(q^\alpha\widetilde{T}_{2\alpha\nu}), \tag*{(71)} T 1 P T 2 = T 1 μν T 2 μν − M 2 2 ( q μ T 1 μν ) ( q α T 2 α ν ) , (71)
so the longitudinal helicities contribute for massive fermions. Since T ~ μ ν \widetilde{T}^{\mu\nu} T μν is not conserved and generates a term proportional to 1 / M 2 1/M^2 1/ M 2 in the T ~ T ~ \widetilde{T}\widetilde{T} T T channel, the resulting 1 / M 2 1/M^2 1/ M 2 enhancement is a signal of sensitivity to its ultraviolet completion. The axial-tensor spin-2 interaction should therefore be understood as an effective description with a finite cutoff and not extrapolated to arbitrarily small mediator mass.
The master nonrelativistic potential is
V ~ 2 ( q ) = − 1 4 m 1 m 2 J 1 μ ν P μ ν , α β J 2 α β q 2 + M 2 , (72) \widetilde{V}_2(\mathbf{q})=-\frac{1}{4m_1m_2}\frac{J_1^{\mu\nu}\mathcal{P}_{\mu\nu,\alpha\beta}J_2^{\alpha\beta}}{\mathbf{q}^2+M^2}, \tag*{(72)} V 2 ( q ) = − 4 m 1 m 2 1 q 2 + M 2 J 1 μν P μν , α β J 2 α β , (72)
with J a μ ν = g 2 a T a μ ν + g ~ 2 a T ~ a μ ν J_a^{\mu\nu}=g_{2a}T_a^{\mu\nu}+\widetilde{g}_{2a}\widetilde{T}_a^{\mu\nu} J a μν = g 2 a T a μν + g 2 a T a μν .
Nonrelativistic currents
To the leading orders needed here,
T 1 00 ≃ 2 m 1 2 , T 1 0 i ≃ 2 m 1 p 1 i − i m 1 2 ( q × σ 1 ) i , (73) T_1^{00}\simeq2m_1^2,\qquad T_1^{0i}\simeq2m_1p_1^i-\frac{im_1}{2}(\mathbf{q}\times\boldsymbol{\sigma}_1)^i, \tag*{(73)} T 1 00 ≃ 2 m 1 2 , T 1 0 i ≃ 2 m 1 p 1 i − 2 i m 1 ( q × σ 1 ) i , (73)
T 2 00 ≃ 2 m 2 2 , T 2 0 i ≃ 2 m 2 p 2 i + i m 2 2 ( q × σ 2 ) i , (74) T_2^{00}\simeq2m_2^2,\qquad T_2^{0i}\simeq2m_2p_2^i+\frac{im_2}{2}(\mathbf{q}\times\boldsymbol{\sigma}_2)^i, \tag*{(74)} T 2 00 ≃ 2 m 2 2 , T 2 0 i ≃ 2 m 2 p 2 i + 2 i m 2 ( q × σ 2 ) i , (74)
and
T ~ a 00 ≃ 2 m a σ a ⋅ p a , (75) \widetilde{T}_a^{00}\simeq2m_a\boldsymbol{\sigma}_a\cdot\mathbf{p}_a, \tag*{(75)} T a 00 ≃ 2 m a σ a ⋅ p a , (75)
T ~ a 0 i ≃ m a 2 σ a i , (76) \widetilde{T}_a^{0i}\simeq m_a^2\sigma_a^i, \tag*{(76)} T a 0 i ≃ m a 2 σ a i , (76)
T ~ a i j ≃ m a ( p a i σ a j + p a j σ a i ) . (77) \widetilde{T}_a^{ij}\simeq m_a\left(p_a^i\sigma_a^j+p_a^j\sigma_a^i\right). \tag*{(77)} T a ij ≃ m a ( p a i σ a j + p a j σ a i ) . (77)
The displayed currents are sufficient for the static and mixed terms below. In deriving the spin–orbit term in Eq. (80) we additionally retain the subleading spin-dependent pieces of the ordinary energy–momentum tensor that are quadratic in nonrelativistic momenta; still higher terms are otherwise omitted.
Tensor–tensor spin-2 exchange
For two conserved currents, the leading central and static spin-dependent potential is
V ~ 22 ( q ) = − 2 g 21 g 22 m 1 m 2 3 ( q 2 + M 2 ) + g 21 g 22 8 ( q 2 + M 2 ) [ q 2 σ 1 ⋅ σ 2 − ( σ 1 ⋅ q ) ( σ 2 ⋅ q ) ] . (78) \widetilde{V}_{22}(\mathbf{q})=-\frac{2g_{21}g_{22}m_1m_2}{3(\mathbf{q}^2+M^2)}+\frac{g_{21}g_{22}}{8(\mathbf{q}^2+M^2)}\left[\mathbf{q}^2\boldsymbol{\sigma}_1\cdot\boldsymbol{\sigma}_2-(\boldsymbol{\sigma}_1\cdot\mathbf{q})(\boldsymbol{\sigma}_2\cdot\mathbf{q})\right]. \tag*{(78)} V 22 ( q ) = − 3 ( q 2 + M 2 ) 2 g 21 g 22 m 1 m 2 + 8 ( q 2 + M 2 ) g 21 g 22 [ q 2 σ 1 ⋅ σ 2 − ( σ 1 ⋅ q ) ( σ 2 ⋅ q ) ] . (78)
Therefore
V 22 ( r ) = − 2 3 g 21 g 22 m 1 m 2 Y M ( r ) + 1 8 g 21 g 22 Q 12 ( r ) + O ( v 2 ) . (79) V_{22}(r)=-\frac{2}{3}g_{21}g_{22}m_1m_2Y_M(r)+\frac{1}{8}g_{21}g_{22}Q_{12}(r)+O(v^2). \tag*{(79)} V 22 ( r ) = − 3 2 g 21 g 22 m 1 m 2 Y M ( r ) + 8 1 g 21 g 22 Q 12 ( r ) + O ( v 2 ) . (79)
At the same quadratic order in nonrelativistic momenta there is also a spin–orbit term. Keeping the terms linear in P \mathbf{P} P and q \mathbf{q} q gives
V ~ 22 S O ( q ) = i g 21 g 22 q 2 + M 2 ( a 1 σ 1 + a 2 σ 2 ) ⋅ ( P × q ) , (80) \widetilde{V}_{22}^{\mathrm{SO}}(\mathbf{q})=\frac{ig_{21}g_{22}}{\mathbf{q}^2+M^2}(a_1\boldsymbol{\sigma}_1+a_2\boldsymbol{\sigma}_2)\cdot(\mathbf{P}\times\mathbf{q}), \tag*{(80)} V 22 SO ( q ) = q 2 + M 2 i g 21 g 22 ( a 1 σ 1 + a 2 σ 2 ) ⋅ ( P × q ) , (80)
a 1 = 1 2 + m 2 3 m 1 , a 2 = 1 2 + m 1 3 m 2 . (81) a_1=\frac{1}{2}+\frac{m_2}{3m_1},\qquad a_2=\frac{1}{2}+\frac{m_1}{3m_2}. \tag*{(81)} a 1 = 2 1 + 3 m 1 m 2 , a 2 = 2 1 + 3 m 2 m 1 . (81)
After the Hermitian replacement in Eq. (14) ,
V 22 S O ( r ) = − g 21 g 22 2 { ( a 1 σ 1 + a 2 σ 2 ) ⋅ ( P × r ^ ) , F M ( r ) } . (82) V_{22}^{\mathrm{SO}}(r)=-\frac{g_{21}g_{22}}{2}\left\{\left(a_1\boldsymbol{\sigma}_1+a_2\boldsymbol{\sigma}_2\right)\cdot(\mathbf{P}\times\hat{\mathbf{r}}),F_M(r)\right\}. \tag*{(82)} V 22 SO ( r ) = − 2 g 21 g 22 { ( a 1 σ 1 + a 2 σ 2 ) ⋅ ( P × r ^ ) , F M ( r ) } . (82)
This is the V 4 , 5 V_{4,5} V 4 , 5 sector. For m 1 = m 2 m_1=m_2 m 1 = m 2 , a 1 = a 2 = 5 / 6 a_1=a_2=5/6 a 1 = a 2 = 5/6 and only the symmetric V 4 + 5 V_{4+5} V 4 + 5 combination remains. The first term in Eq. (79) is the massive-spin-2 Yukawa force; the second contains the V 2 + V 3 V_2+V_3 V 2 + V 3 spin–spin and tensor structures, including their contact completion. All terms in Eqs. (79) and (82) conserve P P P and T T T . At still higher order the conventional energy–momentum coupling also produces the q 2 V 8 q^2V_8 q 2 V 8 structure; this is consistent with the complete fermionic spin-2 expansion of Ref. [5 ] .
Tensor–axial-tensor spin-2 exchange
For one ordinary and one axial-tensor vertex we find
V ~ 22 ( q ) = g 21 g ~ 22 m 1 m 2 q 2 + M 2 σ 2 ⋅ v − i g 21 g ~ 22 m 2 4 ( q 2 + M 2 ) ( σ 1 × σ 2 ) ⋅ q − g ~ 21 g 22 m 1 m 2 q 2 + M 2 σ 1 ⋅ v − i g ~ 21 g 22 m 1 4 ( q 2 + M 2 ) ( σ 1 × σ 2 ) ⋅ q . (83) \begin{aligned}
\widetilde{V}_{22}(\mathbf{q})={}&\frac{g_{21}\widetilde{g}_{22}m_1m_2}{\mathbf{q}^2+M^2}\boldsymbol{\sigma}_2\cdot\mathbf{v}
-\frac{i g_{21}\widetilde{g}_{22}m_2}{4(\mathbf{q}^2+M^2)}(\boldsymbol{\sigma}_1\times\boldsymbol{\sigma}_2)\cdot\mathbf{q}\\
&-\frac{\widetilde{g}_{21}g_{22}m_1m_2}{\mathbf{q}^2+M^2}\boldsymbol{\sigma}_1\cdot\mathbf{v}
-\frac{i\widetilde{g}_{21}g_{22}m_1}{4(\mathbf{q}^2+M^2)}(\boldsymbol{\sigma}_1\times\boldsymbol{\sigma}_2)\cdot\mathbf{q}. \tag*{(83)}
\end{aligned} V 22 ( q ) = q 2 + M 2 g 21 g 22 m 1 m 2 σ 2 ⋅ v − 4 ( q 2 + M 2 ) i g 21 g 22 m 2 ( σ 1 × σ 2 ) ⋅ q − q 2 + M 2 g 21 g 22 m 1 m 2 σ 1 ⋅ v − 4 ( q 2 + M 2 ) i g 21 g 22 m 1 ( σ 1 × σ 2 ) ⋅ q . (83)
In coordinate space,
V ~ 22 ( r ) = m 1 m 2 g 21 g ~ 22 2 { σ 2 ⋅ v , Y M } − m 1 m 2 g ~ 21 g 22 2 { σ 1 ⋅ v , Y M } + m 2 g 21 g ~ 22 + m 1 g ~ 21 g 22 4 ( σ 1 × σ 2 ) ⋅ r ^ F M . (84) \widetilde{V}_{22}(\mathbf{r})=\frac{m_1m_2g_{21}\widetilde{g}_{22}}{2}\{\boldsymbol{\sigma}_2\cdot\mathbf{v},Y_M\}
-\frac{m_1m_2\widetilde{g}_{21}g_{22}}{2}\{\boldsymbol{\sigma}_1\cdot\mathbf{v},Y_M\}
+\frac{m_2g_{21}\widetilde{g}_{22}+m_1\widetilde{g}_{21}g_{22}}{4}(\boldsymbol{\sigma}_1\times\boldsymbol{\sigma}_2)\cdot\hat{\mathbf{r}}F_M. \tag*{(84)} V 22 ( r ) = 2 m 1 m 2 g 21 g 22 { σ 2 ⋅ v , Y M } − 2 m 1 m 2 g 21 g 22 { σ 1 ⋅ v , Y M } + 4 m 2 g 21 g 22 + m 1 g 21 g 22 ( σ 1 × σ 2 ) ⋅ r ^ F M . (84)
The first line has the V 12 , 13 V_{12,13} V 12 , 13 spin–velocity structure and the second the V 11 V_{11} V 11 structure. Both are P P P odd and T T T even. Consequently the minimal one-derivative tensor–axial-tensor coupling of a massive spin-2 boson violates parity but not time reversal.
