betaFeedback

WorksPhysics and Astronomy

Complete theory of radiative corrections to Kℓ3K_{\ell 3} decays and the VusV_{us} update

Abstract

We fill up the missing piece in our own re-analysis of the long-distance electromagnetic radiative corrections to semileptonic kaon decays by performing a rigorous study in the K→πμ+νμ(γ)K\rightarrowπμ^+ν_μ(γ) channels. With appropriate experimental and lattice inputs, we achieve a precision level of 10−410^{-4} in these channels. This is comparable to our previous analysis in the K→πe+νe(γ)K\rightarrowπe^+ν_e(γ) channels. With this new result, we present an updated global analysis to extract the Standard Model parameter ∣Vus∣|V_{us}| from semileptonic kaon decays. We obtain ∣Vus∣=0.22308(55)|V_{us}|=0.22308(55) and 0.22356(73)0.22356(73), using the lattice average of the K0→π−K^0\rightarrowπ^- transition form factor at Nf=2+1+1N_f=2+1+1 and Nf=2+1N_f=2+1, respectively.

Introduction

A series of new applications of the classical Sirlin representation of the Standard Model (SM) radiative corrections (RCs) [1, 2] has recently brought our understanding of semileptonic decays of mesons, nucleons and nuclei to the next level. For instance, the implementation of a dispersion relation analysis on top of this framework has significantly improved the accuracy of the SM predictions of the free neutron [3, 4, 5, 6, 7] and nuclear beta decays [4, 8], leading to a more precise extraction of the top-row Cabibbo-Kobayashi-Maskawa (CKM) matrix element VudV_{ud}. These new developments unveiled an apparent violation of the top-row CKM unitarity [9], which provides one of the most promising evidences of the breakdown of precise SM theory predictions at low-energy precision experiments, alongside with the anomalies observed in the muon anomalous magnetic moment [10, 11, 12, 13, 14] and decays of B-mesons [15, 16, 17, 18].

It was first thought that Sirlin’s representation is only useful for decay processes where the parent and daughter particles are nearly degenerate, because some of the more complicated hadronic matrix elements involving products of three electroweak currents vanish in this limit. However, some of the authors in this paper pointed out that this framework is equally applicable to semileptonic decay with large mass gaps, e.g. the kaon semileptonic decays (Kℓ3K_{\ell3}), upon combining it with more modern techniques such as Chiral Perturbation Theory (ChPT) [19]. In a follow-up work [20] we outlined an efficient approach for lattice Quantum Chromodynamics (QCD) to help fixing the large theory uncertainties in the Kℓ3K_{\ell3} RC that originate from non-perturbative QCD, which was then partially put into practice [21]. With these new theory basis and lattice inputs, we improved the existing theory precision of the Ke3K_{e3} RC [22] by almost an order of magnitude [23, 24], and presented an updated global analysis to the value of the top-row CKM matrix element ∣Vus∣|V_{us}| extracted from semileptonic kaon decays [25]. This new analysis supports the current top-row CKM unitarity violation, and further sharpens a previously-observed discrepancy between the values of ∣Vus∣|V_{us}| obtained from Kℓ3K_{\ell3} and the leptonic kaon decay (Kμ2K_{\mu2}) [9].

This paper is a generalization of the aforementioned works to cover all channels in semileptonic kaon decays, in particular the Kμ3K_{\mu3} channels. A reason why only the Ke3K_{e3} decays were previously focused on is that in these channels the effect of the (more) poorly-constrained RCs to the intrinsically non-forward charged weak form factor f−(t)f_{-}(t) to the decay rate is suppressed by me2/MK2∼10−6m_e^2/M_K^2 \sim10^{-6} (on top of α/π\alpha/\pi), making them theoretically clean. In this work we show that, with the precision goal of 10−410^{-4}, this can even be done for the Kμ3K_{\mu3} channels by first separating the infrared (IR)-divergent and IR-finite corrections to f−(t)f_-(t), calculating the former precisely and matching the latter to a fixed-order ChPT expression. This simple prescription enables us to calculate the Kμ3K_{\mu3} RCs to a same level of precision with their Ke3K_{e3} counterparts.

The content of this work is arranged as follows. In § 2 we introduce the basic notations and formula for the Kℓ3K_{\ell3} decay rate. In § 3 we outline the theory framework of the O(α)\mathcal{O}(\alpha) virtual corrections, most of them are similar to Refs. [23, 24] except a novel and more rigorous treatment of the RCs to f−(t)f_-(t). In § 4 we discuss the bremsstrahlung contributions. The numerical results and error analysis are presented in § 5 together with a new global analysis of ∣Vus∣|V_{us}| from semileptonic kaon decays. In § 6 we state our final conclusions.

Basic formalism

This work is a straightforward extension of our previous Ke3K_{e3} calculation. It is rather meaningless to copy everything from the existing literature, so most of the time we simply refer the reader to the notations, formulas and descriptions in specific sections of Ref. [24]. Some trivial generalizations of notations, e.g. e→ℓe \to\ell, are automatically implied if not stated otherwise.

First of all, we define Kℓ3K_{\ell3} (where ℓ=e,μ\ell=e,\mu) as the fully-inclusive semileptonic decay process K(p)→π(p′)+ℓ+(pℓ)+νℓ(pν)+nγK(p) \to\pi(p')+\ell^+(p_\ell)+\nu_\ell(p_\nu)+n\gamma, with n≥0n \ge0. At tree-level, the decay amplitude is given by:

M0=−GF2uˉνγμ(1−γ5)vℓFμ(p′,p),(1)M_0=-\frac{G_F}{\sqrt{2}}\bar{u}_\nu\gamma^\mu(1-\gamma_5)v_\ell F_\mu(p',p), \tag*{(1)}

where

Fμ(p′,p)=⟨π(p′)∣(JWμ)†∣K(p)⟩=Vus∗[f+Kπ(t)(p+p′)μ+f−Kπ(t)(p−p′)μ](2)F_\mu(p',p)=\langle\pi(p')|(J_W^\mu)^\dagger|K(p)\rangle=V_{us}^*\left[f_+^{K\pi}(t)(p+p')_\mu+f_-^{K\pi}(t)(p-p')_\mu\right] \tag*{(2)}

defines the (real) charged weak form factors f±Kπ(t)f_\pm^{K\pi}(t) (for notational simplicity, from now on we will suppress the superscript KπK\pi whenever it does not cause any confusion). The standard Mandelstam variables are defined as s=(p′+pℓ)2s=(p'+p_\ell)^2, t=(p−p′)2t=(p-p')^2 and u=(p−pℓ)2u=(p-p_\ell)^2, but for n=0n=0 only two kinematic variables are independent, which are often chosen as y=2p⋅pℓ/MK2y = 2p \cdot p_\ell/M_K^2 and z=2p⋅p′/MK2z = 2p \cdot p^{\prime}/M_K^2. The total tree-level decay rate is then given by:

(ΓKℓ3)tree=MK256π3∫D3dydz  GF2∣Vus∣2{f+2H(+1,+1)+2f+f−H(+1,−1)+f−2H(−1,−1)},(3)(\Gamma_{K\ell3})_{\mathrm{tree}} = \frac{M_K}{256\pi^3}\int_{\mathcal{D}_3} dydz\; G_F^2 |V_{us}|^2 \left\{f_+^2 H(+1,+1) + 2f_+f_-H(+1,-1) + f_-^2 H(-1,-1)\right\}, \tag*{(3)}

where the integration region D3\mathcal{D}_3 is defined in Appendix A of [24], while the function H(a,b)H(a,b) is defined in Eq. (2.7) of the same reference.

The full O(α)\mathcal{O}(\alpha) electroweak RC to Kℓ3K_{\ell3} include one-loop virtual corrections at n=0n=0 and tree-level real (i.e. bremsstrahlung) corrections at n=1n=1. It is most conveniently expressed as a fractional correction to the tree-level decay rate:

δKℓ3=δΓKℓ3(ΓKℓ3)tree.(4)\delta_{K\ell3} = \frac{\delta\Gamma_{K\ell3}}{(\Gamma_{K\ell3})_{\mathrm{tree}}}. \tag*{(4)}

However, one must be aware that in the existing ChPT treatment some of the short-distance electromagnetic corrections are not included in the definition of δKℓ3\delta_{K\ell3}, but rather redistributed into the charged weak form factors as well as the isospin-breaking corrections (see Sec. 8 of [24] for a discussion). After removing those terms, our result can be directly compared to the “long-distance” electromagnetic RC δEMKℓ\delta_{\mathrm{EM}}^{K\ell} studied in the standard ChPT literature by further removing a short-distance, channel-independent electroweak correction factor:

δEMKℓ=δKℓ3−(SEW−1),(5)\delta_{\mathrm{EM}}^{K\ell} = \delta_{K\ell3} - (S_{\mathrm{EW}} - 1), \tag*{(5)}

where SEW=1.0232(3)HOS_{\mathrm{EW}} = 1.0232(3)_{\mathrm{HO}} [26].

Below we discuss the virtual and real corrections separately.

Virtual corrections

Using on-shell relations, the virtual corrections to the amplitude M0M_0 can always be expressed in terms of a shift to f±f_\pm, i.e.

