--- title: All-Line Transverse Momentum Shift url: https://www.emergentmind.com/topics/all-line-transverse-momentum-shift type: topic --- # All-Line Transverse Momentum Shift The **All-Line Transverse (ALT) momentum shift** is a complex deformation used in tree-level on-shell recursion for amplitudes with arbitrary masses, especially in theories with particles of spin $\le 1$. In the 2024 formulation, every external momentum is shifted simultaneously by a null vector proportional to a transverse polarization vector, with the deformation chosen so that external legs remain on shell and external wavefunctions or polarization vectors remain unchanged [2403.15538]. The construction was developed for massive amplitudes, where standard BCFW-type shifts are often restrictive and where contact-term ambiguities can obstruct purely factorization-based reconstructions. In subsequent work on electroweak amplitudes, the same framework was used to connect constructibility under the ALT shift to Ward identities and to show that quartic gauge-boson contact terms arise automatically in recursive constructions [2407.14587]. ## 1. Definition and kinematic realization The ALT shift starts from the standard deformation \[ \hat p_i(z)=p_i+z\,q_i, \] subject to \[ \sum_i \hat p_i(z)=0, \qquad \hat p_i^2(z)=p_i^2=m_i^2. \] The distinctive feature is that, for massive particles, the deformation vector is chosen proportional to a **transverse polarization vector** of each external leg [2403.15538]. In the little-group-covariant massive spinor-helicity formalism, a massive momentum is written as \[ p_{a\dot a}=\lambda_a^I \tilde\lambda_{\dot a I}, \] and in the helicity basis the paper uses \[ |i\rangle^I_a = |i\rangle_a \delta^I_- + |\eta_i\rangle_a \delta^I_+, \qquad [i|^I_{\dot a} = [i|_{\dot a}\delta^I_+ + [\eta_i|_{\dot a}\delta^I_-. \] The momentum is then \[ (p_i)_{a\dot a}=|i\rangle_a[i|_{\dot a}-|\eta_i\rangle_a[\eta_i|_{\dot a}, \] with on-shell condition \[ \langle i\,\eta_i\rangle=[i\,\eta_i]=m_i. \] The transverse and longitudinal polarization vectors of a massive spin-1 particle are \[ \epsilon_i^{(+)}=\sqrt{2}\frac{|\eta_i\rangle[i|}{m_i}, \qquad \epsilon_i^{(-)}=-\sqrt{2}\frac{|i\rangle[\eta_i|}{m_i}, \qquad \epsilon_i^{(L)}=\frac{|i\rangle[i|+|\eta_i\rangle[\eta_i|}{m_i}. \] The ALT deformation for spin-$\frac12$ particles and transverse spin-1 modes is defined by \[ \begin{cases} |i\rangle \to |\hat i\rangle = |i\rangle + z\,c_i\,|\eta_i\rangle & \text{for } I_i=+,\\[1mm] |i] \to |\hat i] = |i] + z\,c_i\,|\eta_i] & \text{for } I_i=-, \end{cases} \] which induces \[ p_i \to \hat p_i = p_i + z\,\frac{c_i m_i}{\sqrt2}\,\epsilon_i^{(I_i)}. \] Thus each momentum is shifted **along the transverse polarization vector** of the corresponding external state [2403.15538]. For longitudinal massive spin-1 modes, the construction uses paired spinor deformations such as \[ \begin{cases} |i\rangle \to |\hat i\rangle = |i\rangle + z\frac{c_i}{2}|\eta_i\rangle,\\[1mm] [\eta_i| \to [\hat\eta_i|=[\eta_i|-z\frac{c_i}{2}[i| \end{cases} \qquad \text{or} \qquad \begin{cases} [i| \to [\hat i|=[i|+z\frac{c_i}{2}[\eta_i|,\\[1mm] |\eta_i\rangle \to |\hat\eta_i\rangle = |\eta_i\rangle - z\frac{c_i}{2}|i\rangle. \end{cases} \] The purpose of these paired shifts is to keep external wavefunctions and polarization vectors $z$-independent, at least up to spin 1 [2407.14587]. A recurring point in the literature is the meaning of the word **transverse**. In the ALT framework it refers specifically