Axial-tensor–axial-tensor spin-2 exchange
Using Eq. (71) , the leading result is
V ~ 22 ( q ) = g ~ 21 g ~ 22 m 1 m 2 2 ( q 2 + M 2 ) [ σ 1 ⋅ σ 2 + ( σ 1 ⋅ q ) ( σ 2 ⋅ q ) M 2 ] . (85) \widetilde{V}_{22}(\mathbf{q})=\frac{\widetilde{g}_{21}\widetilde{g}_{22}m_1m_2}{2(\mathbf{q}^2+M^2)}
\left[\boldsymbol{\sigma}_1\cdot\boldsymbol{\sigma}_2+
\frac{(\boldsymbol{\sigma}_1\cdot\mathbf{q})(\boldsymbol{\sigma}_2\cdot\mathbf{q})}{M^2}\right]. \tag*{(85)} V 22 ( q ) = 2 ( q 2 + M 2 ) g 21 g 22 m 1 m 2 [ σ 1 ⋅ σ 2 + M 2 ( σ 1 ⋅ q ) ( σ 2 ⋅ q ) ] . (85)
Its coordinate-space form is
V ~ 22 ( r ) = g ~ 21 g ~ 22 m 1 m 2 2 [ ( σ 1 ⋅ σ 2 ) Y M ( r ) + M − 2 σ 1 i σ 2 j H i j ( r ) ] + O ( v 2 ) . (86) \widetilde{V}_{22}(\mathbf{r})=\frac{\widetilde{g}_{21}\widetilde{g}_{22}m_1m_2}{2}
\left[(\boldsymbol{\sigma}_1\cdot\boldsymbol{\sigma}_2)Y_M(r)+M^{-2}\sigma_{1i}\sigma_{2j}H_{ij}(r)\right]+O(v^2). \tag*{(86)} V 22 ( r ) = 2 g 21 g 22 m 1 m 2 [ ( σ 1 ⋅ σ 2 ) Y M ( r ) + M − 2 σ 1 i σ 2 j H ij ( r ) ] + O ( v 2 ) . (86)
This potential is P , T P,T P , T even. The explicit 1 / M 2 1/M^2 1/ M 2 contribution is due to the longitudinal helicities and reflects the nonconservation of Θ ~ μ ν \widetilde{\Theta}^{\mu\nu} Θ μν for a massive fermion.
Relation to the sixteen rotationally invariant potentials
For comparison with experimental searches we now make the relation of the relativistic interactions above to the standard sixteen-operator basis of Dobrescu and Mocioiu explicit [2 , 4 ] . In the center-of-mass frame we use the momentum P \mathbf{P} P defined in Eq. (6) , with q = p 1 ′ − p 1 \mathbf{q}=\mathbf{p}_1'-\mathbf{p}_1 q = p 1 ′ − p 1 and q ⋅ P = 0 \mathbf{q}\cdot\mathbf{P}=0 q ⋅ P = 0 , and introduce an arbitrary reference mass m 0 m_0 m 0 only to make the operators dimensionless. A convenient Hermitian basis is
O 1 = 1 , O 2 = σ 1 ⋅ σ 2 , \mathcal{O}_1=1,\qquad\mathcal{O}_2=\boldsymbol{\sigma}_1\cdot\boldsymbol{\sigma}_2, O 1 = 1 , O 2 = σ 1 ⋅ σ 2 ,
O 3 = ( σ 1 ⋅ q ) ( σ 2 ⋅ q ) m 0 2 , \mathcal{O}_3=\frac{(\boldsymbol{\sigma}_1\cdot\mathbf{q})(\boldsymbol{\sigma}_2\cdot\mathbf{q})}{m_0^2}, O 3 = m 0 2 ( σ 1 ⋅ q ) ( σ 2 ⋅ q ) ,
O 4 , 5 = i 2 m 0 2 ( σ 1 ± σ 2 ) ⋅ ( P × q ) , \mathcal{O}_{4,5}=\frac{i}{2m_0^2}(\boldsymbol{\sigma}_1\pm\boldsymbol{\sigma}_2)\cdot(\mathbf{P}\times\mathbf{q}), O 4 , 5 = 2 m 0 2 i ( σ 1 ± σ 2 ) ⋅ ( P × q ) ,
O 6 , 7 = i 2 m 0 2 [ ( σ 1 ⋅ P ) ( σ 2 ⋅ q ) ± ( σ 1 ⋅ q ) ( σ 2 ⋅ P ) ] , \mathcal{O}_{6,7}=\frac{i}{2m_0^2}\left[(\boldsymbol{\sigma}_1\cdot\mathbf{P})(\boldsymbol{\sigma}_2\cdot\mathbf{q})\pm(\boldsymbol{\sigma}_1\cdot\mathbf{q})(\boldsymbol{\sigma}_2\cdot\mathbf{P})\right], O 6 , 7 = 2 m 0 2 i [ ( σ 1 ⋅ P ) ( σ 2 ⋅ q ) ± ( σ 1 ⋅ q ) ( σ 2 ⋅ P ) ] ,
O 8 = ( σ 1 ⋅ P ) ( σ 2 ⋅ P ) m 0 2 , \mathcal{O}_8=\frac{(\boldsymbol{\sigma}_1\cdot\mathbf{P})(\boldsymbol{\sigma}_2\cdot\mathbf{P})}{m_0^2}, O 8 = m 0 2 ( σ 1 ⋅ P ) ( σ 2 ⋅ P ) ,
O 9 , 10 = i 2 m 0 ( σ 1 ± σ 2 ) ⋅ q , \mathcal{O}_{9,10}=\frac{i}{2m_0}(\boldsymbol{\sigma}_1\pm\boldsymbol{\sigma}_2)\cdot\mathbf{q}, O 9 , 10 = 2 m 0 i ( σ 1 ± σ 2 ) ⋅ q ,
O 11 = i m 0 ( σ 1 × σ 2 ) ⋅ q , \mathcal{O}_{11}=\frac{i}{m_0}(\boldsymbol{\sigma}_1\times\boldsymbol{\sigma}_2)\cdot\mathbf{q}, O 11 = m 0 i ( σ 1 × σ 2 ) ⋅ q ,
O 12 , 13 = 1 2 m 0 ( σ 1 ± σ 2 ) ⋅ P , O 14 = ( σ 1 × σ 2 ) ⋅ P m 0 , \mathcal{O}_{12,13}=\frac{1}{2m_0}(\boldsymbol{\sigma}_1\pm\boldsymbol{\sigma}_2)\cdot\mathbf{P},\qquad
\mathcal{O}_{14}=\frac{(\boldsymbol{\sigma}_1\times\boldsymbol{\sigma}_2)\cdot\mathbf{P}}{m_0}, O 12 , 13 = 2 m 0 1 ( σ 1 ± σ 2 ) ⋅ P , O 14 = m 0 ( σ 1 × σ 2 ) ⋅ P ,
O 15 = 1 2 m 0 3 { [ σ 1 ⋅ ( P × q ) ] ( σ 2 ⋅ q ) + ( σ 1 ⋅ q ) [ σ 2 ⋅ ( P × q ) ] } , \mathcal{O}_{15}=\frac{1}{2m_0^3}\left\{[\boldsymbol{\sigma}_1\cdot(\mathbf{P}\times\mathbf{q})](\boldsymbol{\sigma}_2\cdot\mathbf{q})+(\boldsymbol{\sigma}_1\cdot\mathbf{q})[\boldsymbol{\sigma}_2\cdot(\mathbf{P}\times\mathbf{q})]\right\}, O 15 = 2 m 0 3 1 { [ σ 1 ⋅ ( P × q )] ( σ 2 ⋅ q ) + ( σ 1 ⋅ q ) [ σ 2 ⋅ ( P × q )] } ,
O 16 = i 2 m 0 3 { [ σ 1 ⋅ ( P × q ) ] ( σ 2 ⋅ P ) + ( σ 1 ⋅ P ) [ σ 2 ⋅ ( P × q ) ] } . (87) \mathcal{O}_{16}=\frac{i}{2m_0^3}\left\{[\boldsymbol{\sigma}_1\cdot(\mathbf{P}\times\mathbf{q})](\boldsymbol{\sigma}_2\cdot\mathbf{P})+(\boldsymbol{\sigma}_1\cdot\mathbf{P})[\boldsymbol{\sigma}_2\cdot(\mathbf{P}\times\mathbf{q})]\right\}. \tag*{(87)} O 16 = 2 m 0 3 i { [ σ 1 ⋅ ( P × q )] ( σ 2 ⋅ P ) + ( σ 1 ⋅ P ) [ σ 2 ⋅ ( P × q )] } . (87)
Scalar functions of q 2 \mathbf{q}^2 q 2 and P 2 \mathbf{P}^2 P 2 multiply these operators, so a factor q 2 \mathbf{q}^2 q 2 does not define a new rotational structure. We nevertheless keep an explicit notation such as q 2 V i q^2V_i q 2 V i below because it is physically important: it changes the radial dependence and, in particular, may turn a long-range potential into a contact-dominated one.
The discrete symmetries of the basis are
Table 88.
In particular, V 6 , 7 V_{6,7} V 6 , 7 are the only P P P -even, T T T -odd structures in this basis.
Table 1 (PDF p. 9) gives the mapping of the interactions derived in this work. For spin 1, the tensor and pseudotensor labels refer to the dipole vertices q μ σ μ ν q_\mu\sigma^{\mu\nu} q μ σ μν and q μ σ μ ν γ 5 q_\mu\sigma^{\mu\nu}\gamma_5 q μ σ μν γ 5 . For spin 2, “tensor” denotes the symmetric energy–momentum current and “axial tensor” its symmetric γ 5 \gamma_5 γ 5 counterpart; these are different Lorentz representations from the spin-1 dipole operators.
Table 1 (PDF p. 9). TABLE I. Mapping of the new spin-1 and spin-2 interactions onto the sixteen rotationally invariant potentials of Ref. [2 ] . An explicit q 2 q^2 q 2 indicates a momentum-suppressed realization of the same V i V_i V i operator. The entries refer to the orders retained in the present calculation.
The same information is displayed in Figure 3 , organized by the discrete symmetries rather than by the relativistic coupling. Several observations follow immediately. First, the finite-range spin-1 cross terms considered here fill precisely the gap left by the usual scalar/pseudoscalar and vector/axial-vector exchanges: g A g T g_Ag_T g A g T produces V 6 , 7 V_{6,7} V 6 , 7 at tree level. Second, the spin-2 basis considered here produces only the T T T -even sectors of Eq. (88); the conventional g 2 g 2 g_2g_2 g 2 g 2 interaction generates V 4 , 5 V_{4,5} V 4 , 5 already at the same quadratic momentum order as its static spin–spin term, while q 2 V 8 q^2V_8 q 2 V 8 starts at the next order. Hence no choice of real diagonal g 2 g_2 g 2 and g ~ 2 \tilde g_2 g ~ 2 generates either the TOPE V 6 , 7 V_{6,7} V 6 , 7 structures or the P , T P,T P , T -odd V 9 , 10 , 14 , 15 V_{9,10,14,15} V 9 , 10 , 14 , 15 structures at this derivative order. Third, several spin-1 and spin-2 amplitudes share an identical V i V_i V i component and radial function. Those identities permit direct coefficient reinterpretations of existing limits, but only when the experimental limit was obtained from the shared component. A complete relativistic coupling may generate additional operators, so equality of one V i V_i V i component must not be confused with equality of the full potentials.
Figure 3. Mapping of single-boson exchange onto the sixteen rotationally invariant potentials, organized by the discrete symmetries of Eq. (88). Each panel names the V i V_i V i belonging to that P , T P,T P , T class and lists the exchange channels that populate it for a spin-0 (S = 0 S=0 S = 0 ), spin-1 (S = 1 S=1 S = 1 ) and spin-2 (S = 2 S=2 S = 2 ) mediator. The S = 0 S=0 S = 0 entries are the scalar/pseudoscalar potentials of Refs. [1 , 3 ] ; g 2 g_2 g 2 and g ~ 2 \widetilde{g}_2 g 2 denote the energy–momentum and axial-tensor spin-2 couplings of Eq. (59) . Blue entries denote previously known interactions, green entries denote interactions generalized here, and orange entries denote the remaining channels derived in this work. In particular, the spin-1 g A g ~ T g_A\widetilde{g}_T g A g T TOPE interaction is generalized to finite mediator mass, while the conventional spin-2 g 2 g 2 g_2g_2 g 2 g 2 interaction is rederived and generalized in the present framework. The parity-even, time-reversal-odd sector is populated only by the spin-1 g A g ~ T g_A\widetilde{g}_T g A g T cross term, while the minimal spin-2 interactions populate only the T T T -even sectors.