M0→M0+δMvir,withf±(t)→f±(t)+δf±(y,z).(6)M_0 \to M_0 + \delta M_{\mathrm{vir}}, \qquad\text{with}\qquad f_\pm(t) \to f_\pm(t) + \delta f_\pm(y,z). \tag*{(6)}

The corrections δf±\delta f_\pm are generally complex functions of two variables y,zy,z instead of a single variable tt. The corresponding change in the decay rate is:

(δΓKℓ3)vir=MK256π3∫D3dydz  2GF2∣Vus∣2{H(+1,+1)f+Re⁡δf++H(+1,−1)[f+Re⁡δf−+f−Re⁡δf+]+H(−1,−1)f−Re⁡δf−}.(7)(\delta\Gamma_{K\ell3})_{\mathrm{vir}} = \frac{M_K}{256\pi^3}\int_{\mathcal{D}_3} dydz\; 2G_F^2|V_{us}|^2 \left\{H(+1,+1)f_+\operatorname{Re}\delta f_+ + H(+1,-1)\left[f_+\operatorname{Re}\delta f_- + f_-\operatorname{Re}\delta f_+\right] + H(-1,-1)f_-\operatorname{Re}\delta f_-\right\}. \tag*{(7)}

An important feature of the equation above is that both H(+1,−1)H(+1,-1) and H(−1,−1)H(-1,-1) contain an explicit factor rℓ≡mℓ2/MK2r_{\ell} \equiv m_{\ell}^{2}/M_{K}^{2} which suppresses the contribution of δf−\delta f_{-} to the decay rate. In the Ke3K_{e3} channels, re∼10−6r_{e} \sim10^{-6} and δf−\delta f_{-} is completely negligible, which is one of the basic assumptions of our previous work. However, even in the Kμ3K_{\mu3} channels, the suppression rμ∼0.04r_{\mu} \sim0.04 is still quite significant. Therefore, apart from the IR-singular pieces in δf−\delta f_{-} that need to be rigorously calculated for the exact cancellation of IR-divergences, one is allowed to adopt an approximate representation for the remaining, IR-finite pieces without introducing a large error budget to the decay rate. This is the main spirit of this work.

The general structure of the virtual corrections to the decay amplitude can be summarized following Section 3 of Ref. [24] as

δMvir=δMI+(δM2+δMγWa)int+δMγWb+δM3.(8)\delta M_{\mathrm{vir}} = \delta M_{\mathrm{I}} + \left(\delta M_{2} + \delta M_{\gamma W}^{a}\right)_{\mathrm{int}} + \delta M_{\gamma W}^{b} + \delta M_{3}. \tag*{(8)}

The first term collects all universal, model-independent analytical pieces that originate from corrections sensitive to IR and UV physics. The second and third terms combine the remnants of the two-point function and the γW\gamma W-box contributions upon accounting for the respective analytic contributions in the first term. The last term represents the contribution of the three-point function. As explained in Ref. [24], the full γW\gamma W-box correction stems from the generalized Compton tensor TμνT^{\mu\nu} which contains a symmetric and an antisymmetric pieces. Correspondingly, the two contributions δMγWa\delta M_{\gamma W}^{a} and δMγWb\delta M_{\gamma W}^{b} are distinguished. Below, we discuss all different contributions to δf±\delta f_{\pm} in Sirlin’s representation in necessary detail.

A Analytic pieces

Firstly, part of the loop integrals in the O(α)\mathcal{O}(\alpha) electroweak RC can be calculated analytically (bearing errors of the order O(mℓ2/MW2)\mathcal{O}(m_{\ell}^{2}/M_{W}^{2})) and the outcome is proportional to the tree-level amplitude,

(δf±)I={α2π[ln⁡MZ2mℓ2−14ln⁡MW2mℓ2+12ln⁡mℓ2Mγ2−38+12a~g]+12δHOQED}f±(t),(9)(\delta f_{\pm})_{\mathrm{I}} = \left\{ \frac{\alpha}{2\pi} \left[ \ln\frac{M_{Z}^{2}}{m_{\ell}^{2}} - \frac{1}{4}\ln\frac{M_{W}^{2}}{m_{\ell}^{2}} + \frac{1}{2}\ln\frac{m_{\ell}^{2}}{M_{\gamma}^{2}} - \frac{3}{8} + \frac{1}{2}\widetilde{a}_{g} \right] + \frac{1}{2}\delta_{\mathrm{HO}}^{\mathrm{QED}} \right\} f_{\pm}(t), \tag*{(9)}

see Eq. (3.11) and below of Ref. [24] for explanations of the notations. Throughout this work we introduce a fictitious photon mass MγM_{\gamma} to regularize the IR-divergence.

From (δM2+δMγWa)int+δMγWb,V(\delta M_2+\delta M_{\gamma W}^{a})_{\mathrm{int}}+\delta M_{\gamma W}^{b,V}

Next we have the integrals (δM2+δMγWa)int+δMγWb,V(\delta M_2+\delta M_{\gamma W}^{a})_{\mathrm{int}}+\delta M_{\gamma W}^{b,V} defined in Eqs. (3.9) and (3.10) of Ref. [24]. Two relevant comments are in order:

  • They give IR-divergent contributions to both δf±\delta f_{\pm}.

  • One can show that their contributions to δf+\delta f_{+} cannot depend on physics at large loop momentum; the same argument, however, does not apply to δf−\delta f_{-}.

Therefore, we adopt the following strategy:

  1. To study δf+\delta f_{+} from these integrals, we apply the same technique as in Sec. 4 of Ref. [24], to write

(δf+)this subsection=(δf+)II+(δf+)convfin+(δf+)Born−conv+(δf+)inel.(10)(\delta f_{+})_{\mathrm{this\ subsection}}=(\delta f_{+})_{\mathrm{II}}+(\delta f_{+})_{\mathrm{conv}}^{\mathrm{fin}}+(\delta f_{+})_{\mathrm{Born-conv}}+(\delta f_{+})_{\mathrm{inel}} . \tag*{(10)}

The calculable piece is the so-called “Born contribution” to the integral, and within it is the “convection term” contribution that is analytically calculable and contains the full IR-divergence. We split the latter into the IR divergent piece (δf+)II(\delta f_{+})_{\mathrm{II}} and the IR-finite piece (δf+)convfin(\delta f_{+})_{\mathrm{conv}}^{\mathrm{fin}}. The remaining “Born minus convection” contribution can be computed numerically making use of the π−\pi^{-} and K+K^{+} electromagnetic form factors, for which we adopt a monopole parameterization [27, 28]. Finally, the systematic uncertainty due to the incalculable inelastic (or “non-Born”) pieces is estimated using the resonance chiral theory [29, 30, 31], see Appendix B of Ref. [24] for details. We find that their corrections to δKℓ3\delta K_{\ell3} in all channels are at most 1×10−41 \times10^{-4}, so we simply assign a conservative uncertainty of 2×10−42 \times10^{-4} to them in each channel.

  1. For δf−\delta f_{-}, we do not attempt to calculate the full Born contribution because the convergence of the loop integrals depends on the ultraviolet (UV)-behavior of the form factors, which introduces extra model-dependence to the outcome. Instead, we compute only the convection term contribution (both the IR-divergent and IR-finite pieces, of which loop integrals are explicitly UV-finite) and leave the remaining, non-convection contribution as an incalculable piece at the moment:

(δf−)this subsection=(δf−)II+(δf−)convfin+(δf−)non−conv.(11)(\delta f_{-})_{\mathrm{this\ subsection}}=(\delta f_{-})_{\mathrm{II}}+(\delta f_{-})_{\mathrm{conv}}^{\mathrm{fin}}+(\delta f_{-})_{\mathrm{non-conv}} . \tag*{(11)}

Below we provide the analytic expressions of the convection term contribution to δf±\delta f_{\pm} from (δM2+δMγWa)int+δMγWb(\delta M_{2}+\delta M^{a}_{\gamma W})_{\mathrm{int}}+\delta M^{b}_{\gamma W}:

(δf±K0π−)II=−α4π{−4p′⋅pℓxsMπmℓ(1−xs2)ln⁡xsln⁡(Mγ2Mπmℓ)f±K0π−±(52−ln⁡Mπ2Mγ2)(p′⋅(p+p′)2Mπ2f+K0π−+p′⋅(p−p′)2Mπ2f−K0π−)}\begin{aligned} (\delta f_{\pm}^{K^{0}\pi^{-}})_{\mathrm{II}}={}&-\frac{\alpha}{4\pi}\left\{-\frac{4p'\cdot p_{\ell}x_{s}}{M_{\pi}m_{\ell}(1-x_{s}^{2})}\ln x_{s}\ln\left(\frac{M_{\gamma}^{2}}{M_{\pi}m_{\ell}}\right)f_{\pm}^{K^{0}\pi^{-}}\right.\\ &\left.\pm\left(\frac{5}{2}-\ln\frac{M_{\pi}^{2}}{M_{\gamma}^{2}}\right)\left(\frac{p'\cdot(p+p')}{2M_{\pi}^{2}}f_{+}^{K^{0}\pi^{-}}+\frac{p'\cdot(p-p')}{2M_{\pi}^{2}}f_{-}^{K^{0}\pi^{-}}\right)\right\} \end{aligned}
(δf±K+π0)II=−α4π{4p⋅pℓxuMKmℓ(1−xu2)ln⁡xuln⁡(Mγ2MKmℓ)f±K+π0+(52−ln⁡MK2Mγ2)(p⋅(p+p′)2MK2f+K+π0+p⋅(p−p′)2MK2f−K+π0)}(12)\begin{aligned} (\delta f_{\pm}^{K^{+}\pi^{0}})_{\mathrm{II}}={}&-\frac{\alpha}{4\pi}\left\{\frac{4p\cdot p_{\ell}x_{u}}{M_{K}m_{\ell}(1-x_{u}^{2})}\ln x_{u}\ln\left(\frac{M_{\gamma}^{2}}{M_{K}m_{\ell}}\right)f_{\pm}^{K^{+}\pi^{0}}\right.\\ &\left.+\left(\frac{5}{2}-\ln\frac{M_{K}^{2}}{M_{\gamma}^{2}}\right)\left(\frac{p\cdot(p+p')}{2M_{K}^{2}}f_{+}^{K^{+}\pi^{0}}+\frac{p\cdot(p-p')}{2M_{K}^{2}}f_{-}^{K^{+}\pi^{0}}\right)\right\} \tag*{(12)} \end{aligned}
(δf+K0π−)convfin=−α4π{(C00fin+4p′⋅pℓC0fin+2p′⋅pℓC1−2mℓ2C2)f+K0π−+(p′⋅(p+p′)f+K0π−+p′⋅(p−p′)f−K0π−)(C1+C11/2)−(pℓ⋅(p+p′)f+K0π−+pℓ⋅(p−p′)f−K0π−)C12/2+(pℓ⋅(p′−p)+mℓ2)(f+K0π−+f−K0π−)C2}\begin{aligned} (\delta f_{+}^{K^{0}\pi^{-}})_{\mathrm{conv}}^{\mathrm{fin}}={}&-\frac{\alpha}{4\pi}\left\{\left(C_{00}^{\mathrm{fin}}+4p'\cdot p_{\ell}C_{0}^{\mathrm{fin}}+2p'\cdot p_{\ell}C_{1}-2m_{\ell}^{2}C_{2}\right)f_{+}^{K^{0}\pi^{-}}\right.\\ &+\left(p'\cdot(p+p')f_{+}^{K^{0}\pi^{-}}+p'\cdot(p-p')f_{-}^{K^{0}\pi^{-}}\right)(C_{1}+C_{11}/2)\\ &-\left(p_{\ell}\cdot(p+p')f_{+}^{K^{0}\pi^{-}}+p_{\ell}\cdot(p-p')f_{-}^{K^{0}\pi^{-}}\right)C_{12}/2\\ &\left.+\left(p_{\ell}\cdot(p'-p)+m_{\ell}^{2}\right)\left(f_{+}^{K^{0}\pi^{-}}+f_{-}^{K^{0}\pi^{-}}\right)C_{2}\right\} \end{aligned}
(δf−K0π−)convfin=−α4π{(C00fin+4p′⋅pℓC0fin+2p′⋅pℓC1−2mℓ2C2)f−K0π−−(p′⋅(p+p′)f+K0π−+p′⋅(p−p′)f−K0π−)(C1+C11/2+2C2+C12)+(pℓ⋅(p+p′)f+K0π−+pℓ⋅(p−p′)f−K0π−)(C12/2+C22)+(pℓ⋅(p−3p′)+2p⋅p′−2Mπ2−mℓ2)(f+K0π−+f−K0π−)C2}(13)\begin{aligned} (\delta f_{-}^{K^{0}\pi^{-}})_{\mathrm{conv}}^{\mathrm{fin}}={}&-\frac{\alpha}{4\pi}\left\{\left(C_{00}^{\mathrm{fin}}+4p'\cdot p_{\ell}C_{0}^{\mathrm{fin}}+2p'\cdot p_{\ell}C_{1}-2m_{\ell}^{2}C_{2}\right)f_{-}^{K^{0}\pi^{-}}\right.\\ &-\left(p'\cdot(p+p')f_{+}^{K^{0}\pi^{-}}+p'\cdot(p-p')f_{-}^{K^{0}\pi^{-}}\right)(C_{1}+C_{11}/2+2C_{2}+C_{12})\\ &+\left(p_{\ell}\cdot(p+p')f_{+}^{K^{0}\pi^{-}}+p_{\ell}\cdot(p-p')f_{-}^{K^{0}\pi^{-}}\right)(C_{12}/2+C_{22})\\ &\left.+\left(p_{\ell}\cdot(p-3p')+2p\cdot p'-2M_{\pi}^{2}-m_{\ell}^{2}\right)\left(f_{+}^{K^{0}\pi^{-}}+f_{-}^{K^{0}\pi^{-}}\right)C_{2}\right\} \tag*{(13)} \end{aligned}
(δf+K+π0)convfin=−α4π{(C00fin−4p⋅pℓC0fin−2p⋅pℓC1−2mℓ2C2)f+K+π0+(p⋅(p+p′)f+K+π0+p⋅(p−p′)f−K+π0)(C1+C11/2)+(pℓ⋅(p+p′)f+K+π0+pℓ⋅(p−p′)f−K+π0)C12/2+(pℓ⋅(p′−p)+mℓ2)(f+K+π0−f−K+π0)C2}\begin{aligned} (\delta f_{+}^{K^{+}\pi^{0}})_{\mathrm{conv}}^{\mathrm{fin}}={}&-\frac{\alpha}{4\pi}\left\{\left(C_{00}^{\mathrm{fin}}-4p\cdot p_{\ell}C_{0}^{\mathrm{fin}}-2p\cdot p_{\ell}C_{1}-2m_{\ell}^{2}C_{2}\right)f_{+}^{K^{+}\pi^{0}}\right.\\ &+\left(p\cdot(p+p')f_{+}^{K^{+}\pi^{0}}+p\cdot(p-p')f_{-}^{K^{+}\pi^{0}}\right)(C_{1}+C_{11}/2)\\ &+\left(p_{\ell}\cdot(p+p')f_{+}^{K^{+}\pi^{0}}+p_{\ell}\cdot(p-p')f_{-}^{K^{+}\pi^{0}}\right)C_{12}/2\\ &\left.+\left(p_{\ell}\cdot(p'-p)+m_{\ell}^{2}\right)\left(f_{+}^{K^{+}\pi^{0}}-f_{-}^{K^{+}\pi^{0}}\right)C_{2}\right\} \end{aligned}
(δf−K+π0)convfin=−α4π{(C00fin−4p⋅pℓC0fin−2p⋅pℓC1−2mℓ2C2)f−K+π0+(p⋅(p+p′)f+K+π0+p⋅(p−p′)f−K+π0)(C1+C11/2+2C2+C12)+(pℓ⋅(p+p′)f+K+π0+pℓ⋅(p−p′)f−K+π0)(C12/2+C22)+(pℓ⋅(p−3p′)+2p⋅p′−2Mπ2−mℓ2)(f+K+π0−f−K+π0)C2}(14)\begin{aligned} (\delta f_{-}^{K^{+}\pi^{0}})_{\mathrm{conv}}^{\mathrm{fin}}={}&-\frac{\alpha}{4\pi}\left\{\left(C_{00}^{\mathrm{fin}}-4p\cdot p_{\ell}C_{0}^{\mathrm{fin}}-2p\cdot p_{\ell}C_{1}-2m_{\ell}^{2}C_{2}\right)f_{-}^{K^{+}\pi^{0}}\right.\\ &+\left(p\cdot(p+p')f_{+}^{K^{+}\pi^{0}}+p\cdot(p-p')f_{-}^{K^{+}\pi^{0}}\right)(C_{1}+C_{11}/2+2C_{2}+C_{12})\\ &+\left(p_{\ell}\cdot(p+p')f_{+}^{K^{+}\pi^{0}}+p_{\ell}\cdot(p-p')f_{-}^{K^{+}\pi^{0}}\right)(C_{12}/2+C_{22})\\ &\left.+\left(p_{\ell}\cdot(p-3p')+2p\cdot p'-2M_{\pi}^{2}-m_{\ell}^{2}\right)\left(f_{+}^{K^{+}\pi^{0}}-f_{-}^{K^{+}\pi^{0}}\right)C_{2}\right\} \tag*{(14)} \end{aligned}

We refer the reader to Appendix C in Ref. [24] for the definitions of xvx_v and the CC-functions; in Kℓ30K^0_{\ell3} their arguments are m1=Mπm_1=M_\pi, m2=mℓm_2=m_\ell, v=sv=s, whereas in Kℓ3+K^+_{\ell3} their arguments are m1=MKm_1=M_K, m2=mℓm_2=m_\ell, v=uv=u. We defer the discussion of their numerical impact on δKℓ3\delta_{K\ell3} (as well as the “Born minus convection” contribution to δf+\delta f_+) to Sections IV and V.