to the physical transverse polarization vectors $\epsilon^{(\pm)}$, not to the longitudinal vector $\epsilon^{(L)}$. The longitudinal polarization is not used directly as a shift direction because its norm is nonzero and then $\hat p_i^2\neq p_i^2$ would spoil on-shellness [2403.15538]. ## 2. On-shell constraints, momentum conservation, and invariance of external states The ALT shift is designed to preserve the basic kinematic constraints of recursion. On-shellness follows because \[ \langle \hat i\,\eta_i\rangle = \langle i\,\eta_i\rangle + z c_i \langle \eta_i\,\eta_i\rangle = m_i, \] and similarly \[ [\hat i\,\eta_i]=[i\,\eta_i]+z c_i[\eta_i\,\eta_i]=m_i, \] using \[ \langle \eta_i\,\eta_i\rangle = [\eta_i\,\eta_i]=0. \] Momentum conservation imposes \[ 0=\sum_i c_i\,\epsilon_i^{(I_i)}. \] The construction emphasizes that these transverse deformation vectors are purely spatial, with \[ \langle i|\gamma^0|\eta_i]=\langle \eta_i|\gamma^0|i]=0, \] so there are effectively only three independent constraints and nontrivial solutions for the coefficients $c_i$ can exist already at four points [2403.15538]. A central practical advantage is that the shift can be chosen so that **external Dirac spinors and polarization vectors are unshifted**. In the electroweak formulation, the massive Dirac spinors are written as \[ u^{(+)}=\begin{pmatrix}|\eta_i\rangle_a\\ |i]^{\dot a}\end{pmatrix}, \qquad u^{(-)}=\begin{pmatrix}|i\rangle_a\\ |\eta_i]^{\dot a}\end{pmatrix}, \qquad v^{(+)}=\begin{pmatrix}|\eta_i\rangle_a\\ -|i]^{\dot a}\end{pmatrix}, \qquad v^{(-)}=\begin{pmatrix}|i\rangle_a\\ -|\eta_i]^{\dot a}\end{pmatrix}, \] and the shift is arranged so that the combinations entering $u$, $v$, and $\epsilon$ remain undeformed [2407.14587]. This feature is one of the main reasons the ALT shift is well suited to amplitudes with massive fermions and massive vector bosons. The framework is not manifestly little-group covariant in its definition. The full amplitudes remain little-group covariant, but a **helicity basis** must be chosen to define a good complex deformation. This suggests, as the papers themselves emphasize, that the method resembles massless BCFW practice: the recursion respects the physical symmetry, but the shift is implemented in a basis adapted to the external spin configuration [2403.15538]. ## 3. Large-$z$ behavior, constructibility, and the role of Ward identities The validity of any on-shell recursion depends on the behavior of the shifted amplitude at large complex parameter $z$. For ALT, the starting point is a dimensional estimate. A generic $n$-point amplitude is written schematically as \[ A_n=\left(\sum_{\mathrm{diagrams}} g\times F\right)\times \prod_{\mathrm{vectors}} \epsilon \times \prod_{\mathrm{fermions}} u, \] where $F=N/D$ contains the stripped kinematic dependence. Because external wavefunctions and polarization vectors are $z$-independent under ALT, the large-$z$ behavior is controlled entirely by $F$ [2403.15538]. For an all-line shift and generic kinematics, each internal propagator denominator scales as $z^2$, since the internal shifted momentum is a sum of several shifted external momenta and generally satisfies $q_I^2\neq 0$. The numerator obeys \[ N\sim z^{\gamma_N}, \qquad \gamma_N\le [N], \] leading to the bound \[ \hat A_n \sim z^\gamma, \qquad \gamma=\gamma_N-[D]\le [N]-[D] =4-n-[g]-\frac{N_F}{2}. \] For QED, where $[g]=0$, this becomes \[ \gamma \le 4-n-\frac{N_F}{2}. \] The papers conclude from this that in renormalizable