For later use, two such component identities are particularly simple. From Eqs. (54) and (79) ,
V 22 ∣ Q 12 = g 21 g 22 8 Q 12 , (88) \left.V_{22}\right|_{Q_{12}}=\frac{g_{21}g_{22}}{8}Q_{12}, \tag*{(88)} V 22 ∣ Q 12 = 8 g 21 g 22 Q 12 , (88)
V T T ( 1 ) ∣ Q 12 = − 4 g T 1 g T 2 Q 12 , (89) \left.V_{TT}^{(1)}\right|_{Q_{12}}=-4g_T^1g_T^2Q_{12}, \tag*{(89)} V T T ( 1 ) Q 12 = − 4 g T 1 g T 2 Q 12 , (89)
so equal V 2 + V 3 V_2+V_3 V 2 + V 3 components correspond to
g 21 g 22 = − 32 g T 1 g T 2 . (90) g_{21}g_{22}=-32g_T^1g_T^2. \tag*{(90)} g 21 g 22 = − 32 g T 1 g T 2 . (90)
It is also instructive to compare ordinary spin-1 vector exchange with the spin-2 energy–momentum coupling. To the same static order,
V V V ( 1 ) = g V 1 g V 2 Y M − g V 1 g V 2 4 m 1 m 2 Q 12 + ⋯ . (91) V_{VV}^{(1)}=g_V^1g_V^2Y_M-\frac{g_V^1g_V^2}{4m_1m_2}Q_{12}+\cdots. \tag*{(91)} V V V ( 1 ) = g V 1 g V 2 Y M − 4 m 1 m 2 g V 1 g V 2 Q 12 + ⋯ . (91)
The V 1 V_1 V 1 terms alone would be equal for
g 21 g 22 = − 3 2 m 1 m 2 g V 1 g V 2 , (92) g_{21}g_{22}=-\frac{3}{2m_1m_2}g_V^1g_V^2, \tag*{(92)} g 21 g 22 = − 2 m 1 m 2 3 g V 1 g V 2 , (92)
whereas equality of the V 2 + V 3 V_2+V_3 V 2 + V 3 components requires
g 21 g 22 = − 2 m 1 m 2 g V 1 g V 2 . (93) g_{21}g_{22}=-\frac{2}{m_1m_2}g_V^1g_V^2. \tag*{(93)} g 21 g 22 = − m 1 m 2 2 g V 1 g V 2 . (93)
The two rescalings are different, so a spin-1 vector potential and a spin-2 energy–momentum potential are not globally proportional even though they share the same V 1 V_1 V 1 and V 2 + V 3 V_2+V_3 V 2 + V 3 operator structures.
Likewise the common V 11 V_{11} V 11 component of Eqs. (22) and (84) gives, for the ordering shown,
− 2 g A 1 g T 2 = m 2 4 g 21 g ~ 22 , -2g_A^{1}g_T^{2}=\frac{m_2}{4}g_{21}\tilde g_{22}, − 2 g A 1 g T 2 = 4 m 2 g 21 g ~ 22 ,
i.e.
g 21 g ~ 22 = − 8 m 2 g A 1 g T 2 . (94) g_{21}\tilde g_{22}=-\frac{8}{m_2}g_A^{1}g_T^{2}. \tag*{(94)} g 21 g ~ 22 = − m 2 8 g A 1 g T 2 . (94)
The qualifications are important: V 22 V_{22} V 22 also contains the central V 1 V_1 V 1 term, while V 2 ~ 2 ~ V_{\tilde2\tilde2} V 2 ~ 2 ~ contains an unsuppressed V 12 , 13 V_{12,13} V 12 , 13 term. Equations (90) and (94) are therefore operator-by-operator translations, not universal rescalings of the full spin-2 models.
A third useful component identity follows from the H i j H_{ij} H ij , or V 3 V_3 V 3 , part of Eqs. (55) and (86) . At fixed mediator mass,
4 g ~ T 1 g ~ T 2 = m 1 m 2 2 M 2 g ~ 21 g ~ 22 , 4\tilde g_T^1\tilde g_T^2=\frac{m_1m_2}{2M^2}\tilde g_{21}\tilde g_{22}, 4 g ~ T 1 g ~ T 2 = 2 M 2 m 1 m 2 g ~ 21 g ~ 22 ,
i.e.
g ~ 21 g ~ 22 = 8 M 2 m 1 m 2 g T 1 g T 2 . (95) \tilde g_{21}\tilde g_{22}=\frac{8M^2}{m_1m_2}g_T^1g_T^2. \tag*{(95)} g ~ 21 g ~ 22 = m 1 m 2 8 M 2 g T 1 g T 2 . (95)
This maps the identical V 3 V_3 V 3 radial function only. The spin-2 g ~ 2 g ~ 2 \tilde g_2\tilde g_2 g ~ 2 g ~ 2 interaction additionally contains the isotropic V 2 V_2 V 2 term. Alternatively, its complete leading radial bracket is identical to the ordinary spin-1 axial–axial Proca potential, for which
g ~ 21 g ~ 22 = − 2 g A 1 g A 2 m 1 m 2 (96) \tilde g_{21}\tilde g_{22}=-\frac{2g_A^1g_A^2}{m_1m_2} \tag*{(96)} g ~ 21 g ~ 22 = − m 1 m 2 2 g A 1 g A 2 (96)
(up to the overall interaction-sign convention). Thus axial–axial limits may be converted directly when the published analysis uses the full V 2 + V 3 V_2+V_3 V 2 + V 3 response in the same normalization.
For M 2 ≫ q 2 M^2\gg q^2 M 2 ≫ q 2 ,
1 q 2 + M 2 = 1 M 2 − q 2 M 4 + ⋯ , (97) \frac{1}{\mathbf{q}^2+M^2}=\frac{1}{M^2}-\frac{\mathbf{q}^2}{M^4}+\cdots, \tag*{(97)} q 2 + M 2 1 = M 2 1 − M 4 q 2 + ⋯ , (97)
Y M ( r ) ⟶ δ 3 ( r ) M 2 . (98) Y_M(r)\longrightarrow\frac{\delta^3(\mathbf{r})}{M^2}. \tag*{(98)} Y M ( r ) ⟶ M 2 δ 3 ( r ) . (98)
For spin 2 the tensor–tensor interaction becomes
V 22 c o n t a c t = − 2 g 21 g 22 m 1 m 2 3 M 2 δ 3 ( r ) − g 21 g 22 8 M 2 ( σ 1 ⋅ σ 2 ) ∇ 2 δ 3 ( r ) + g 21 g 22 8 M 2 ( σ 1 ⋅ ∇ ) ( σ 2 ⋅ ∇ ) δ 3 ( r ) + O ( M − 4 ) . (99) V_{22}^{\mathrm{contact}}=-\frac{2g_{21}g_{22}m_1m_2}{3M^2}\delta^3(\mathbf{r})
-\frac{g_{21}g_{22}}{8M^2}(\boldsymbol{\sigma}_1\cdot\boldsymbol{\sigma}_2)\nabla^2\delta^3(\mathbf{r})
+\frac{g_{21}g_{22}}{8M^2}(\boldsymbol{\sigma}_1\cdot\nabla)(\boldsymbol{\sigma}_2\cdot\nabla)\delta^3(\mathbf{r})+O(M^{-4}). \tag*{(99)} V 22 contact = − 3 M 2 2 g 21 g 22 m 1 m 2 δ 3 ( r ) − 8 M 2 g 21 g 22 ( σ 1 ⋅ σ 2 ) ∇ 2 δ 3 ( r ) + 8 M 2 g 21 g 22 ( σ 1 ⋅ ∇ ) ( σ 2 ⋅ ∇ ) δ 3 ( r ) + O ( M − 4 ) . (99)
The parity-odd mixed interaction gives
V 2 ~ 2 c o n t a c t = m 1 m 2 g 21 g ~ 22 2 M 2 { σ 2 ⋅ v , δ 3 ( r ) } − m 1 m 2 g ~ 21 g 22 2 M 2 { σ 1 ⋅ v , δ 3 ( r ) } − m 2 g 21 g ~ 22 4 M 2 ( σ 1 × σ 2 ) ⋅ ∇ δ 3 ( r ) − m 1 g ~ 21 g 22 4 M 2 ( σ 1 × σ 2 ) ⋅ ∇ δ 3 ( r ) + O ( M − 4 ) . (100) \begin{aligned}
V_{\tilde2 2}^{\mathrm{contact}}={}&\frac{m_1m_2g_{21}\tilde g_{22}}{2M^2}\{\boldsymbol{\sigma}_2\cdot\mathbf{v},\delta^3(\mathbf{r})\}
-\frac{m_1m_2\tilde g_{21}g_{22}}{2M^2}\{\boldsymbol{\sigma}_1\cdot\mathbf{v},\delta^3(\mathbf{r})\}\\
&-\frac{m_2g_{21}\tilde g_{22}}{4M^2}(\boldsymbol{\sigma}_1\times\boldsymbol{\sigma}_2)\cdot\nabla\delta^3(\mathbf{r})
-\frac{m_1\tilde g_{21}g_{22}}{4M^2}(\boldsymbol{\sigma}_1\times\boldsymbol{\sigma}_2)\cdot\nabla\delta^3(\mathbf{r})+O(M^{-4}). \tag*{(100)}
\end{aligned} V 2 ~ 2 contact = 2 M 2 m 1 m 2 g 21 g ~ 22 { σ 2 ⋅ v , δ 3 ( r )} − 2 M 2 m 1 m 2 g ~ 21 g 22 { σ 1 ⋅ v , δ 3 ( r )} − 4 M 2 m 2 g 21 g ~ 22 ( σ 1 × σ 2 ) ⋅ ∇ δ 3 ( r ) − 4 M 2 m 1 g ~ 21 g 22 ( σ 1 × σ 2 ) ⋅ ∇ δ 3 ( r ) + O ( M − 4 ) . (100)
The leading axial-tensor–axial-tensor result is
V 22 c o n t a c t = g ~ 21 g ~ 22 m 1 m 2 2 M 2 ( σ 1 ⋅ σ 2 ) δ 3 ( r ) − g ~ 21 g ~ 22 m 1 m 2 2 M 4 ( σ 1 ⋅ ∇ ) ( σ 2 ⋅ ∇ ) δ 3 ( r ) + ⋯ . (101) V_{22}^{\mathrm{contact}}=\frac{\widetilde{g}_{21}\widetilde{g}_{22}m_1m_2}{2M^2}(\boldsymbol{\sigma}_1\cdot\boldsymbol{\sigma}_2)\delta^3(\mathbf{r})-\frac{\widetilde{g}_{21}\widetilde{g}_{22}m_1m_2}{2M^4}(\boldsymbol{\sigma}_1\cdot\nabla)(\boldsymbol{\sigma}_2\cdot\nabla)\delta^3(\mathbf{r})+\cdots. \tag*{(101)} V 22 contact = 2 M 2 g 21 g 22 m 1 m 2 ( σ 1 ⋅ σ 2 ) δ 3 ( r ) − 2 M 4 g 21 g 22 m 1 m 2 ( σ 1 ⋅ ∇ ) ( σ 2 ⋅ ∇ ) δ 3 ( r ) + ⋯ . (101)
These spin-2 contact operators are not the conventional atomic-EDM tensor–pseudotensor interaction ( e ˉ σ μ ν e ) ( N ˉ i σ μ ν γ 5 N ) (\bar{e}\sigma_{\mu\nu}e)(\bar{N}i\sigma^{\mu\nu}\gamma_5N) ( e ˉ σ μν e ) ( N ˉ i σ μν γ 5 N ) , whose Lorentz indices are antisymmetric 2 . A pure symmetric spin-2 field cannot couple linearly to that antisymmetric current.
Massless limit and the graviton
The M → 0 M\to0 M → 0 limit of a massive Fierz–Pauli propagator is not identical to the propagator of a genuinely massless spin-2 gauge field. For conserved sources the massive exchange contains
T 1 μ ν T 2 μ ν − 1 3 T 1 T 2 , (102) T_1^{\mu\nu}T_{2\mu\nu}-\frac{1}{3}T_1T_2, \tag*{(102)} T 1 μν T 2 μν − 3 1 T 1 T 2 , (102)
whereas the massless graviton contains
T 1 μ ν T 2 μ ν − 1 2 T 1 T 2 . (103) T_1^{\mu\nu}T_{2\mu\nu}-\frac{1}{2}T_1T_2. \tag*{(103)} T 1 μν T 2 μν − 2 1 T 1 T 2 . (103)
This is the linear van Dam–Veltman–Zakharov discontinuity; nonlinear screening in consistent massive-gravity theories changes the physical interpretation at sufficiently short distances [8 ] .