From the three-point function

Next we have the so-called “three-point function” correction to the form factors which we denote as δf±,3\delta f_{\pm,3}, see Eq. (5.14) in Ref. [19]. The same reference provides an approximate expression for these quantities to O(e2p2)\mathcal{O}(e^2p^2) in the chiral power counting (see Sec. 8 of that paper), but naїvely applying those expressions here will cause an incomplete cancellation of IR-divergences. Therefore, adopting the strategy in Sec. 5 of Ref. [24], we split the full δf±,3\delta f_{\pm,3} into the IR-divergent and the IR-finite pieces:

δf±,3=(δf±)III+(δf±,3)fin.(15)\delta f_{\pm,3}=(\delta f_{\pm})_{\mathrm{III}}+(\delta f_{\pm,3})_{\mathrm{fin}} . \tag*{(15)}

Starting from its O(e2p2)\mathcal{O}(e^2p^2) expression, one could resum the IR-divergent piece (δf±)III(\delta f_{\pm})_{\mathrm{III}} to all orders in the chiral expansion by simply adding back the full form factors and to ensure the exact cancellation of IR-divergences between the real and virtual corrections, as we pointed out in Sec. 5 of Ref. [24]. The resummed version reads:

(δf±K0π−)III=∓α4π(ln⁡Mπ2Mγ2−52)p′⋅(p∓p′)2Mπ2(f+K0π−+f−K0π−)(\delta f^{K^0\pi^-}_{\pm})_{\mathrm{III}}=\mp\frac{\alpha}{4\pi}\left(\ln\frac{M_\pi^2}{M_\gamma^2}-\frac{5}{2}\right)\frac{p'\cdot(p\mp p')}{2M_\pi^2}\left(f^{K^0\pi^-}_++f^{K^0\pi^-}_-\right)
(δf±K+π0)III=±α4π(ln⁡MK2Mγ2−52)p⋅(p∓p′)2MK2(f+K+π0−f−K+π0).(16)(\delta f^{K^+\pi^0}_{\pm})_{\mathrm{III}}=\pm\frac{\alpha}{4\pi}\left(\ln\frac{M_K^2}{M_\gamma^2}-\frac{5}{2}\right)\frac{p\cdot(p\mp p')}{2M_K^2}\left(f^{K^+\pi^0}_+-f^{K^+\pi^0}_-\right) . \tag*{(16)}

In the following, it is sufficient to adopt an O(e2p2)\mathcal{O}(e^2p^2) approximation to (δf±,3)fin(\delta f_{\pm,3})_{\mathrm{fin}}, which could be inferred from Ref. [19]. However, there is one further complication as discussed in Sec. 8 of Ref. [24], namely: Following the standard ChPT treatment, some of the terms in (δf±,3)fin(\delta f_{\pm,3})_{\mathrm{fin}} are in fact not counted as a part of the long-distance electromagnetic corrections, but rather redistributed into the extra tt-dependence of the form factors (Eqs. (5.3), (5.7), (5.11), (5.12) in Ref. [32]) as well as the isospin-breaking correction δSU(2)Kπ\delta^{K\pi}_{\mathrm{SU}(2)} through the electromagnetically-induced π0−η\pi^0-\eta mixing (Eq. (5.5) in Ref. [32]). Therefore, in order to appropriately compare with the existing literature, we have to remove these terms from our definitions of (δf±,3)fin(\delta f_{\pm,3})_{\mathrm{fin}} as well. After doing so, we obtain the following O(e2p2)\mathcal{O}(e^2p^2) expressions:

(δf+,3K0π−)fine2p2=0(δf+,3K+π0)fine2p2=0(δf−,3K0π−)fine2p2=α4π[34ln⁡Mπ2μ2−1]−Zα8πln⁡Mπ2μ2+8πα3(K5r+K6r)(δf−,3K+π0)fine2p2=−α42π[34ln⁡MK2μ2−1]−5Zα82πln⁡MK2μ2+8πα2(−2K3r+K4r+13K5r+13K6r).(17)\begin{aligned} (\delta f_{+,3}^{K^0\pi^-})_{\mathrm{fin}}^{e^2p^2} &= 0 \\ (\delta f_{+,3}^{K^+\pi^0})_{\mathrm{fin}}^{e^2p^2} &= 0 \\ (\delta f_{-,3}^{K^0\pi^-})_{\mathrm{fin}}^{e^2p^2} &= \frac{\alpha}{4\pi}\left[\frac{3}{4}\ln\frac{M_\pi^2}{\mu^2}-1\right]-\frac{Z\alpha}{8\pi}\ln\frac{M_\pi^2}{\mu^2}+\frac{8\pi\alpha}{3}(K_5^r+K_6^r) \\ (\delta f_{-,3}^{K^+\pi^0})_{\mathrm{fin}}^{e^2p^2} &= -\frac{\alpha}{4\sqrt{2}\pi}\left[\frac{3}{4}\ln\frac{M_K^2}{\mu^2}-1\right]-\frac{5Z\alpha}{8\sqrt{2}\pi}\ln\frac{M_K^2}{\mu^2}+\frac{8\pi\alpha}{\sqrt{2}}\left(-2K_3^r+K_4^r+\frac{1}{3}K_5^r+\frac{1}{3}K_6^r\right). \tag*{(17)} \end{aligned}

where Z≈0.8Z\approx0.8 and KirK_i^r are low-energy constants (LECs) in ChPT, with μ\mu the scale of dimensional regularization, which will be taken as Mρ=770M_\rho=770 MeV in the numerical analysis.

From δMγWb,A\delta M_{\gamma W}^{b,A}

Finally, we have the contribution from the axial γW\gamma W-box amplitude δMγWb,A\delta M_{\gamma W}^{b,A}, namely the axial charged weak current contribution to the loop integral δMγWb\delta M_{\gamma W}^{b} defined in Eq. (3.10) of Ref. [24]. In Sec. 6 of that paper, we outlined the strategy to fix its contribution to δf+\delta f_+, namely to make use of the existing lattice QCD calculations of the forward axial γW\gamma W-box diagrams in the meson sector. In short, we can write:

(δf+)γWb,A={□γWVA> ⁣+[□γWVA<(K,π,Mπ)+O(MK2Λχ2)]}f+(t).(18)(\delta f_+)_{\gamma W}^{b,A}=\left\{\Box_{\gamma W}^{VA>}\!+\left[\Box_{\gamma W}^{VA<}(K,\pi,M_\pi)+\mathcal{O}\left(\frac{M_K^2}{\Lambda_\chi^2}\right)\right]\right\}f_+(t). \tag*{(18)}

First, □γWVA>≈2.16×10−3\Box_{\gamma W}^{VA>}\approx2.16\times10^{-3} is the channel-independent component originating from the integral of the forward box diagram at Q2>2Q^2>2 GeV2^2, which was calculated to high precision from perturbative QCD. In the meantime, □γWVA<(K,π,Mπ)\Box_{\gamma W}^{VA<}(K,\pi,M_\pi) is the remaining, channel-dependent part of the forward KπK\pi box diagram calculated in the flavor SU(3) limit MK=MπM_K=M_\pi. Existing lattice calculations are for □γWVA<(π+,π0,Mπ)\Box_{\gamma W}^{VA<}(\pi^+,\pi^0,M_\pi) [33] and □γWVA<(K0,π−,Mπ)\Box_{\gamma W}^{VA<}(K^0,\pi^-,M_\pi) [21], from which we can also obtain □γWVA<(K+,π0,Mπ)=2□γWVA<(π+,π0,Mπ)−□γWVA<(K0,π−,Mπ)\Box_{\gamma W}^{VA<}(K^+,\pi^0,M_\pi)=2\Box_{\gamma W}^{VA<}(\pi^+,\pi^0,M_\pi)-\Box_{\gamma W}^{VA<}(K^0,\pi^-,M_\pi) through a ChPT matching [20]. The uncertainties due to the non-forward (NF) corrections are conservatively estimated by multiplying the central values of □γWVA<\Box_{\gamma W}^{VA<} by the factor MK2/Λχ2M_K^2/\Lambda_\chi^2, where Λ=4πFπ≈1.2\Lambda=4\pi F_\pi\approx1.2 GeV is the chiral symmetry breaking scale.

Unfortunately, these lattice calculations of the forward axial γW\gamma W-box diagrams do not provide any useful information for (δf−)γWb,A(\delta f_-)_{\gamma W}^{b,A} because p−p′=0p-p'=0 by definition. So at this point we just leave it as another incalculable piece.

Matching the incalculable pieces in δf−\delta f_- to ChPT

So far we have discussed the full O(α)\mathcal{O}(\alpha) electroweak RC to δf±\delta f_{\pm}. For δf+\delta f_+ we have:

δf+=(δf+)I+II+III+(δf+)convfin+(δf+)Born−conv+(δf+)inel+(δf+,3)fin+(δf+)γWb,A.(19)\delta f_+ = (\delta f_+)_{\mathrm{I+II+III}} + (\delta f_+)_{\mathrm{conv}}^{\mathrm{fin}} + (\delta f_+)_{\mathrm{Born-conv}} + (\delta f_+)_{\mathrm{inel}} + (\delta f_{+,3})_{\mathrm{fin}} + (\delta f_+)_{\gamma W}^{\mathrm{b,A}} . \tag*{(19)}

From the discussions above, we see that each term at the right-hand side can either be determined to high precision or is small enough to assign a controllable theory uncertainty.