theories with spin $\le 1$, all amplitudes with $n\ge 5$ are constructible under ALT, and that for $n=4$ amplitudes are constructible if they involve at least one fermion [2403.15538]. The electroweak analysis sharpens this statement by relating ALT directly to **Ward identities**. At large $z$, since the shift vector is proportional to a transverse polarization vector, \[ \lim_{z\to\infty}\hat p_i \propto \epsilon_i^{(\lambda_i)}. \] For a transverse external gauge boson, the leading term may therefore be written schematically as proportional to \[ q_1\cdot F(q_1,\dots,q_n). \] If the theory obeys the massless Ward identity \[ q_1\cdot F(q_1,\dots,q_n)=0, \] the leading large-$z$ term vanishes, improving the scaling to \[ \gamma \le 4-n-[g]-\frac{N_F}{2}-c_T, \] with \[ c_T= \begin{cases} 1, & \text{if at least one external transverse mode is present},\\ 0, & \text{if all external gauge bosons are longitudinal}. \end{cases} \] This is the main general constructibility criterion derived for ALT in the electroweak setting [2407.14587]. The resulting interpretation is precise. The ALT shift is not merely a convenient deformation; its large-$z$ validity is tightly linked to gauge invariance in the massless limit. The papers present this as the on-shell form of the statement that a consistent massive vector theory descends from a spontaneously broken gauge theory [2407.14587]. Two limitations are stated explicitly. First, all-longitudinal four-vector amplitudes are not guaranteed constructible by the basic ALT power-counting argument alone. Second, four-scalar amplitudes can contain an independent $\lambda \phi^4$ term and are not constructible by momentum shifts in general [2407.14587]. ## 4. Recursion relation, pole structure, and paired roots The contour argument is standard in form but acquires distinctive features under ALT because internal shifted momenta typically satisfy $q_I^2\neq 0$. For an internal channel $I$, \[ \hat p_I(z)=p_I+z q_I, \qquad p_I=\sum_{i\in L} p_i, \qquad q_I=\sum_{i\in L} q_i, \] and the pole condition \[ \hat p_I^2-m_I^2=0 \] has two roots, \[ z_I^\pm = \frac{1}{q_I^2}\left[ -p_I\cdot q_I \pm \sqrt{(p_I\cdot q_I)^2-(p_I^2-m_I^2)q_I^2} \right]. \] The propagator can then be rewritten as \[ \frac{1}{\hat p_I^2-m_I^2} = \frac{1}{p_I^2-m_I^2} \frac{z_I^+ z_I^-}{(z-z_I^+)(z-z_I^-)}. \] This paired-pole structure is one of the main differences from ordinary BCFW-type deformations, where often $q_I^2=0$ and there is a single root per channel [2403.15538]. At tree level, the general recursion formula is \[ A_n = -\sum_{z=z_I}\sum_\lambda \mathrm{Res}\left[ \hat A_{n-m+2}^{(\lambda)} \frac{1}{z}\frac{1}{\hat p_I^2-m_I^2} \hat A_m^{(-\lambda)} \right] +B_\infty. \] When the large-$z$ condition implies $B_\infty=0$, this yields a recursive construction from lower-point on-shell amplitudes. In the electroweak presentation the paired-root version is written as \[ A_n = \sum_I \frac{1}{p_I^2-m_I^2} \frac{1}{z_I^+ - z_I^-} \sum_\lambda \left[ z_I^+\hat A^{(\lambda)}_{n-m+2}(z_I^-)\, \hat A^{(\bar\lambda)}_m(z_I^-) - (z_I^+\leftrightarrow z_I^-) \right] + B_\infty \] [2407.14587]. A practical consequence is that intermediate expressions can contain square roots, but the papers stress that poles occur in pairs and final answers simplify strongly. In the five-point QED example this simplification is organized by identities of the form \[ f_j=\sum_{k=1}^I z_k^j \prod_{i\neq k}\frac{z_i}{z_k-z_i}, \] with \[ f_0=(-1)^{I+1},\qquad f_j=0 \quad (1\le j