For a genuine massless graviton, gauge invariance requires coupling to a conserved symmetric source. With
L g r a v = − κ 2 h μ ν T μ ν , κ 2 = 32 π G , (104) \mathcal{L}_{\mathrm{grav}}=-\frac{\kappa}{2}h_{\mu\nu}T^{\mu\nu},\qquad\kappa^2=32\pi G, \tag*{(104)} L grav = − 2 κ h μν T μν , κ 2 = 32 π G , (104)
the leading nonrelativistic potential is the Newton interaction
V N ( r ) = − G m 1 m 2 r , (105) V_N(r)=-\frac{Gm_1m_2}{r}, \tag*{(105)} V N ( r ) = − r G m 1 m 2 , (105)
which is P , T P,T P , T even. The axial-tensor current of Eq. (58) is not conserved for massive electrons, nucleons, or quarks and therefore cannot be introduced as an independent minimal source for an ordinary massless graviton. The same conservation requirement restricts the couplings available to strictly massless spin-1 fields [17 ] . In particular, the 1 / M 2 1/M^2 1/ M 2 longitudinal term in Eq. (86) has no smooth massless-graviton limit.
Existing limits and their reinterpretation
Connection with the conventions and exclusion plots of Ref. Cong2025
Review of Modern Physics paper [4 ] (abbreviated as RMP) is particularly useful for converting the results of this paper into experimentally constrained quantities. It presents both the Dobrescu–Mocioiu (DM) convention, in which a potential is written as V i = f i X Y V i V_i=f_i^{XY}V_i V i = f i X Y V i , and the coordinate-space particle-exchange convention of Fadeev et al., in which the coefficients are products of physical fermion–boson couplings. Appendix A of [4 ] , and in particular its Table VI, gives the conversion between these conventions. We adopt its normalization in this subsection. The labels X X X and Y Y Y specify the ordering of the two fermions, and
m r = m X m Y m X + m Y , M = λ − 1 (106) m_r=\frac{m_Xm_Y}{m_X+m_Y},\qquad M=\lambda^{-1} \tag*{(106)} m r = m X + m Y m X m Y , M = λ − 1 (106)
in natural units. Overall signs depend on the sign convention chosen for the interaction Lagrangian; limits below therefore use absolute values.
In terms of the kernels in Eqs. (9) –(12) , the RMP/DM potentials relevant to the interactions derived here may be written in the compact forms collected in Table II . These are the quantum, center-of-mass forms (including contact terms where appropriate). For macroscopic nonoverlapping sources the delta functions are omitted and p X = m r v \mathbf{p}_X=m_r\mathbf{v} p X = m r v may be treated classically. The V 6 + 7 V_{6+7} V 6 + 7 line is written in the original DM normalization because this term is not included in Table VI of [4 ] ; [4 ] nevertheless explicitly notes that the extended spin-1 Lagrangian containing g A g T e ˉ g_Ag_T\bar{e} g A g T e ˉ produces V 6 , 7 V_{6,7} V 6 , 7 .
Table II.
For reference, the leading Fadeev/RMP potentials needed most often in the translations can be written especially simply in the notation of this paper,
V V V = g V X g V Y Y M − g V X g V Y 4 m X m Y Q X Y + ⋯ , (107) V_{VV}=g_V^Xg_V^Y\mathcal{Y}_M-\frac{g_V^Xg_V^Y}{4m_Xm_Y}Q_{XY}+\cdots, \tag*{(107)} V V V = g V X g V Y Y M − 4 m X m Y g V X g V Y Q X Y + ⋯ , (107)
V A A = − g A X g A Y [ ( σ X ⋅ σ Y ) Y M + σ X i σ Y j H i j M 2 ] + ⋯ , (108) V_{AA}=-g_A^Xg_A^Y\left[(\boldsymbol{\sigma}_X\cdot\boldsymbol{\sigma}_Y)\mathcal{Y}_M+\frac{\sigma_X^i\sigma_Y^jH_{ij}}{M^2}\right]+\cdots, \tag*{(108)} V AA = − g A X g A Y [ ( σ X ⋅ σ Y ) Y M + M 2 σ X i σ Y j H ij ] + ⋯ , (108)
V A V X Y = g A X g V Y ( σ X ⋅ v ) Y M − g A X g V Y 2 m Y ( σ X × σ Y ) ⋅ r ^ F M + ⋯ , (109) V_{AV}^{XY}=g_A^Xg_V^Y(\boldsymbol{\sigma}_X\cdot\mathbf{v})\mathcal{Y}_M-\frac{g_A^Xg_V^Y}{2m_Y}(\boldsymbol{\sigma}_X\times\boldsymbol{\sigma}_Y)\cdot\hat{\mathbf{r}}F_M+\cdots, \tag*{(109)} V A V X Y = g A X g V Y ( σ X ⋅ v ) Y M − 2 m Y g A X g V Y ( σ X × σ Y ) ⋅ r ^ F M + ⋯ , (109)
V p s X Y = − g p X g s Y 2 m X ( σ X ⋅ r ^ ) F M + ⋯ , (110) V_{ps}^{XY}=-\frac{g_p^Xg_s^Y}{2m_X}(\boldsymbol{\sigma}_X\cdot\hat{\mathbf{r}})F_M+\cdots, \tag*{(110)} V p s X Y = − 2 m X g p X g s Y ( σ X ⋅ r ^ ) F M + ⋯ , (110)
V p p = − g p X g p Y 4 m X m Y σ X i σ Y j H i j . (111) V_{pp}=-\frac{g_p^Xg_p^Y}{4m_Xm_Y}\sigma_X^i\sigma_Y^jH_{ij}. \tag*{(111)} V pp = − 4 m X m Y g p X g p Y σ X i σ Y j H ij . (111)
The ellipses denote the paired higher-order terms displayed explicitly in Eqs. (16) –(24) of [4 ] .
The most useful result for the present paper is the coefficient dictionary in Table III . It permits a limit on either an RMP coefficient f i X Y f_i^{XY} f i X Y or a Fadeev-style coupling product to be converted directly into a limit on a coupling product introduced here. The table is intentionally component-by-component: when two relativistic models generate the same V i V_i V i with different ratios to other V j V_j V j terms, there is no universal rescaling of the complete potentials.
Table III. Component-by-component matching of the new spin-1 and spin-2 couplings to the RMP conventions.
Two features of Table III are worth stressing. The spin-1 g T g V g_Tg_V g T g V and spin-2 g 2 g 2 g_2g_2 g 2 g 2 interactions each give more than one standard V i V_i V i component, and the effective g V g V g_Vg_V g V g V rescaling is different for the different components. Likewise, the spin-2 g 2 g ~ 2 g_2\tilde{g}_2 g 2 g ~ 2 interaction has a different relative coefficient of V 11 V_{11} V 11 and V 12 , 13 V_{12,13} V 12 , 13 from a standard vector–axial-vector interaction. Therefore the strongest rigorous bound on a complete model requires folding the complete potential through the response function of the experiment; the relations in Table III are exact translations of the stated components.
Why exclusion plots are not repeated here. Section V of [4 ] already collects the up-to-date mass-dependent exclusion regions for the standard physical products: g A g V g_Ag_V g A g V in Fig. 9, g A g A g_Ag_A g A g A in Figs. 10 and 11, g V g V g_Vg_V g V g V in Fig. 12, g p g s g_pg_s g p g s in Fig. 13, g p g p g_pg_p g p g p in Fig. 14, and g s g s g_sg_s g s g s in Fig. 15. Appendix A and Table VI of that review provide the corresponding conversion to the DM coefficients
TABLE II. Relevant Dobrescu–Mocioiu/RMP potentials in natural units. The normalization follows Appendix A of Ref. [4 ] . H i j H_{ij} H ij is defined in Eq. (12) . The full V 14 q + 15 V_{14q+15} V 14 q + 15 expression is Eq. (A9) of Ref. [4 ] ; only its characteristic operator content is displayed here.
f i X Y f_i^{XY} f i X Y Thus the combination of those figures with Table III gives the continuous exclusion curves for the new products considered here, and reproducing essentially the same plots would add little information. Strictly speaking, Ref. [4 ] does not draw a separate plot for every individual f i f_i f i ; rather, its Table VI supplies the conversion between f i f_i f i and the physical coupling products. The new TOPE product g A g T g_Ag_T g A g T is also an exception: it is discussed in Sec. IV.E of Ref. [4 ] , where the relevant V 6 , 7 V_{6,7} V 6 , 7 searches are listed, but it is not one of the standard physical products plotted in Figs. 9–15. This is why we retain a small table of representative V 6 + 7 V_{6+7} V 6 + 7 limits below.
For convenience, Table IV gives a direct guide from the figures of Ref. [4 ] to the coupling products in this work. This also makes explicit which published exclusion plot should be used when a continuous mass-dependent limit, rather than one of the representative numbers below, is required.
Let Q Q Q denote the characteristic momentum transfer of a given experiment. For M ≪ Q M\ll Q M ≪ Q the Yukawa propagator approaches the long-range form, whereas for M ≫ Q M\gg Q M ≫ Q
1 q 2 + M 2 = 1 M 2 − q 2 M 4 + ⋯ . (112) \frac{1}{\mathbf{q}^{2}+M^{2}}=\frac{1}{M^{2}}-\frac{\mathbf{q}^{2}}{M^{4}}+\cdots. \tag*{(112)} q 2 + M 2 1 = M 2 1 − M 4 q 2 + ⋯ . (112)
Consequently a short-range experiment constrains a Wilson coefficient built from the coupling product divided by M 2 M^{2} M 2 , with additional derivatives when the vertices themselves contain momenta. These are scaling statements; a numerical contact bound is experiment dependent because the matrix element of the delta function or its derivatives depends on the source geometry or atomic/molecular wave function. Appendix A tabulates the resulting limits for each coupling combination.
The spin-2 axial-tensor–axial-tensor channel (Table XII) has a direct analogue in the massive spin-1 axial–axial potential. For fixed couplings, the coefficient of the longitudinal V 3 ∣ A A V_{3|AA} V 3∣ AA contribution scales as 1 / M 2 1/M^{2} 1/ M 2 and therefore becomes strongly enhanced as M → 0 M\to0 M → 0 . In the small-M M M regime, experimental limits on this contribution are naturally expressed in terms of the finite combination g A X g A Y λ 2 g_A^Xg_A^Y\lambda^2 g A X g A Y λ 2 , equivalently g A X g A Y / M 2 g_A^Xg_A^Y/M^2 g A X g A Y / M 2 , as discussed in Ref. [4 ] . Through the exact leading mapping g A X g A Y = − ( m X m Y / 2 ) g ~ 2 X g ~ 2 Y g_A^Xg_A^Y=-(m_Xm_Y/2)\tilde{g}_2^X\tilde{g}_2^Y g A X g A Y = − ( m X m Y /2 ) g ~ 2 X g ~ 2 Y , the corresponding spin-2 quantity is
∣ g ~ 2 X g ~ 2 Y M 2 ∣ = 2 m X m Y ∣ g A X g A Y M 2 ∣ . (113) \left|\frac{\tilde{g}_2^X\tilde{g}_2^Y}{M^2}\right|=\frac{2}{m_Xm_Y}\left|\frac{g_A^Xg_A^Y}{M^2}\right|. \tag*{(113)} M 2 g ~ 2 X g ~ 2 Y = m X m Y 2 M 2 g A X g A Y . (113)
This statement concerns a small but nonzero massive mediator. It must not be replaced by the bounds for a genuinely massless axial vector, for which the longitudinal 1 / M 2 1/M^2 1/ M 2 term is absent.
For all component maps whose matching factor K X Y K_{XY} K X Y contains only fermion masses, a complete RMP exclusion curve G R M P ( M ) G_{\mathrm{RMP}}(M) G RMP ( M ) immediately gives
∣ G n e w ∣ lim = K X Y ∣ G R M P ∣ lim , |G_{\mathrm{new}}|_{\lim}=K_{XY}|G_{\mathrm{RMP}}|_{\lim}, ∣ G new ∣ l i m = K X Y ∣ G RMP ∣ l i m ,
∣ G n e w M 2 ∣ lim = K X Y ∣ G R M P M 2 ∣ lim . (114) \left|\frac{G_{\mathrm{new}}}{M^2}\right|_{\lim}=K_{XY}\left|\frac{G_{\mathrm{RMP}}}{M^2}\right|_{\lim}. \tag*{(114)} M 2 G new l i m = K X Y M 2 G RMP l i m . (114)
TABLE III. Component-by-component matching of the new spin-1 and spin-2 couplings to the RMP conventions. Magnitudes are shown because the exclusion limits are bounds on absolute values. a X = 1 / 2 + m Y / ( 3 m X ) a_X = 1/2 + m_Y/(3m_X) a X = 1/2 + m Y / ( 3 m X ) . “Equivalent RMP product” means the physical coupling product that gives the same stated V i V_i V i component, not an equality of complete relativistic models.