Meanwhile, for δf−\delta f_- we have:

δf−=(δf−)I+II+III+(δf−)convfin+(δf−)non−conv+(δf−,3)fin+(δf−)γWb,A,(20)\delta f_- = (\delta f_-)_{\mathrm{I+II+III}} + (\delta f_-)_{\mathrm{conv}}^{\mathrm{fin}} + (\delta f_-)_{\mathrm{non-conv}} + (\delta f_{-,3})_{\mathrm{fin}} + (\delta f_-)_{\gamma W}^{\mathrm{b,A}} , \tag*{(20)}

among which (δf−)I+II+III(\delta f_-)_{\mathrm{I+II+III}} and (δf−)convfin(\delta f_-)_{\mathrm{conv}}^{\mathrm{fin}} are exactly known, (δf−,3)fin(\delta f_{-,3})_{\mathrm{fin}} is known to O(e2p2)\mathcal{O}(e^2p^2), while (δf−)non−conv(\delta f_-)_{\mathrm{non-conv}} and (δf−)γWb,A(\delta f_-)_{\gamma W}^{\mathrm{b,A}} are so far completely unknown. Fortunately, since they are IR-regular terms whose contributions to δKℓ3\delta K_{\ell3} are naturally suppressed by rℓr_\ell, there is a simple strategy to deal with them: Existing ChPT studies provided the O(e2p2)\mathcal{O}(e^2p^2) expression for the full δf−\delta f_- (i.e. the left-hand side of Eq. (20)) [32], and we can further take the O(e2p2)\mathcal{O}(e^2p^2) approximation for (δf−)I+II+III(\delta f_-)_{\mathrm{I+II+III}} and (δf−)convfin(\delta f_-)_{\mathrm{conv}}^{\mathrm{fin}} by replacing:

f+K0π−(t)→−1,f+K+π0(t)→−12,f−(t)→0.(21)f_+^{K^0\pi^-}(t) \to-1,\qquad f_+^{K^+\pi^0}(t) \to-\frac{1}{\sqrt{2}},\qquad f_-(t) \to0 . \tag*{(21)}

By doing so, one could then equate both sides in Eq. (20) to extract the O(e2p2)\mathcal{O}(e^2p^2) expression of (δf−)non−conv+(δf−)γWb,A(\delta f_-)_{\mathrm{non-conv}} + (\delta f_-)_{\gamma W}^{\mathrm{b,A}}. As a consequence, we arrive the following representation:

δf−=(δf−)I+II+III+(δf−)convfin+(δf−)rem(22)\delta f_- = (\delta f_-)_{\mathrm{I+II+III}} + (\delta f_-)_{\mathrm{conv}}^{\mathrm{fin}} + (\delta f_-)_{\mathrm{rem}} \tag*{(22)}

where the first two terms at the right-hand side are exactly known, while the remaining piece (δf−)rem(\delta f_-)_{\mathrm{rem}} adopts an O(e2p2)\mathcal{O}(e^2p^2) approximation:

(δf−K0π−)reme2p2=−α4π[r0,1Λ(s,Mπ,mℓ)+r0,2ln⁡Mπ2mℓ2+r0,3ln⁡Mπ2μ2+r0,4](\delta f_-^{K^0\pi^-})_{\mathrm{rem}}^{e^2p^2} = -\frac{\alpha}{4\pi}\left[r_{0,1}\Lambda(s,M_\pi,m_\ell)+r_{0,2}\ln\frac{M_\pi^2}{m_\ell^2}+r_{0,3}\ln\frac{M_\pi^2}{\mu^2}+r_{0,4}\right]
(δf−K+π0)reme2p2=−α42π[r+,1Λ(u,MK,mℓ)+r+,2ln⁡MK2mℓ2+r+,3ln⁡MK2μ2+r+,4],(23)(\delta f_-^{K^+\pi^0})_{\mathrm{rem}}^{e^2p^2} = -\frac{\alpha}{4\sqrt{2}\pi}\left[r_{+,1}\Lambda(u,M_K,m_\ell)+r_{+,2}\ln\frac{M_K^2}{m_\ell^2}+r_{+,3}\ln\frac{M_K^2}{\mu^2}+r_{+,4}\right] , \tag*{(23)}

where the function Λ(v,m1,m2)\Lambda(v,m_1,m_2) is defined in Eq. (C.5) of Ref. [24]. The coefficients ri,jr_{i,j} are given by:

r0,1=−18λs{MK2[2s(mℓ2−2Mπ2)+(mℓ2−Mπ2)2+3s2]−mℓ6+mℓ4[3Mπ2+s]−mℓ2[3Mπ4+13Mπ2s+s(t−20s)]+Mπ6+12Mπ4s+7Mπ2s2+st[Mπ2−5s]−20s3}\begin{aligned} r_{0,1}={}&-\frac{1}{8\lambda s}\Bigl\{M_K^2[2s(m_\ell^2-2M_\pi^2)+(m_\ell^2-M_\pi^2)^2+3s^2]-m_\ell^6+m_\ell^4[3M_\pi^2+s]\\ &-m_\ell^2[3M_\pi^4+13M_\pi^2s+s(t-20s)]+M_\pi^6+12M_\pi^4s+7M_\pi^2s^2+st[M_\pi^2-5s]\\ &-20s^3\Bigr\} \end{aligned}
r0,2=116s2[MK2(−mℓ2+Mπ2−3s)+mℓ4−2mℓ2Mπ2+Mπ4+13Mπ2s+28s2+st]r_{0,2}=\frac{1}{16s^2}\left[M_K^2(-m_\ell^2+M_\pi^2-3s)+m_\ell^4-2m_\ell^2M_\pi^2+M_\pi^4+13M_\pi^2s+28s^2+st\right]
r0,3=Z−62r_{0,3}=\frac{Z-6}{2}
r0,4=32π2[X1+X2r−X3r−(K5r+K6r)/3]+mℓ2−MK2−Mπ28s+34(24)r_{0,4}=32\pi^2\left[X_1+X_2^{\mathrm{r}}-X_3^{\mathrm{r}}-(K_5^{\mathrm{r}}+K_6^{\mathrm{r}})/3\right]+\frac{m_\ell^2-M_K^2-M_\pi^2}{8s}+\frac{3}{4} \tag*{(24)}

and

r+,1=18λu{u2[7MK2+20mℓ2+3Mπ2−5t]+[MK2−mℓ2]2[MK2−mℓ2+Mπ2]+u[12MK4+MK2(−13mℓ2−4Mπ2+t)+mℓ2(mℓ2+2Mπ2−t)]−20u3}\begin{aligned} r_{+,1}={}&\frac{1}{8\lambda u}\Bigl\{u^2[7M_K^2+20m_\ell^2+3M_\pi^2-5t]+[M_K^2-m_\ell^2]^2[M_K^2-m_\ell^2+M_\pi^2]\\ &+u[12M_K^4+M_K^2(-13m_\ell^2-4M_\pi^2+t)+m_\ell^2(m_\ell^2+2M_\pi^2-t)]-20u^3\Bigr\} \end{aligned}
r+,2=−116u2[(MK2−mℓ2)(MK2−mℓ2+Mπ2)+u(13MK2−3Mπ2+t)+28u2]r_{+,2}=-\frac{1}{16u^2}\left[(M_K^2-m_\ell^2)(M_K^2-m_\ell^2+M_\pi^2)+u(13M_K^2-3M_\pi^2+t)+28u^2\right]
r+,3=5Z+62r_{+,3}=\frac{5Z+6}{2}
r+,4=32π2[X1−X2r+X3r+2K3r−K4r−(K5r+K6r)/3]+MK2+Mπ2−mℓ28u−34(25)r_{+,4}=32\pi^2\left[X_1-X_2^{\mathrm{r}}+X_3^{\mathrm{r}}+2K_3^{\mathrm{r}}-K_4^{\mathrm{r}}-(K_5^{\mathrm{r}}+K_6^{\mathrm{r}})/3\right]+\frac{M_K^2+M_\pi^2-m_\ell^2}{8u}-\frac{3}{4} \tag*{(25)}

where λ=λ(m12,m22,v)\lambda=\lambda(m_1^2,m_2^2,v) is the Källén function.

Real corrections

Our previous treatment of the Ke3K_{e3} bremsstrahlung correction detailed in Sec. 7 of Ref. [24] can be directly applied to all Kℓ3K_{\ell3} channels without any further modification. During the bremsstrahlung, the real photon can be emitted either by the charged lepton or by the meson; in the latter case, the decay amplitude involves the generalized, non-forward Compton tensor Tμν(q′;p′,p)T_{\mu\nu}(q';p',p) (see Eq. (3.5) in Ref. [24]). Our strategy is to split TμνT_{\mu\nu} into the convection term contribution and a remainder:

Tμν=Tconvμν+[Tμν−Tconvμν].(26)T^{\mu\nu}=T_{\mathrm{conv}}^{\mu\nu}+[T^{\mu\nu}-T_{\mathrm{conv}}^{\mu\nu}] . \tag*{(26)}

In practice, we retain the full TconvμνT_{\mathrm{conv}}^{\mu\nu} but take the O(p2)\mathcal{O}(p^2) approximation for Tμν−TconvμνT^{\mu\nu}-T_{\mathrm{conv}}^{\mu\nu}; this simple implementation ensures exact electromagnetic gauge invariance. Consequently, we split the full tree-level K→πℓ+νℓγK\to\pi\ell^+\nu_\ell\gamma amplitudes into two pieces:

MK→πℓ+νℓγ=MA+MB,(27)M_{K\to\pi\ell^+\nu_\ell\gamma}=M_A+M_B , \tag*{(27)}

where Tμν−TconvμνT^{\mu\nu}-T_{\mathrm{conv}}^{\mu\nu} is fully contained in MBM_B. Therefore, in calculating the squared amplitude, the result of ∣MA∣2|M_A|^2 is exact while 2ℜ{MA∗MB}+∣MB∣22\Re\{M_A^*M_B\}+|M_B|^2 acquires a chiral expansion uncertainty.