TABLE IV. Guide to the exclusion plots in Ref. [4 ] . The third column lists the principal RMP/Fadeev potential components entering those plots; the last column gives examples of the new products in this paper that can be constrained component by component using Table III. The TOPE V 6 , 7 V_{6,7} V 6 , 7 interaction is discussed separately in Sec. IV.E of Ref. [4 ] .
Thus the same translation table applies to both the long-range plateau and the contact-normalized large-M M M limit. Equation (114) , together with Figs. 9–15 of Ref. [4 ] , is more informative than a second set of exclusion plots in this paper.
Spin-1 coupling products
The mapping above permits existing searches for individual V i V_i V i structures to be translated into limits on the new spin-1 coupling products. The tables in Appendix A quote limits on these products. Sec. A1 provides limits in the small and large mediator mass M M M . These are quoted explicitly from the source papers. Sec. A2 provides representative limits obtained by digitizing published exclusion curves using WebPlotDigitizer and translating a single representative point near the strongest sensitivity region. Therefore, the values presented are approximate. A complete and accurate determination of the constraints would require digitising and translating the entire and more up-to-date exclusion curves, which should be taken from Ref. [4 ] as described in § 3 A. Limits in both of these sections are component-wise single-operator translations , in which the coupling combination under consideration is assumed to be the only non-zero interaction. However, it is important to note coupling product can generate several V i V_i V i structures simultaneously.
For the RMP/DM normalization of Table II , and with fermion X X X carrying the displayed spin in V 4 + 5 V_{4+5} V 4 + 5 , the relevant component relations are
∣ g T ~ X g V ~ Y ∣ = m r 4 m X 2 ∣ f 4 + 5 X Y ∣ → m Y ≫ m X ∣ f 4 + 5 X Y ∣ 4 m X , (115) \left|g_{\widetilde{T}}^{X}g_{\widetilde{V}}^{Y}\right|=\frac{m_r}{4m_X^2}\left|f_{4+5}^{XY}\right|\xrightarrow{m_Y\gg m_X}\frac{\left|f_{4+5}^{XY}\right|}{4m_X}, \tag*{(115)} g T X g V Y = 4 m X 2 m r f 4 + 5 X Y m Y ≫ m X 4 m X f 4 + 5 X Y , (115)
∣ g T ~ X g V Y ∣ = ∣ g p ~ X g s Y ∣ 4 m X , (116) \left|g_{\widetilde{T}}^{X}g_{V}^{Y}\right|=\frac{\left|g_{\widetilde{p}}^{X}g_s^{Y}\right|}{4m_X}, \tag*{(116)} g T X g V Y = 4 m X g p X g s Y , (116)
∣ g A X g T ~ Y ∣ = ∣ g A X g V Y ∣ 4 m Y , (117) \left|g_{A}^{X}g_{\widetilde{T}}^{Y}\right|=\frac{\left|g_{A}^{X}g_{V}^{Y}\right|}{4m_Y}, \tag*{(117)} g A X g T Y = 4 m Y g A X g V Y , (117)
∣ g A X g T ~ Y ∣ = ∣ f 6 + 7 X Y ∣ R M P 4 m X , (118) \left|g_{A}^{X}g_{\widetilde{T}}^{Y}\right|=\frac{\left|f_{6+7}^{XY}\right|_{\mathrm{RMP}}}{4m_X}, \tag*{(118)} g A X g T Y = 4 m X f 6 + 7 X Y RMP , (118)
∣ g T X g T Y ∣ = ∣ g T ~ X g T ~ Y ∣ = ∣ g p X g p Y ∣ 16 m X m Y . (119) \left|g_{T}^{X}g_{T}^{Y}\right|=\left|g_{\widetilde{T}}^{X}g_{\widetilde{T}}^{Y}\right|=\frac{\left|g_p^{X}g_p^{Y}\right|}{16m_Xm_Y}. \tag*{(119)} g T X g T Y = g T X g T Y = 16 m X m Y g p X g p Y . (119)
The TOPE numbers quoted below follow the legacy experimental f 6 + 7 exp f_{6+7}^{\exp} f 6 + 7 e x p convention. In the notation of that table, ∣ g A e g T n ∣ = ∣ f 6 + 7 exp ∣ / ( 2 m r ) \left|g_A^e g_T^n\right|=\left|f_{6+7}^{\exp}\right|/(2m_r) ∣ g A e g T n ∣ = f 6 + 7 e x p / ( 2 m r ) , so the conversion to the RMP/DM normalization used above is
∣ f 6 + 7 X Y ∣ R M P = 2 m X m r ∣ f 6 + 7 ∣ exp . (120) \left|f_{6+7}^{XY}\right|_{\mathrm{RMP}}=\frac{2m_X}{m_r}\left|f_{6+7}\right|_{\exp}. \tag*{(120)} f 6 + 7 X Y RMP = m r 2 m X ∣ f 6 + 7 ∣ e x p . (120)
Here m r = m 1 m 2 / ( m 1 + m 2 ) m_r=m_1m_2/(m_1+m_2) m r = m 1 m 2 / ( m 1 + m 2 ) is the reduced mass. All numerical conversions below use m e = 0.510 998 950 M e V m_e=0.510\,998\,950\ \mathrm{MeV} m e = 0.510 998 950 MeV , m n = 939.565 421 M e V m_n=939.565\,421\ \mathrm{MeV} m n = 939.565 421 MeV , m p = 938.272 088 M e V m_p=938.272\,088\ \mathrm{MeV} m p = 938.272 088 MeV , and m μ = 105.658 375 M e V m_\mu=105.658\,375\ \mathrm{MeV} m μ = 105.658 375 MeV . The e e e –n n n entry Ji et al. in the TOPE table are sensitivity projections from Ref. [18 ] , which is a proposal rather than a measurement; they are labelled accordingly.
The same V 3 V_3 V 3 operator limits can be used to display particularly simple small-mass combinations for the massive spin-2 theory. Using Eqs. (90) and (95) , Table X and Table XII show (i) a component-wise limit on g 2 X g 2 Y g_2^Xg_2^Y g 2 X g 2 Y obtained from the common V 3 V_3 V 3 term and (ii) the finite longitudinal combination g ~ 2 X g ~ 2 Y / M 2 \widetilde{g}_2^X\widetilde{g}_2^Y/M^2 g 2 X g 2 Y / M 2 . The latter is the natural quantity when M M M is small but nonzero. These translations use the experimental operator bounds summarized by the RMP; they do not identify a strictly massless paraphoton with a massless spin-2 field.
Table X. Limits for small and large mediator masses on the tensor–tensor coupling ∣ g 2 1 g 2 2 ∣ |g_2^1g_2^2| ∣ g 2 1 g 2 2 ∣ of a massive spin-2 (S = 2 S=2 S = 2 ) mediator. The limits are obtained from the common V 3 V_3 V 3 component of the spin-2 tensor–tensor and spin-1 tensor–tensor interactions, using ∣ g 2 1 g 2 2 ∣ = 32 ∣ g T 1 g T 2 ∣ |g_2^1g_2^2|=32|g_T^1g_T^2| ∣ g 2 1 g 2 2 ∣ = 32∣ g T 1 g T 2 ∣ . The small-M M M limits are obtained by rescaling the corresponding tensor–tensor/pseudotensor–pseudotensor constraints for a massless paraphoton reported in Table IV of Ref. [4 ] . This is a translation based only on the common V 3 V_3 V 3 interaction structure and should not be interpreted as the strictly massless limit of the spin-2 theory; the mediator considered here remains a massive Fierz–Pauli spin-2 field. The large-M M M (contact) limit is obtained using ∣ g 2 1 g 2 2 ∣ / M 2 = 32 ∣ g T 1 g T 2 ∣ / M 2 |g_2^1g_2^2|/M^2=32|g_T^1g_T^2|/M^2 ∣ g 2 1 g 2 2 ∣/ M 2 = 32∣ g T 1 g T 2 ∣/ M 2 . This is a component-wise reinterpretation: the complete spin-2 tensor–tensor potential also contains the central V 1 V_1 V 1 term, the V 2 V_2 V 2 component accompanying V 3 V_3 V 3 , and V 4 , 5 V_{4,5} V 4 , 5 terms, which are not included in this V 3 V_3 V 3 -only translation. An em dash indicates that no corresponding small- or large-M M M limit is quoted.
Table XII. Constraints on the axial-tensor–axial-tensor coupling g ~ 2 1 g ~ 2 2 \tilde{g}_{2}^{1}\tilde{g}_{2}^{2} g ~ 2 1 g ~ 2 2 of a massive spin-2 (S = 2 S = 2 S = 2 ) mediator for small and large mediator masses. Both regimes constrain the combination ∣ g ~ 2 1 g ~ 2 2 ∣ / M 2 \lvert\tilde{g}_{2}^{1}\tilde{g}_{2}^{2}\rvert/M^{2} ∣ g ~ 2 1 g ~ 2 2 ∣ / M 2 , but for different physical reasons. In the small-M M M regime, the 1 / M 2 1/M^{2} 1/ M 2 dependence arises from the longitudinally enhanced V 3 V_{3} V 3 component of the massive spin-2 interaction; the quoted small-M M M limits ∣ g ~ 2 1 g ~ 2 2 ∣ / M 2 \lvert\tilde{g}_{2}^{1}\tilde{g}_{2}^{2}\rvert/M^{2} ∣ g ~ 2 1 g ~ 2 2 ∣ / M 2 are therefore component-wise V 3 V_{3} V 3 reinterpretations. In the large-M M M regime, the V 3 V_{3} V 3 contribution is O ( M − 4 ) \mathcal{O}(M^{-4}) O ( M − 4 ) , while the leading V 2 V_{2} V 2 term acquires the usual 1 / M 2 1/M^{2} 1/ M 2 dependence in the contact limit; the quoted large-M M M limits ∣ g ~ 2 1 g ~ 2 2 ∣ / M 2 \lvert\tilde{g}_{2}^{1}\tilde{g}_{2}^{2}\rvert/M^{2} ∣ g ~ 2 1 g ~ 2 2 ∣ / M 2 are therefore component-wise V 2 V_{2} V 2 reinterpretations. The positronium, muonium and helium entries (rows 2, 3 and 4) are obtained from axial–axial constraints using the mapping ∣ g ~ 2 1 g ~ 2 2 ∣ / M 2 = 2 ∣ g A 1 g A 2 ∣ / ( M 2 ( m 1 m 2 ) ) \lvert\tilde{g}_{2}^{1}\tilde{g}_{2}^{2}\rvert/M^{2} = 2\lvert g_{A}^{1}g_{A}^{2}\rvert/(M^{2}(m_{1}m_{2})) ∣ g ~ 2 1 g ~ 2 2 ∣ / M 2 = 2 ∣ g A 1 g A 2 ∣ / ( M 2 ( m 1 m 2 )) . Other small-M M M entries are component-wise V 3 V_{3} V 3 reinterpretations of pseudotensor–pseudotensor constraints using ∣ g ~ 2 1 g ~ 2 2 ∣ / M 2 = ∣ g T 1 g T 2 ∣ / ( m 1 m 2 ) \lvert\tilde{g}_{2}^{1}\tilde{g}_{2}^{2}\rvert/M^{2} = \lvert g_{T}^{1}g_{T}^{2}\rvert/(m_{1}m_{2}) ∣ g ~ 2 1 g ~ 2 2 ∣ / M 2 = ∣ g T 1 g T 2 ∣ / ( m 1 m 2 ) . Only the corresponding V 3 V_{3} V 3 or V 2 V_{2} V 2 component is included in each translation; other terms in the complete spin-2 axial-tensor–axial-tensor interaction may contribute to the response of a particular experiment. The small-M M M results apply to a massive Fierz–Pauli mediator whose range is large compared with the characteristic length scale of the experiment; they should not be interpreted as constraints on a strictly massless spin-2 mediator. For N N N , the neutron mass is used in the numerical conversion. An em dash indicates no corresponding limit is quoted.