The bremsstrahlung process resides at a larger allowed phase space area D4=D3⊕D4−3\mathcal{D}_4=\mathcal{D}_3\oplus\mathcal{D}_{4-3} (see Appendix A of Ref. [24]). The IR-divergence is fully contained in ∣MA∣2|M_A|^2 integrating over the D3\mathcal{D}_3 region, and is exactly canceled by those in (δf±)I+II+III(\delta f_\pm)_{\mathrm{I+II+III}} from the virtual corrections.

Numerical results, errors and correlations

Now we are ready to present the numerical results of δKℓ3\delta K_{\ell3} in all channels. The IR-finite contribution from the virtual corrections is summarized in Table I, the combination of the IR-divergent parts in the virtual corrections and the bremsstrahlung corrections in the D3\mathcal{D}_3 region is given in Table II, and the bremsstrahlung contribution in the D4−3\mathcal{D}_{4-3} region is given in tbl:3. Columns in these three tables with an asterisk (*) denote quantities that have taken an O(e2p2)\mathcal{O}(e^2p^2) approximation and are subject to chiral uncertainties at O(e2p4)\mathcal{O}(e^2p^4).

δKℓ3\delta K_{\ell3}(δf+)convfin(\delta f_+)_{\mathrm{conv}}^{\mathrm{fin}}(δf+)Born−conv(\delta f_+)_{\mathrm{Born-conv}}(δf+)γWb,A(\delta f_+)_{\gamma W}^{b,A}(∗)  (δf−)rem(*)\;(\delta f_-)_{\mathrm{rem}}
Ke30K^0_{e3}−5.0-5.04.14.14.9(1)lat(1)NF4.9(1)_{\mathrm{lat}}(1)_{\mathrm{NF}}0.00.0
Ke3+K^+_{e3}9.69.60.10.16.4(1)lat(4)NF6.4(1)_{\mathrm{lat}}(4)_{\mathrm{NF}}0.00.0
Kμ30K^0_{\mu3}18.618.60.40.44.9(1)lat(1)NF4.9(1)_{\mathrm{lat}}(1)_{\mathrm{NF}}0.3(2)LEC0.3(2)_{\mathrm{LEC}}
Kμ3+K^+_{\mu3}−0.1-0.11.01.06.5(1)lat(4)NF6.5(1)_{\mathrm{lat}}(4)_{\mathrm{NF}}−1.1(2)LEC-1.1(2)_{\mathrm{LEC}}

Table I. Contribution to δKℓ3\delta K_{\ell3} from (δf+)convfin(\delta f_+)_{\mathrm{conv}}^{\mathrm{fin}}, (δf+)Born−conv(\delta f_+)_{\mathrm{Born-conv}}, (δf+)γWb,A(\delta f_+)_{\gamma W}^{b,A} and (δf−)rem(\delta f_-)_{\mathrm{rem}}, in units of 10−310^{-3}.

(δKℓ3)I+II+III+brem(D3)(\delta K_{\ell3})_{\mathrm{I+II+III+brem}(D_3)}(∗)(*) From 2ℜ{MA∗MB}+∣MB∣22\Re\{M_A^*M_B\}+|M_B|^2Remainder
Ke30K^0_{e3}1.01.024.124.1
Ke3+K^+_{e3}−0.3-0.34.44.4
Kμ30K^0_{\mu3}0.60.613.413.4
Kμ3+K^+_{\mu3}−0.1-0.117.417.4

Table II. Sum of the IR-divergent one-loop contributions I, II, III and the bremsstrahlung contribution in the D3\mathcal{D}_3 region, in units of 10−310^{-3}.

(δKℓ3)brem(D4−3)(\delta K_{\ell3})_{\mathrm{brem}}(D_{4-3})(∗)(*) From 2Re⁡{MA∗MB}+∣MB∣22\operatorname{Re}\{M_A^*M_B\}+|M_B|^2From ∣MA∣2|M_A|^2
Ke30K^0_{e3}0.20.25.65.6
Ke3+K^+_{e3}−0.1-0.15.35.3
Kμ30K^0_{\mu3}0.10.10.20.2
Kμ3+K^+_{\mu3}0.00.00.10.1

caption: Table III: The bremsstrahlung contribution in the D4−3D_{4-3} region, in units of 10−310^{-3}.

:::

For the error analysis we include only theory uncertainties that after a single rounding-up are of the size 1×10−41 \times10^{-4} or larger. After subtracting out SEW−1S_{\mathrm{EW}}-1 and its associated short-distance error, there are five major sources of theory uncertainties:1

  1. The lattice uncertainty in (δf+)γWb,A(\delta f_+)_{\gamma W}^{b,A},

  2. The LEC uncertainties in (δf−)rem(\delta f_-)_{\mathrm{rem}},

  3. The NF uncertainty in (δf+)γWb,A(\delta f_+)_{\gamma W}^{b,A},

  4. The O(e2p4)\mathcal{O}(e^2p^4) chiral uncertainty, and

  5. The unknown (δf+)inel(\delta f_+)_{\mathrm{inel}} in Eq. (10).

The first three are already displayed in Table I, but the last two not yet. In any case, in order to discuss later the average of ∣Vus∣|V_{us}| over different channels, it is not sufficient to know just the uncertainties in each separate channel, but their correlations must also be understood. To do so one needs to first identify the independent theory inputs that carry the uncertainties, and then the correlation matrix can be calculated using the formula outlined in the Appendix of Ref. [25]. This can be rigorously done for the first two types, which we describe below:

  • For (δf+)γWb,A(\delta f_+)_{\gamma W}^{b,A}, the independent lattice QCD inputs are [21, 33]:

VγWA<(π+,π0,Mπ)=0.671(28)lat×10−3,VγWA<(K0,π−,Mπ)=0.278(44)lat×10−3.(28)\mathcal{V}_{\gamma W}^{A<}(\pi^+,\pi^0,M_\pi)=0.671(28)_{\mathrm{lat}}\times10^{-3},\qquad\mathcal{V}_{\gamma W}^{A<}(K^0,\pi^-,M_\pi)=0.278(44)_{\mathrm{lat}}\times10^{-3}. \tag*{(28)}
  • For (δf−)rem(\delta f_{-})_{\mathrm{rem}}, the independent combinations of LECs are X1X_{1}, C1≡X2r−X3rC_{1} \equiv X_{2}^{r} - X_{3}^{r}, C2≡2K3r−K4rC_{2} \equiv2K_{3}^{r} - K_{4}^{r} and C3≡K5r+K6rC_{3} \equiv K_{5}^{r} + K_{6}^{r}. Among them, X1=−2.2(4)×10−3X_{1} = -2.2(4) \times10^{-3} was fixed to good precision with the recent lattice calculations [21], and its resulting uncertainty to δKℓ3\delta_{K\ell3} is negligible. Similar calculations are not yet done for C1−3C_{1-3}, so we infer their values at μ=Mρ\mu= M_{\rho} from resonance models [34, 35, 36], and assign a 100% uncertainty to each of them:

C1=−1.4(1.4)LEC×10−3,C2=4.0(4.0)LEC×10−3,C3=14.4(14.4)LEC×10−3.(29)C_{1} = -1.4(1.4)_{\mathrm{LEC}} \times10^{-3}, \quad C_{2} = 4.0(4.0)_{\mathrm{LEC}} \times10^{-3}, \quad C_{3} = 14.4(14.4)_{\mathrm{LEC}} \times10^{-3}. \tag*{(29)}

Meanwhile, the next three uncertainties are estimated as follows:

  • The NF uncertainty in (δf+)γWb,A(\delta f_{+})_{\gamma W}^{b,A} is estimated by multiplying □γWVA<\square_{\gamma W}^{VA<} in each channel by MK2/Λχ2M_{K}^{2}/\Lambda_{\chi}^{2};

  • The O(e2p4)\mathcal{O}(e^{2}p^{4}) chiral uncertainty is obtained by first adding all the columns with asterisks in Table I–III, and then multiply the sum by MK2/Λχ2M_{K}^{2}/\Lambda_{\chi}^{2};

  • Finally, a conservative uncertainty of 2×10−42 \times10^{-4} is assign to each channel to account for the poorly-constrained contribution from (δf+)inel(\delta f_{+})_{\mathrm{inel}} (see discussions in § 3 B).

Unlike the first two, these three errors are deduced using naïve power counting and order-of-magnitude estimations, and it is difficult to identify independent sources of uncertainties within each type. In fact, we consider it as arbitrary to take these uncertainties to be uncorrelated as to assume any correlation. Therefore, we simply take them to be uncorrelated, following the same strategy adopted by some of us in Ref. [25].