Reference [4 ] also emphasizes that no experimental or observational constraint on the complete massless tensor–pseudotensor V 14 q + 15 V_{14q+15} V 14 q + 15 interaction was available at the time of that review. Hence a model-independent strict-M = 0 M=0 M = 0 bound on g T ~ X g T Y g_{\widetilde{T}}^Xg_T^Y g T X g T Y cannot be obtained by using a V 15 V_{15} V 15 limit alone.
We therefore quote no numerical limit on the mixed g T ~ e g T T g_{\widetilde{T}}^e g_T^T g T e g T T product: the required complete q 2 V 14 ± V 15 q^2V_{14\pm V_{15}} q 2 V 14 ± V 15 response has not been constrained independently.
Spin-2 reinterpretations
The minimal spin-2 interactions of Eq. (59) are T T T even. Consequently the P , T P,T P , T -odd constraints in Table XIV and Table XV , and the TOPE constraints in Table XVII , have no counterpart in this minimal spin-2 operator basis. The remaining shared V i V_i V i components can be reinterpreted directly, provided one keeps clear that the complete spin-2 interaction generally contains additional structures.
Table XIV. Representative laboratory g p g s g_{p}g_{s} g p g s constraints and translated spin-1 (S = 1 S = 1 S = 1 ) vector–pseudotensor bounds, checked with Eq. (116) . These bounds are component-wise V 9 + 10 V_{9+10} V 9 + 10 reinterpretations. The complete vector–pseudotensor interaction also contains the momentum-suppressed q 2 V 14 , 15 q^{2}V_{14,15} q 2 V 14 , 15 contributions, which are not included in this translation.
Table XV. EDM-derived g p g s g_{p}g_{s} g p g s constraints and the corresponding translated bounds on the spin-1 (S = 1 S = 1 S = 1 ) vector–pseudotensor coupling, obtained by matching the V 9 + 10 V_{9+10} V 9 + 10 component. These bounds are component-wise V 9 + 10 V_{9+10} V 9 + 10 reinterpretations. The complete vector–pseudotensor interaction also contains the momentum-suppressed q 2 V 14 , 15 q^{2}V_{14,15} q 2 V 14 , 15 contributions, which are not included in this translation.
Table XVII. Representative TOPE V 6 + 7 V_{6+7} V 6 + 7 limits and the corresponding translated bounds on the spin-1 (S = 1 S = 1 S = 1 ) axial–pseudotensor coupling, obtained by matching the common V 6 + 7 V_{6+7} V 6 + 7 component. Eq. (120) gives the conversion between the legacy expression f 6 + 7 f_{6+7} f 6 + 7 normalization used in the source table and the RMP/DM normalization. The e e e –n n n entry from [18 ] is a projected limit from a proposed experiment.
First consider V 4 + 5 V_{4+5} V 4 + 5 . For an unpolarized fermion b b b and a polarized fermion a a a , define
A a = 1 2 + m b 3 m a . (121) A_a=\frac{1}{2}+\frac{m_b}{3m_a}. \tag*{(121)} A a = 2 1 + 3 m a m b . (121)
Using P = m r v \mathbf{P}=m_r\mathbf{v} P = m r v and the classical source–sensor limit of the anticommutator in Eq. (82) , comparison with the standard V 4 + 5 V_{4+5} V 4 + 5 normalization gives
∣ g 2 a g 2 b ∣ = ∣ f 4 + 5 a b ∣ 2 m a 2 A a . (122) \left|g_{2a}g_{2b}\right|=\frac{\left|f_{4+5}^{ab}\right|}{2m_a^2A_a}. \tag*{(122)} ∣ g 2 a g 2 b ∣ = 2 m a 2 A a f 4 + 5 ab . (122)
Equation (122) uses the exact RMP Appendix-A normalization. The form based on the commonly used macroscopic f ⊥ f_\perp f ⊥ normalization differs by the factor m r / m a m_r/m_a m r / m a , which is negligible for e e e –N N N .
The V 2 + V 3 V_2+V_3 V 2 + V 3 component of the spin-2 g 2 g 2 g_2g_2 g 2 g 2 potential and the spin-1 g T g T g_Tg_T g T g T potential are related by Eq. (90) . Likewise the V 3 V_3 V 3 component of the spin-2 g ~ 2 g ~ 2 \widetilde{g}_2\widetilde{g}_2 g 2 g 2 potential and the spin-1 g ~ T g ~ T \widetilde{g}_T\widetilde{g}_T g T g T potential are related by Eq. (95) . Thus
∣ g 21 g 22 ∣ = 32 ∣ g T 1 g T 2 ∣ , (123) \left|g_{21}g_{22}\right|=32\left|g_T^1g_T^2\right|, \tag*{(123)} ∣ g 21 g 22 ∣ = 32 g T 1 g T 2 , (123)
∣ g ~ 21 g ~ 22 ∣ V 3 = 8 M 2 m 1 m 2 ∣ g T ~ 1 g T ~ 2 ∣ , M = ℏ c λ . (124) \left|\widetilde{g}_{21}\widetilde{g}_{22}\right|_{V_3}=\frac{8M^2}{m_1m_2}\left|g_{\widetilde{T}}^1g_{\widetilde{T}}^2\right|,\qquad M=\frac{\hbar c}{\lambda}. \tag*{(124)} ∣ g 21 g 22 ∣ V 3 = m 1 m 2 8 M 2 g T 1 g T 2 , M = λ ℏ c . (124)
The second relation is explicitly a V 3 V_3 V 3 -component translation; the full spin-2 axial–tensor–axial–tensor potential also contains V 2 V_2 V 2 .
Finally, the common V 11 V_{11} V 11 component of the spin-1 axial–tensor and spin-2 tensor–axial–tensor interactions gives
∣ g 21 g ~ 22 ∣ = 8 m 2 2 ∣ g A 1 g T 2 ∣ = 2 m 2 2 ∣ g A 1 g V 2 ∣ . (125) \begin{aligned}
\left|g_{21}\widetilde{g}_{22}\right|=\frac{8}{m_2^2}\left|g_A^1g_T^2\right| \\
&=\frac{2}{m_2^2}\left|g_A^1g_V^2\right|. \tag*{(125)}
\end{aligned} ∣ g 21 g 22 ∣ = m 2 2 8 g A 1 g T 2 = m 2 2 2 g A 1 g V 2 . (125)
The particle ordering is the one in Eq. (94) : a source limit on g A 1 g V 2 g_A^1g_V^2 g A 1 g V 2 maps to ordinary spin-2 coupling on particle 1 and axial-tensor spin-2 coupling on particle 2. The complete spin-2 mixed potential also contains an unsuppressed V 12 , 13 V_{12,13} V 12 , 13 contribution.
For completeness, Eq. (96) gives an exact leading-order functional rescaling between the spin-2 g ~ 2 g ~ 2 \tilde{g}_2\tilde{g}_2 g ~ 2 g ~ 2 interaction and ordinary spin-1 axial–axial exchange whenever the experimental analysis uses the full V 2 + V 3 V_2+V_3 V 2 + V 3 Proca potential. This provides an independent route to spin-2 bounds beyond the component translations tabulated above. If a publication constrains only V 2 V_2 V 2 or only V 3 V_3 V 3 , the full-model limit should instead be obtained by recomputing the experimental response to the complete spin-2 potential.
Conclusion
We have derived finite-range two-fermion potentials for a massive spin-1 mediator carrying simultaneously vector/axial-vector and tensor/pseudotensor interactions. The mixed vertices generate operator combinations not present when the vector and dipole sectors are treated separately. In particular,
g A g T ~ ⟹ V 6 , 7 , P = + 1 , T = − 1 , (126) g_A g_{\widetilde{T}} \Longrightarrow V_{6,7}, \qquad P=+1,\qquad T=-1, \tag*{(126)} g A g T ⟹ V 6 , 7 , P = + 1 , T = − 1 , (126)
providing a tree-level single-boson mechanism for the TOPE spin–velocity interaction in the standard sixteen-potential basis. The results given here include both vertex orderings and the distributional contact terms required for short-range and atomic-scale applications.
We also extended the analysis to a massive spin-2 field. For the conventional conserved energy–momentum coupling the longitudinal propagator terms vanish exactly. A symmetric axial-tensor fermion current is not conserved for massive fermions, so its exchange probes the longitudinal spin-2 helicities. The mixed tensor–axial-tensor interaction is P P P odd but T T T even, while the tensor–tensor and axial-tensor–axial-tensor interactions are P , T P,T P , T even. Thus no T T T -odd force is produced by the minimal spin-2 operator basis considered here. The heavy-mediator limit yields local derivative operators, whereas a genuine massless graviton is restricted by gauge invariance to conserved symmetric sources and gives the usual P , T P,T P , T -even gravitational interaction.
We have also mapped the new interactions onto the sixteen-potential Dobrescu–Mocioiu basis and used this mapping to obtain limits on the new coupling products, both in the small- and large-mediator-mass M M M regime, where the constraints are quoted directly from the source papers, and at a representative finite masses obtained by digitising the published exclusion curves. Among the new tensor/pseudotensor and spin-2 channels derived explicitly here, V 16 V_{16} V 16 is not generated at the order retained. The conventional axial-vector–vector Proca interaction can generate V 16 V_{16} V 16 at higher order [4 ] . Where spin-1 and spin-2 exchange share an identical operator and radial function, we supplied explicit coefficient maps and corresponding spin-2 component bounds. These results provide a unified starting point for interpreting macroscopic fifth-force measurements, precision spin experiments, and atomic or molecular tests in terms of massive spin-1 and spin-2 mediators. Dedicated analyses incorporating all
operators generated by a given relativistic coupling combination will be needed for rigorous full-model bounds.
Acknowledgments. — This work was supported by the Australian Research Council Grant No. DP230101058.
Constraints on coupling combinations
This appendix presents the constraints on the spin-1 and spin-2 interaction constants considered in this work. The bounds are obtained by matching components of the potentials derived above to existing experimental constraints, assuming in each case that the coupling combination under consideration is the only nonzero new interaction. We first present the limits for small- and large-mediator-mass regimes, where available, followed by representative constraints at finite mediator masses. The representative limits are obtained by digitising the published exclusion curves and are intended to illustrate the sensitivity to the newly derived coupling combinations; they should therefore be regarded as approximate rather than as a complete reanalysis of the experimental data.
Constraints in the regimes of large and small masses M of the mediator.
In the case of a small mass M M M of the mediator, the Yukawa exponent exp ( − M r ) ≈ 1 \exp(-Mr)\approx1 exp ( − M r ) ≈ 1 , and the constraints on the product of the interaction constants g i g k g_i g_k g i g k usually do not depend on M M M . In experiments with atoms and small molecules, r ∼ a B r\sim a_B r ∼ a B , so we consider the mass regime M a B ≪ 1 Ma_B\ll1 M a B ≪ 1 , giving M ≪ 4 M\ll4 M ≪ 4 keV, where a B a_B a B is the Bohr radius. In the opposite, large-mass regime, we assume the contact limit, Y M ( r ) → δ 3 ( r ) / M 2 Y_M(r)\to\delta^3(r)/M^2 Y M ( r ) → δ 3 ( r ) / M 2 , so constraints are usually expressed in terms of g i g k / M 2 g_i g_k/M^2 g i g k / M 2 . In atomic experiments this typically corresponds to M a B / Z ≫ 1 Ma_B/Z\gg1 M a B / Z ≫ 1 , giving M ≫ 4 Z M\gg4Z M ≫ 4 Z keV, where Z Z Z is the nuclear charge. These are characteristic atomic estimates, since the relevant momentum scale is experiment- and observable-dependent.
Table V. Limits for small and large mediator masses on the spin-1 vector–tensor coupling ∣ g V 1 g T 2 ∣ |g_V^1g_T^2| ∣ g V 1 g T 2 ∣ . The M → 0 M \to0 M → 0 limits are obtained by matching the common V 3 V_3 V 3 component of the interactions to the tensor–tensor/pseudotensor–pseudotensor constraints for a massless paraphoton reported in Table IV of Ref. [4 ] , using ∣ g V 1 g T 2 ∣ = 4 v h 2 m 1 ∣ Re ( C 1 ) Re ( C 2 ) / Λ 4 ∣ |g_V^1g_T^2| = 4v_h^2m_1|\operatorname{Re}(C_1)\operatorname{Re}(C_2)/\Lambda^4| ∣ g V 1 g T 2 ∣ = 4 v h 2 m 1 ∣ Re ( C 1 ) Re ( C 2 ) / Λ 4 ∣ , with normalisation g T = v h 2 Re ( C ) / Λ 2 g_T = v_h^2\operatorname{Re}(C)/\Lambda^2 g T = v h 2 Re ( C ) / Λ 2 , and v h = 246 v_h = 246 v h = 246 GeV. The large-M M M (contact) limit is obtained from the pseudoscalar–pseudoscalar V 3 V_3 V 3 constraint using ∣ g V 1 g T 2 ∣ / M 2 = ∣ g p 1 g p 2 ∣ / ( 4 m 2 M 2 ) |g_V^1g_T^2|/M^2 = |g_p^1g_p^2|/(4m_2M^2) ∣ g V 1 g T 2 ∣/ M 2 = ∣ g p 1 g p 2 ∣/ ( 4 m 2 M 2 ) . This is a component-wise V 3 V_3 V 3 reinterpretation. The additional V 2 V_2 V 2 contribution present in the full combination ( V 2 + V 3 ) ∣ V T (V_2+V_3)|_{V T} ( V 2 + V 3 ) ∣ V T is not included and may become important in the short-range (contact) regime. The complete vector–tensor interaction also contains the momentum-suppressed q 2 V 1 q^2V_1 q 2 V 1 and V 4 , 5 V_{4,5} V 4 , 5 terms, which are likewise omitted here but may contribute to the response of a particular experiment. An em dash indicates that no corresponding small- or large-M M M limit is quoted.
Table VI. Limits for small and large mediator masses on the spin-1 vector–pseudotensor coupling ∣ g T 1 g V 2 ∣ |g_T^1g_V^2| ∣ g T 1 g V 2 ∣ , obtained from constraints on the V 9 + 10 V_{9+10} V 9 + 10 interaction using ∣ g T 1 g V 2 ∣ = ∣ g p 1 g s 2 ∣ / ( 4 m 1 ) |g_T^1g_V^2| = |g_p^1g_s^2|/(4m_1) ∣ g T 1 g V 2 ∣ = ∣ g p 1 g s 2 ∣/ ( 4 m 1 ) . The small-M M M limits apply in the long-range regime, M ≪ 1 M \ll1 M ≪ 1 keV, while the large-M M M limits apply in the contact regime, M ≫ 1 M \gg1 M ≫ 1 GeV; see Table III of Ref. [25 ] . These limits are component-wise V 9 + 10 V_{9+10} V 9 + 10 reinterpretations. The complete vector–pseudotensor interaction also contains momentum-suppressed q 2 V 14 ± V 15 q^2V_{14}\pm V_{15} q 2 V 14 ± V 15 terms, which are not included in this translation.
Table VII. Limits for small and large mediator masses on the spin-1 axial–tensor coupling ∣ g A 2 g T 1 ∣ |g_A^2g_T^1| ∣ g A 2 g T 1 ∣ . The small-M M M limit is obtained from the V 11 V_{11} V 11 constraint using ∣ g A 2 g T 1 ∣ = ∣ g A 2 g V 1 ∣ / ( 4 m 2 ) |g_A^2g_T^1| = |g_A^2g_V^1|/(4m_2) ∣ g A 2 g T 1 ∣ = ∣ g A 2 g V 1 ∣/ ( 4 m 2 ) and applies for M ≲ 0.1 μ e V M \lesssim0.1\,\mu\mathrm{eV} M ≲ 0.1 μ eV . This is a component-wise V 11 V_{11} V 11 reinterpretation. The complete axial–tensor interaction also contains the momentum-suppressed q 2 V 12 + 13 q^2V_{12+13} q 2 V 12 + 13 contribution, which is not included in this translation. The quoted e − e e-e e − e constraint is a 1 σ 1\sigma 1 σ limit from Ref. [26 ] . No corresponding large-M M M V 11 V_{11} V 11 limit was identified.
Table VIII. Constraints on the spin-1 axial–pseudotensor coupling g A g T ~ g_Ag_{\tilde T} g A g T ~ , obtained by translating the dimensionless Conti–Khriplovich constants β \beta β [14 ] via the dictionary derived here, ∣ g A g T ~ ∣ = ( π / m p ) ∣ β ∣ = 3.35 ∣ β ∣ |g_Ag_{\tilde T}| = (\pi/m_p)|\beta| = 3.35|\beta| ∣ g A g T ~ ∣ = ( π / m p ) ∣ β ∣ = 3.35∣ β ∣ GeV− 1 ^{-1} − 1 . Here m p m_p m p is not a dynamical fermion mass but the reference mass introduced in Ref. [14 ] through the factor 1 / ( 2 m ) 1/(2m) 1/ ( 2 m ) , with m m m conventionally fixed to the proton mass so that β \beta β is dimensionless. The quoted β \beta β bounds are indirect constraints inferred from electron and neutron EDM limits through two-loop Z Z Z -exchange diagrams. Their mediator-mass dependence enters through ln ( Λ 2 / M 2 ) \ln(\Lambda^2/M^2) ln ( Λ 2 / M 2 ) , with M Z = max ( M , M Z ) M_Z = \max(M,M_Z) M Z = max ( M , M Z ) . The underlying leading-log expression is independent of M M M below M Z M_Z M Z and depends only logarithmically on M M M above M Z M_Z M Z ; for their numerical estimates Ref. [14 ] conservatively set this logarithm to unity. For the nonidentical-fermion channels, the fourth column shows the same translated coupling constraint divided by M Z 2 M_Z^2 M Z 2 , evaluated at M = M Z = 91.19 M=M_Z=91.19 M = M Z = 91.19 GeV. It is included only as a reference contact-form normalization and is not an independent high-mass asymptote of a Yukawa-force exclusion curve. Equivalently, Ref. [14 ] parametrizes the local four-fermion interaction by 4 π β / M 2 = ( G F / 2 ) q ~ 4\pi\beta/M^2 = (G_F/\sqrt{2})\tilde q 4 π β / M 2 = ( G F / 2 ) q ~ , which gives ∣ g A g T ~ ∣ / M 2 = G F ∣ q ~ ∣ / ( 4 2 m p ) = 2.20 × 10 − 6 ∣ q ~ ∣ |g_Ag_{\tilde T}|/M^2 = G_F|\tilde q|/(4\sqrt{2}m_p) = 2.20 \times10^{-6}|\tilde q| ∣ g A g T ~ ∣/ M 2 = G F ∣ q ~ ∣/ ( 4 2 m p ) = 2.20 × 1 0 − 6 ∣ q ~ ∣ GeV− 3 ^{-3} − 3 . The q ~ \tilde q q ~ limits quoted in Ref. [14 ] were evaluated for M ∼ 100 m p M \sim100m_p M ∼ 100 m p and rounded at the order-of-magnitude level, and are therefore used only as a cross-check rather than as the numerical input to the fourth column. For identical fermions the local e − e e-e e − e TOPE interaction vanishes because the scattering contribution is cancelled by the exchange contribution [14 ] ; therefore, no corresponding contact entry is given. The nucleon-level entries follow from quark-level estimates of the nucleon matrix elements and therefore carry an additional order-of-magnitude uncertainty. The first index of β \beta β denotes the pseudoscalar (derivative) vertex. The EDM-derived bounds quoted here are conditional on assumptions about the short-distance realization of the TOPE interaction. A subsequent effective field theory analysis showed that the connection between EDM and direct TOPE constraints depends on the ultraviolet realization of parity violation [27 ] . We therefore treat the Conti–Khriplovich bounds as conditional constraints rather than fully model-independent constraints.
Table IX. Limits for small and large mediator masses on the spin-1 tensor–tensor and pseudotensor–pseudotensor couplings, ∣ g T 1 g T 2 ∣ |g_T^1g_T^2| ∣ g T 1 g T 2 ∣ and ∣ g T ~ 1 g T ~ 2 ∣ |g_{\widetilde{T}}^1g_{\widetilde{T}}^2| ∣ g T 1 g T 2 ∣ , respectively. The small-M M M limits are obtained from the massless-paraphoton constraints in Table IV of Ref. [4 ] , using ∣ g T 1 g T 2 ∣ = v h 2 ∣ Re ( C 1 ) Re ( C 2 ) ∣ / Λ 4 |g_T^1g_T^2|=v_h^2|\operatorname{Re}(C_1)\operatorname{Re}(C_2)|/\Lambda^4 ∣ g T 1 g T 2 ∣ = v h 2 ∣ Re ( C 1 ) Re ( C 2 ) ∣/ Λ 4 and ∣ g T ~ 1 g T ~ 2 ∣ = v h 2 ∣ Im ( C 1 ) Im ( C 2 ) ∣ / Λ 4 |g_{\widetilde{T}}^1g_{\widetilde{T}}^2|=v_h^2|\operatorname{Im}(C_1)\operatorname{Im}(C_2)|/\Lambda^4 ∣ g T 1 g T 2 ∣ = v h 2 ∣ Im ( C 1 ) Im ( C 2 ) ∣/ Λ 4 , with normalisation g T = v h / Λ 2 Re ( C ) g_T=v_h/\Lambda^2\operatorname{Re}(C) g T = v h / Λ 2 Re ( C ) , g T ~ = v h / Λ 2 Im ( C ) g_{\widetilde{T}}=v_h/\Lambda^2\operatorname{Im}(C) g T = v h / Λ 2 Im ( C ) , and v h = 246 G e V v_h=246\ \mathrm{GeV} v h = 246 GeV . The large-M M M limits are obtained from the pseudoscalar–pseudoscalar V 3 V_3 V 3 constraint using ∣ g T 1 g T 2 ∣ / M 2 = ∣ g T ~ 1 g T ~ 2 ∣ / M 2 = ∣ g P 1 g P 2 ∣ / ( 16 m 1 m 2 M 2 ) |g_T^1g_T^2|/M^2=|g_{\widetilde{T}}^1g_{\widetilde{T}}^2|/M^2=|g_P^1g_P^2|/(16m_1m_2M^2) ∣ g T 1 g T 2 ∣/ M 2 = ∣ g T 1 g T 2 ∣/ M 2 = ∣ g P 1 g P 2 ∣/ ( 16 m 1 m 2 M 2 ) . For g T g T g_Tg_T g T g T , the complete tensor–tensor potential also contains a V 2 V_2 V 2 contribution, which is not included in this translation and may contribute to the response of a particular experiment. For g T ~ g T ~ g_{\widetilde{T}}g_{\widetilde{T}} g T g T , the interaction is a pure V 3 V_3 V 3 structure at the order retained. The resulting V 3 V_3 V 3 -based numerical limits are identical for the two couplings. An em dash indicates that no corresponding small- or large-M M M limit is quoted.
Table XI. Limits for small and large mediator masses on the mixed tensor–axial-tensor coupling ∣ g ~ 2 1 g ~ 2 2 ∣ \lvert\tilde{g}_{2}^{1}\tilde{g}_{2}^{2}\rvert ∣ g ~ 2 1 g ~ 2 2 ∣ of a massive spin-2 (S = 2 S = 2 S = 2 ) mediator. The limits are obtained by matching the common V 12 + 13 V_{12+13} V 12 + 13 component of the spin-2 g 2 g ~ 2 2 g_{2}\tilde{g}_{2}^{2} g 2 g ~ 2 2 interaction to the axial–vector interaction of a massive spin-1 mediator, using ∣ g ~ 2 1 g ~ 2 2 ∣ = ∣ g V 1 g A 2 ∣ / ( m 1 m 2 ) \lvert\tilde{g}_{2}^{1}\tilde{g}_{2}^{2}\rvert= \lvert g_{V}^{1}g_{A}^{2}\rvert/(m_{1}m_{2}) ∣ g ~ 2 1 g ~ 2 2 ∣ = ∣ g V 1 g A 2 ∣ / ( m 1 m 2 ) . The 133 C s ^{133}\mathrm{Cs} 133 Cs , 174 Y b ^{174}\mathrm{Yb} 174 Yb , and 205 T l ^{205}\mathrm{Tl} 205 Tl rows are obtained from atomic parity-nonconservation constraints on g A N g V N g_{A}^{N}g_{V}^{N} g A N g V N , where g V N = ( N g V n + Z g V p ) / A g_{V}^{N} = (Ng_{V}^{n} + Zg_{V}^{p})/A g V N = ( N g V n + Z g V p ) / A , and therefore correspond to effective e e e –N N N couplings. The 133 C s ^{133}\mathrm{Cs} 133 Cs anapole row instead constrains g A p g V N g_{A}^{p}g_{V}^{N} g A p g V N and therefore corresponds to a p p p –N N N coupling. For the numerical conversion we use m N ≃ 0.939 m_{N} \simeq0.939 m N ≃ 0.939 GeV. The small-M M M column is obtained from the long-range limits quoted in Ref. [28 ] , whereas the large-M M M column gives the contact-normalized limits. These are component-wise reinterpretations based only on the shared V 12 + 13 V_{12+13} V 12 + 13 interaction structure and should not be interpreted as the strictly massless limit of the spin-2 theory; the mediator considered here remains a massive Fierz–Pauli spin-2 field. The complete spin-2 g 2 g ~ 2 2 g_{2}\tilde{g}_{2}^{2} g 2 g ~ 2 2 interaction also contains a V 11 V_{11} V 11 contribution, which is not included in this translation.