We present our final result of δEMKℓ\delta_{\mathrm{EM}}^{K\ell} in Table IV, with the correlation matrix given by:

δEMKℓ\delta_{\mathrm{EM}}^{K\ell}ChPT
K0eK^{0}e11.6(2)inel(1)lat(1)NF(2)e2p411.6(2)_{\mathrm{inel}}(1)_{\mathrm{lat}}(1)_{\mathrm{NF}}(2)_{e^{2}p^{4}}9.9(1.9)e2p4(1.1)LEC9.9(1.9)_{e^{2}p^{4}}(1.1)_{\mathrm{LEC}}
K+eK^{+}e2.1(2)inel(1)lat(4)NF(1)e2p42.1(2)_{\mathrm{inel}}(1)_{\mathrm{lat}}(4)_{\mathrm{NF}}(1)_{e^{2}p^{4}}1.0(1.9)e2p4(1.6)LEC1.0(1.9)_{e^{2}p^{4}}(1.6)_{\mathrm{LEC}}
K0μK^{0}\mu15.4(2)inel(1)lat(1)NF(2)LEC(2)e2p415.4(2)_{\mathrm{inel}}(1)_{\mathrm{lat}}(1)_{\mathrm{NF}}(2)_{\mathrm{LEC}}(2)_{e^{2}p^{4}}14.0(1.9)e2p4(1.1)LEC14.0(1.9)_{e^{2}p^{4}}(1.1)_{\mathrm{LEC}}
K+μK^{+}\mu0.5(2)inel(1)lat(4)NF(2)LEC(2)e2p40.5(2)_{\mathrm{inel}}(1)_{\mathrm{lat}}(4)_{\mathrm{NF}}(2)_{\mathrm{LEC}}(2)_{e^{2}p^{4}}0.2(1.9)e2p4(1.6)LEC0.2(1.9)_{e^{2}p^{4}}(1.6)_{\mathrm{LEC}}

Table IV. Final result for δEMKℓ\delta_{\mathrm{EM}}^{K\ell}, in units of 10−310^{-3}. The ChPT result from Ref. [22] is given in the last column for comparison.

Corr⁡(δEM)=(1−0.0500.069−0.0431−0.0490.07910.0881)(30)\operatorname{Corr}(\delta_{\mathrm{EM}})= \begin{pmatrix} 1 & -0.050 & 0.069 & -0.043 \\ & 1 & -0.049 & 0.079 \\ & & 1 & 0.088 \\ & & & 1 \end{pmatrix} \tag*{(30)}

where δEM=(δEMK0e δEMK+e δEMK0μ δEMK+μ)T\delta_{\mathrm{EM}}=\left(\delta_{\mathrm{EM}}^{K^{0}e}\ \delta_{\mathrm{EM}}^{K^{+}e}\ \delta_{\mathrm{EM}}^{K^{0}\mu}\ \delta_{\mathrm{EM}}^{K^{+}\mu}\right)^{T}. The results in the two KeK e channels are obviously the same as in Ref. [24]. As a comparison, we also quote the ChPT result from Ref. [22]. Our new determinations agree with them within error bars in all four channels, but with significant improvements in precision by almost an order of magnitude.

With these new theory inputs of δEMKℓ\delta_{\mathrm{EM}}^{K\ell}, we also present here an updated global analysis of ∣Vusf+K0π−(0)∣|V_{us}f_{+}^{K^{0}\pi^{-}}(0)| from semileptonic kaon decays. Essentially, we take all the theory and experimental inputs quoted in Ref. [25], and only replace the values of δEM\delta_{\mathrm{EM}} and Corr⁡(δEM)\operatorname{Corr}(\delta_{\mathrm{EM}}) by our new results. The outcome is summarized in Table V, which has only very minor changes comparing to Table I in Ref. [25]. Supplementing the result with the most recent FLAG averages of ∣f+K0π−(0)∣|f_{+}^{K^{0}\pi^{-}}(0)| with different numbers of active quark flavors [37]:

∣Vusf+K0π−(0)∣|V_{us}f_{+}^{K^{0}\pi^{-}}(0)|Correlation Matrix
KLeK_{L e}0.21617(46)exp⁡(10)IK(4)δEM0.21617(46)_{\exp}(10)_{I_K}(4)_{\delta_{\mathrm{EM}}}10.0210.0250.5670.0040.0171\quad0.021\quad0.025\quad0.567\quad0.004\quad0.017
KSeK_{S e}0.21530(122)exp⁡(10)IK(4)δEM0.21530(122)_{\exp}(10)_{I_K}(4)_{\delta_{\mathrm{EM}}}10.0090.0140.0000.0061\quad0.009\quad0.014\quad0.000\quad0.006
K+eK^{+}e0.21714(88)exp⁡(10)IK(21)δSU(2)(5)δEM0.21714(88)_{\exp}(10)_{I_K}(21)_{\delta_{\mathrm{SU}(2)}}(5)_{\delta_{\mathrm{EM}}}10.0180.0020.8941\quad0.018\quad0.002\quad0.894
KLμK_{L\mu}0.21649(50)exp⁡(16)IK(4)δEM0.21649(50)_{\exp}(16)_{I_K}(4)_{\delta_{\mathrm{EM}}}10.0110.0441\quad0.011\quad0.044
KSμK_{S\mu}0.21251(466)exp⁡(16)IK(4)δEM0.21251(466)_{\exp}(16)_{I_K}(4)_{\delta_{\mathrm{EM}}}10.0051\quad0.005
K+μK^{+}\mu0.21699(108)exp⁡(16)IK(21)δSU(2)(6)δEM0.21699(108)_{\exp}(16)_{I_K}(21)_{\delta_{\mathrm{SU}(2)}}(6)_{\delta_{\mathrm{EM}}}11
Average: KeK e0.21626(40)K(3)HO0.21626(40)_{K}(3)_{\mathrm{HO}}
Average: KμK\mu0.21654(48)K(3)HO0.21654(48)_{K}(3)_{\mathrm{HO}}
Average: tot0.21634(38)K(3)HO0.21634(38)_{K}(3)_{\mathrm{HO}}

Table V. Updated determination of ∣Vusf+K0π−(0)∣|V_{us}f_{+}^{K^{0}\pi^{-}}(0)|, which should be compared to Table I in Ref. [25].

∣f+K0π−(0)∣={0.9698(17)Nf=2+1+1Refs. citep‘[38,39]‘0.9677(27)Nf=2+1Refs. citep‘[40,41]‘(31)|f_{+}^{K^{0}\pi^{-}}(0)|= \begin{cases} 0.9698(17) & N_f=2+1+1\qquad\text{Refs. }{citep}`[38, 39]` \\ 0.9677(27) & N_f=2+1\qquad\text{Refs. }{citep}`[40, 41]` \end{cases} \tag*{(31)}

we obtain:

∣Vus∣Kℓ3={0.22308(39)lat(39)K(3)HONf=2+1+10.22356(62)lat(39)K(3)HONf=2+1(32)\lvert V_{us}\rvert_{K_{\ell3}} = \begin{cases} 0.22308(39)_{\mathrm{lat}}(39)_{K}(3)_{\mathrm{HO}} & N_f = 2 + 1 + 1 \\ 0.22356(62)_{\mathrm{lat}}(39)_{K}(3)_{\mathrm{HO}} & N_f = 2 + 1 \end{cases} \tag*{(32)}

respectively.

The value of ∣Vus∣\lvert V_{us}\rvert can also be extracted by first obtaining ∣Vus/Vud∣\lvert V_{us}/V_{ud}\rvert from the ratio between the leptonic kaon (Kμ2K_{\mu2}) and pion (πμ2\pi_{\mu2}) decay rates, and then substituting ∣Vud∣\lvert V_{ud}\rvert by its most precise determination from superallowed nuclear decays [42]. With this method, we obtain [9]:

∣Vus∣Kμ2={0.2252(5)Nf=2+1+10.2255(8)Nf=2+1(33)\lvert V_{us}\rvert_{K_{\mu2}} = \begin{cases} 0.2252(5) & N_f = 2 + 1 + 1 \\ 0.2255(8) & N_f = 2 + 1 \end{cases} \tag*{(33)}

Comparing Eqs. (32) and (33), we see that the so-called Kℓ3−Kμ2K_{\ell3} - K_{\mu2} discrepancy in the ∣Vus∣\lvert V_{us}\rvert determination is still very much alive.

Conclusion

In this work we present for the first time a unified analysis of the long-distance electromagnetic RC to all channels in Kℓ3K_{\ell3} decays based on the novel hybrid method combining Sirlin’s representation with ChPT, that we developed a few years ago. With an appropriate use of the electromagnetic/charged weak form factors, lattice QCD inputs of the forward axial γW\gamma W-box as well as the existing knowledge of the remaining, poorly-constrained LECs, we are able to control the theory uncertainties in all four independent channels in Kℓ3K_{\ell3} at the level of 10−410^{-4}.

Several important implications of this new calculation are as follows. First, no significant change of δEMKℓ\delta_{\mathrm{EM}}^{K_{\ell}} from the previous ChPT result is observed; this allows us to state with more confidence that the outstanding discrepancy between the ∣Vus∣\lvert V_{us}\rvert extracted from Kμ2/πμ2K_{\mu2}/\pi_{\mu2} and Kℓ3K_{\ell3} does not originate from the long-distance electromagnetic corrections. Second, separate global analysis of ∣Vusf+K0π−(0)∣\lvert V_{us}f_{+}^{K^{0}\pi^{-}}(0)\rvert from KeK e and KμK\mu channels show no significant difference, which supports the SM prediction of lepton flavor universality. Third, the further reduction of theory uncertainties provides an even stronger motivation for next-generation experiments to measure the kaon lifetimes and Kℓ3K_{\ell3} branching ratios with higher precision. From the theory side, the dominant sources of uncertainty are now the phase space factor IKI_K and the isospin-breaking correction δSU(2)\delta_{\mathrm{SU}(2)}, which await new breakthroughs from phenomenology and lattice QCD.

Finally, we stress that every important theory input should always be cross-checked with several independent methods. In this sense, the existing plan to compute the full (virtual and real) Kℓ3K_{\ell3} RC directly on the lattice [43] is very much anticipated.