Representative limits
The representative limits below are obtained by digitizing published exclusion curves using WebPlotDigitizer and translating selected points through the corresponding curve. For each experiment, a representative point is chosen near the region of strongest sensitivity. These values are approximate and are intended to provide representative constraints rather than a complete translation of the exclusion curves over the full mediator-mass range. Table IV provides a guide to the published exclusion curves that can be used, together with the mappings above, to obtain the full mass-dependent constraints.
Table IV. Guide to the exclusion plots in Ref. [4 ] .
Table XIII. Representative limits on f 4 + 5 f_{4+5} f 4 + 5 and the corresponding translated bounds on the spin-1 (S = 1 S = 1 S = 1 ) vector–tensor coupling, obtained by matching the common V 4 + 5 V_{4+5} V 4 + 5 component using Eq. (115) . These bounds are component-wise V 4 + 5 V_{4+5} V 4 + 5 reinterpretations. The complete vector–tensor interaction also contains the q 2 V 1 q^{2}V_{1} q 2 V 1 and V 2 + V 3 V_{2}+V_{3} V 2 + V 3 contributions, which are not included in this translation and may contribute to the response of a particular experiment.
Table XIX. Spin-2 tensor–axial–tensor bounds obtained from the shared V 11 V_{11} V 11 component of Table XVI using Eq. (125) . Spin-2 exchange also generates V 12 , 13 V_{12,13} V 12 , 13 .
Table XVI. Representative V 11 V_{11} V 11 limits on g A g V g_{A}g_{V} g A g V and the corresponding translated bounds on the spin-1 (S = 1 S = 1 S = 1 ) axial–tensor coupling, obtained by matching the common V 11 V_{11} V 11 component using Eq. (117) . These bounds are component-wise V 11 V_{11} V 11 reinterpretations. The complete axial–tensor interaction also contains the momentum-suppressed q 2 V 12 + 13 q^{2}V_{12+13} q 2 V 12 + 13 contribution, which is not included in this translation.
TABLE XX. Spin-2 axial-tensor–axial-tensor bounds from the common V 3 V_{3} V 3 component of the spin-1 pseudotensor–pseudotensor interaction in Table XVIII . Equation (124) is used; the additional spin-2 V 2 V_{2} V 2 term is not included in this translation.
Table XVIII. Representative g p g p g_{p}g_{p} g p g p limits and the corresponding translated bounds on the spin-1 (S = 1 S = 1 S = 1 ) tensor–tensor and pseudotensor–pseudotensor couplings, obtained by matching the common V 3 V_{3} V 3 component using Eq. (119) . These bounds are component-wise V 3 V_{3} V 3 reinterpretations. For g T g T g_{T}g_{T} g T g T , the complete tensor–tensor interaction also contains a V 2 V_{2} V 2 contribution, which is not included in this translation. For g T q g T q g_{T}^{q}g_{T}^{q} g T q g T q , the interaction is a pure V 3 V_{3} V 3 structure at the order retained.
Table XX.
TABLE XXI. Spin-2 g 2 g 2 g_{2}g_{2} g 2 g 2 bounds obtained from the V 3 V_{3} V 3 -based g T g T g_{T}g_{T} g T g T limits in Table XVIII , using ∣ g 2 2 ∣ = 32 ∣ g T 2 ∣ \lvert g_{2}^{2}\rvert= 32\lvert g_{T}^{2}\rvert ∣ g 2 2 ∣ = 32 ∣ g T 2 ∣ . This is a component-wise translation obtained by matching the V 3 V_{3} V 3 term. The additional V 2 V_{2} V 2 term accompanying V 3 V_{3} V 3 , as well as the V 1 V_{1} V 1 and V 4 , 5 V_{4,5} V 4 , 5 terms are not included in this translation.
Table XXI.
Nonrelativistic fermion bilinears
With q = p ′ − p q = p^{\prime} - p q = p ′ − p , p = ( p i n + p o u t ) / 2 \mathbf{p} = (\mathbf{p}^{\mathrm{in}}+\mathbf{p}^{\mathrm{out}})/2 p = ( p in + p out ) /2 , and u ‾ u ≃ 2 m \overline{u}u \simeq2m u u ≃ 2 m , the leading bilinears are
u ‾ ′ γ 0 u ≃ 2 m , (127) \overline{u}^{\prime}\gamma^{0}u \simeq2m, \tag*{(127)} u ′ γ 0 u ≃ 2 m , (127)
u ‾ ′ γ u ≃ 2 p − i q × σ , (128) \overline{u}^{\prime}\boldsymbol{\gamma}u \simeq2\mathbf{p} - i\mathbf{q}\times\boldsymbol{\sigma}, \tag*{(128)} u ′ γ u ≃ 2 p − i q × σ , (128)
u ‾ ′ γ 0 γ 5 u ≃ 2 σ ⋅ p , (129) \overline{u}^{\prime}\gamma^{0}\gamma_{5}u \simeq2\boldsymbol{\sigma}\cdot\mathbf{p}, \tag*{(129)} u ′ γ 0 γ 5 u ≃ 2 σ ⋅ p , (129)
u ‾ ′ γ γ 5 u ≃ 2 m σ , (130) \overline{u}^{\prime}\boldsymbol{\gamma}\gamma_{5}u \simeq2m\boldsymbol{\sigma}, \tag*{(130)} u ′ γ γ 5 u ≃ 2 m σ , (130)
u ‾ ′ γ 5 u ≃ − σ ⋅ q . (131) \overline{u}^{\prime}\gamma_{5}u \simeq-\boldsymbol{\sigma}\cdot\mathbf{q}. \tag*{(131)} u ′ γ 5 u ≃ − σ ⋅ q . (131)
For the contracted tensor current,
T 0 ≡ q α u ‾ ′ σ α 0 u ≃ − i q 2 + 2 σ ⋅ ( p × q ) , (132) T^{0} \equiv q_{\alpha}\overline{u}^{\prime}\sigma^{\alpha0}u \simeq-i\mathbf{q}^{2}+2\boldsymbol{\sigma}\cdot(\mathbf{p}\times\mathbf{q}), \tag*{(132)} T 0 ≡ q α u ′ σ α 0 u ≃ − i q 2 + 2 σ ⋅ ( p × q ) , (132)
T i ≡ q α u ‾ ′ σ α i u ≃ 2 m ( q × σ ) i , (133) T^{i} \equiv q_{\alpha}\overline{u}^{\prime}\sigma^{\alpha i}u \simeq2m(\mathbf{q}\times\boldsymbol{\sigma})^{i}, \tag*{(133)} T i ≡ q α u ′ σ α i u ≃ 2 m ( q × σ ) i , (133)
and for the contracted pseudotensor current,
T ~ 0 ≡ q α u ‾ ′ σ α 0 γ 5 u ≃ 2 i m ( σ ⋅ q ) , (134) \widetilde{T}^{0} \equiv q_{\alpha}\overline{u}^{\prime}\sigma^{\alpha0}\gamma_{5}u \simeq2im(\boldsymbol{\sigma}\cdot\mathbf{q}), \tag*{(134)} T 0 ≡ q α u ′ σ α 0 γ 5 u ≃ 2 im ( σ ⋅ q ) , (134)
T ~ i ≡ q α u ‾ ′ σ α i γ 5 u ≃ 2 i p i ( σ ⋅ q ) . (135) \widetilde{T}^{i} \equiv q_{\alpha}\overline{u}^{\prime}\sigma^{\alpha i}\gamma_{5}u \simeq2ip^{i}(\boldsymbol{\sigma}\cdot\mathbf{q}). \tag*{(135)} T i ≡ q α u ′ σ α i γ 5 u ≃ 2 i p i ( σ ⋅ q ) . (135)
At the second vertex the momentum entering the fermion line is − q -q − q , producing the corresponding sign reversals. The exact Gordon identity
q μ u ‾ ′ σ μ ν u = i u ‾ ′ ( 2 m γ ν − K ν ) u (136) q_{\mu}\overline{u}^{\prime}\sigma^{\mu\nu}u = i\overline{u}^{\prime}(2m\gamma^{\nu}-K^{\nu})u \tag*{(136)} q μ u ′ σ μν u = i u ′ ( 2 m γ ν − K ν ) u (136)
provides a useful check on Eqs. (132) and (B7).
The basic identities used throughout the paper are
∫ d 3 q ( 2 π ) 3 e i q ⋅ r q 2 + M 2 = Y M ( r ) , (137) \int\frac{\mathrm{d}^{3}q}{(2\pi)^{3}}\frac{e^{i\mathbf{q}\cdot\mathbf{r}}}{\mathbf{q}^{2}+M^{2}} = Y_{M}(r), \tag*{(137)} ∫ ( 2 π ) 3 d 3 q q 2 + M 2 e i q ⋅ r = Y M ( r ) , (137)
∫ d 3 q ( 2 π ) 3 e i q ⋅ r q q 2 + M 2 = D M ( r ) , (138) \int\frac{\mathrm{d}^{3}q}{(2\pi)^{3}}\frac{e^{i\mathbf{q}\cdot\mathbf{r}}\mathbf{q}}{\mathbf{q}^{2}+M^{2}} = \mathbf{D}_{M}(r), \tag*{(138)} ∫ ( 2 π ) 3 d 3 q q 2 + M 2 e i q ⋅ r q = D M ( r ) , (138)
∫ d 3 q ( 2 π ) 3 e i q ⋅ r q i q j q 2 + M 2 = H i j ( r ) . (139) \int\frac{\mathrm{d}^{3}q}{(2\pi)^{3}}\frac{e^{i\mathbf{q}\cdot\mathbf{r}}q_{i}q_{j}}{\mathbf{q}^{2}+M^{2}} = H_{ij}(r). \tag*{(139)} ∫ ( 2 π ) 3 d 3 q q 2 + M 2 e i q ⋅ r q i q j = H ij ( r ) . (139)
The contact term in Eq. (12) follows from the distributional identity
∂ i ∂ j 1 r = 3 r ^ i r ^ j − δ i j r 3 − 4 π 3 δ i j δ 3 ( r ) . (140) \partial_{i}\partial_{j}\frac{1}{r} = \frac{3\widehat{r}_{i}\widehat{r}_{j}-\delta_{ij}}{r^{3}}-\frac{4\pi}{3}\delta_{ij}\delta^{3}(\mathbf{r}). \tag*{(140)} ∂ i ∂ j r 1 = r 3 3 r i r j − δ ij − 3 4 π δ ij δ 3 ( r ) . (140)
These terms must be retained when the interaction range is comparable to or shorter than atomic length scales.
For diagonal interactions of a neutral Hermitian boson, it is convenient to choose a basis of Hermitian fermion operators and real coefficients. In particular,
ψ ‾ σ μ ν ψ is Hermitian , ψ ‾ σ μ ν γ 5 ψ is anti-Hermitian , (141) \overline{\psi}\sigma^{\mu\nu}\psi\text{ is Hermitian},\qquad\overline{\psi}\sigma^{\mu\nu}\gamma_{5}\psi\text{ is anti-Hermitian}, \tag*{(141)} ψ σ μν ψ is Hermitian , ψ σ μν γ 5 ψ is anti-Hermitian , (141)
so the pseudotensor term is written as i g T ~ ψ ‾ σ μ ν γ 5 ψ ig_{\widetilde{T}}\overline{\psi}\sigma^{\mu\nu}\gamma_{5}\psi i g T ψ σ μν γ 5 ψ with real g T ~ g_{\widetilde{T}} g T . Equivalently one may omit the explicit i i i and use a purely imaginary coefficient. These are the same Hermitian interaction written in two conventions. A physical T T T -violating phase arises only when a phase cannot be removed by such an operator redefinition, for example in an appropriate non-diagonal coupling matrix. The T T T -odd spin-1 potentials in Sec. III arise from the transformation properties of the Hermitian operators themselves and do not require complex diagonal couplings.
References