Acknowledgements

This work is supported in part by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) and the NSFC through the funds provided to the Sino-German Collaborative Research Center TRR110 “Symmetries and the Emergence of Structure in QCD” (DFG Project-ID 196253076 - TRR 110, NSFC Grant No. 12070131001) (U-G.M and C.Y.S), by the Chinese Academy of Sciences (CAS) through a President’s International Fellowship Initiative (PIFI) (Grant No. 2018DM0034) and by the VolkswagenStiftung (Grant No. 93562) (U-G.M), by EU Horizon 2020 research and innovation programme, STRONG-2020 project under grant agreement No 824093, by the German-Mexican research collaboration Grant No. 278017 (CONACyT) and No. SP 778/4-1 (DFG), and partially by the DFG personal grant No GO 2604/3-1 (M.G).

References

References

  1. [1]A. Sirlin, Current Algebra Formulation of Radiative Corrections in Gauge Theories and the Universality of the Weak Interactions, Rev. Mod. Phys. 50 (1978) 573. [Erratum: Rev. Mod. Phys.50,905(1978)].DOI
  2. [2]C.-Y. Seng, Radiative Corrections to Semileptonic Beta Decays: Progress and Challenges, Particles 4 (2021), no. 4 397–467 [2108.03279].DOI
  3. [3]C.-Y. Seng, M. Gorchtein, H. H. Patel and M. J. Ramsey–Musolf, Reduced Hadronic Uncertainty in the Determination of V_ud, Phys. Rev. Lett. 121 (2018), no. 24 241804 [1807.10197].DOI
  4. [4]C. Y. Seng, M. Gorchtein and M. J. Ramsey–Musolf, Dispersive evaluation of the inner radiative correction in neutron and nuclear β decay, Phys. Rev. D100 (2019), no. 1 013001 [1812.03352].DOI
  5. [5]C.-Y. Seng, X. Feng, M. Gorchtein and L.-C. Jin, Joint lattice QCD–dispersion theory analysis confirms the quark-mixing top-row unitarity deficit, Phys. Rev. D 101 (2020), no. 11 111301 [2003.11264].
  6. [6]K. Shiels, P. G. Blunden and W. Melnitchouk, Electroweak axial structure functions and improved extraction of the Vud CKM matrix element, Phys. Rev. D 104 (2021), no. 3 033003 [2012. 01580].DOI
  7. [7]M. Gorchtein and C.-Y. Seng, Dispersion relation analysis of the radiative corrections to gA in the neutron β-decay, JHEP 10 (2021) 053 [2106.09185].DOI
  8. [8]M. Gorchtein, γW Box Inside Out: Nuclear Polarizabilities Distort the Beta Decay Spectrum, Phys. Rev. Lett. 123 (2019), no. 4 042503 [1812.04229].DOI
  9. [9]Particle Data Group Collaboration, P. Zyla et. al., Review of Particle Physics, PTEP 2020 (2020), no. 8 083C01.
  10. [10]Muon g-2 Collaboration, B. Abi et. al., Measurement of the Positive Muon Anomalous Magnetic Moment to 0.46 ppm, Phys. Rev. Lett. 126 (2021) 141801.
  11. [11]T. Aoyama et. al., The anomalous magnetic moment of the muon in the Standard Model, Phys. Rept. 887 (2020) 1–166 [2006.04822].
  12. [12]J. P. Miller, E. de Rafael and B. L. Roberts, Muon (g-2): Experiment and theory, Rept. Prog. Phys. 70 (2007) 795 [hep-ph/0703049].DOI
  13. [13]J. P. Miller, E. de Rafael, B. L. Roberts and D. Stöckinger, Muon (g-2): Experiment and Theory, Ann. Rev. Nucl. Part. Sci. 62 (2012) 237–264.
  14. [14]F. Jegerlehner and A. Nyffeler, The Muon g-2, Phys. Rept. 477 (2009) 1–110 [0902.3360].DOI
  15. [15]LHCb Collaboration, R. Aaij et. al., Search for lepton-universality violation in B+ → K+ℓ+ℓ− decays, Phys. Rev. Lett. 122 (2019), no. 19 191801 [1903.09252].
  16. [16]LHCb Collaboration, R. Aaij et. al., Test of lepton universality using B+ → K+ℓ+ℓ− decays, Phys. Rev. Lett. 113 (2014) 151601 [1406.6482].pubmed.ncbi.nlm.nih.gov/25375701
  17. [17]LHCb Collaboration, R. Aaij et. al., Measurement of the ratio of branching fractions B(B̄0 → D*+τ−ν̄τ)/B(B̄0 → D*+μ−ν̄μ), Phys. Rev. Lett. 115 (2015), no. 11 111803 [1506.08614]. [Erratum: Phys.Rev.Lett. 115, 159901 (2015)].
  18. [18]LHCb Collaboration, R. Aaij et. al., Angular analysis of the B0 → K*0μ+μ− decay using 3 fb−1 of integrated luminosity, JHEP 02 (2016) 104 [1512.04442].
  19. [19]C.-Y. Seng, D. Galviz and U.-G. Meißner, A New Theory Framework for the Electroweak Radiative Corrections in Kℓ3 Decays, JHEP 02 (2020) 069 [1910.13208].DOI
  20. [20]C.-Y. Seng, X. Feng, M. Gorchtein, L.-C. Jin and U.-G. Meißner, New method for calculating electromagnetic effects in semileptonic beta-decays of mesons, JHEP 10 (2020) 179 [2009.00459].DOI
  21. [21]P.-X. Ma, X. Feng, M. Gorchtein, L.-C. Jin and C.-Y. Seng, Lattice QCD calculation of the electroweak box diagrams for the kaon semileptonic decays, Phys. Rev. D 103 (2021) 114503 [2102.12048].DOI
  22. [22]V. Cirigliano, M. Giannotti and H. Neufeld, Electromagnetic effects in K(l3) decays, JHEP 11 (2008) 006 [0807.4507].
  23. [23]C.-Y. Seng, D. Galviz, M. Gorchtein and U. G. Meißner, High-precision determination of the Ke3 radiative corrections, Phys. Lett. B 820 (2021) 136522 [2103.00975].DOI
  24. [24]C.-Y. Seng, D. Galviz, M. Gorchtein and U.-G. Meißner, Improved Ke3 radiative corrections sharpen the Kμ2−Kℓ3 discrepancy, JHEP 11 (2021) 172 [2103.04843].DOI
  25. [25]C.-Y. Seng, D. Galviz, W. J. Marciano and U.-G. Meißner, Update on |Vus| and |Vus/Vud| from semileptonic kaon and pion decays, Phys. Rev. D 105 (2022), no. 1 013005 [2107.14708].DOI
  26. [26]W. J. Marciano and A. Sirlin, Radiative corrections to pi(lepton 2) decays, Phys. Rev. Lett. 71 (1993) 3629–3632.
  27. [27]NA7 Collaboration, S. Amendolia et. al., A Measurement of the Space - Like Pion Electromagnetic Form-Factor, Nucl. Phys. B 277 (1986) 168.
  28. [28]S. Amendolia et. al., A Measurement of the Kaon Charge Radius, Phys. Lett. B 178 (1986) 435–440.
  29. [29]G. Ecker, J. Gasser, A. Pich and E. de Rafael, The Role of Resonances in Chiral Perturbation Theory, Nucl. Phys. B 321 (1989) 311–342.DOI
  30. [30]G. Ecker, J. Gasser, H. Leutwyler, A. Pich and E. de Rafael, Chiral Lagrangians for Massive Spin 1 Fields, Phys. Lett. B 223 (1989) 425–432.DOI
  31. [31]V. Cirigliano, G. Ecker, M. Eidemuller, R. Kaiser, A. Pich and J. Portoles, Towards a consistent estimate of the chiral low-energy constants, Nucl. Phys. B 753 (2006) 139–177 [hep-ph/0603205].DOI
  32. [32]V. Cirigliano, M. Knecht, H. Neufeld, H. Rupert­sberger and P. Talavera, Radiative corrections to K(l3) decays, Eur. Phys. J. C23 (2002) 121–133 [hep-ph/0110153].DOI
  33. [33]X. Feng, M. Gorchtein, L.-C. Jin, P.-X. Ma and C.-Y. Seng, First-principles calculation of electroweak box diagrams from lattice QCD, Phys. Rev. Lett. 124 (2020), no. 19 192002 [2003.09798].
  34. [34]S. Descotes-Genon and B. Moussallam, Radiative corrections in weak semi-leptonic processes at low energy: A Two-step matching determination, Eur. Phys. J. C42 (2005) 403–417 [hep-ph/0505077].DOI
  35. [35]B. Ananthanarayan and B. Moussallam, Four-point correlator constraints on electromagnetic chiral parameters and resonance effective Lagrangians, JHEP 06 (2004) 047 [hep-ph/0405206].DOI
  36. [36]J. Bijnens and G. Ecker, Mesonic low-energy constants, Ann. Rev. Nucl. Part. Sci. 64 (2014) 149–174 [1405.6488].DOI
  37. [37]Y. Aoki et. al., FLAG Review 2021, 2111.09849.DOI
  38. [42]W. J. Marciano, Precise determination of −V(us)− from lattice calculations of pseudoscalar decay constants, Phys. Rev. Lett. 93 (2004) 231803 [hep-ph/0402299].DOI
  39. [43]P. Boyle et. al. High-precision determination of V_us and V_ud from lattice QCD. [Link].

Paper details

vlnc:W_c7e74bq5
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 9

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents