A Surface Integrand for the Inverse KLT Kernel

1. Problem Statement

The paper addresses the complexity of calculating string-theory scattering amplitudes, specifically focusing on the inverse Kawai-Lewellen-Tye (KLT) kernel, denoted as $m_n^{\alpha'} = (S_n^{\alpha'})^{-1}$.

• Context: The KLT kernel relates closed string amplitudes to products of open string amplitudes. Its inverse, $m_n^{\alpha'}$, acts as a "stringy" generalization of the bi-adjoint scalar ($\text{tr}\phi^3$) theory amplitudes.
• The Challenge: In conventional Feynman diagram formulations, the inverse KLT kernel requires an infinite tower of interaction vertices (odd-point contact terms of valency $v \geq 5$) to reproduce the correct stringy behavior (monodromy relations, infinite resonance structure). This makes calculations at higher multiplicities and loop orders computationally prohibitive and obscures the underlying simplicity of the theory.
• Goal: The authors aim to find a more efficient, constructive definition of the inverse KLT kernel that generalizes to all loop orders, revealing a hidden structural simplicity analogous to the cubic $\text{tr}\phi^3$ theory.

2. Methodology

The authors employ a novel recursive approach based on Berends-Giele (BG) currents, adapted for stringy kinematics and loop integrands.

A. Cubic Berends-Giele Recursion (Tree-Level)

• Reformulation: Instead of summing over an infinite tower of Feynman diagrams with higher-valency vertices, the authors propose a purely cubic recursion.
• Effective Vertex: They introduce an $\alpha'$-dependent effective root vertex that absorbs the infinite tower of contact interactions. The vertex is defined as:
  $$V_{\text{eff}} = 1 + \tan(\pi\alpha' X_L)\tan(\pi\pi\alpha' X_R)$$
  where $X_L$ and $X_R$ are the kinematic invariants of the left and right sub-amplitudes.
• Mechanism: This effective vertex allows the inverse KLT matrix elements $m_n^{\alpha'}$ to be computed using only cubic diagrams (similar to $\text{tr}\phi^3$), but with a modified vertex that generates the necessary contact terms dynamically.

B. Kinematic Surface and Loop Generalization

• Surfaceology Framework: The authors utilize the "kinematic surface" formalism (introduced by Arkani-Hamed et al.), where external momenta and loop momenta are represented as punctures on a disk. Kinematic variables $X_{ij}$ correspond to curves connecting these punctures.
• Homology Sensitivity: A crucial insight is that the effective vertex's behavior depends on the homology of the curves on the kinematic surface.
  • If a region of the surface can be collapsed (e.g., an on-shell leg), the vertex reduces to the standard field theory vertex.
  • If the region contains a loop puncture (non-collapsible), the full $\alpha'$-dependent structure of the effective vertex remains active.
• Loop Recursion: The tree-level cubic recursion is generalized to $L$-loop integrands ($m_{L,n}^{\alpha'}$). The recursion sums over all ways to distribute $L$ loops between left and right sub-integrands, utilizing a symmetric product to ensure permutation symmetry of loop variables.

C. The $\alpha'$-Shift

• The paper utilizes a kinematic shift (the $\alpha'$-shift) to connect the inverse KLT integrand to Non-Linear Sigma Model (NLSM) pion amplitudes.
• By shifting even-parity kinematic variables ($X_{\text{even}} \to X_{\text{even}} \pm 1/\alpha'$) and rescaling $\alpha' \to \alpha'/2$, the inverse KLT integrand transforms into a "stringy" version of the NLSM integrand.

3. Key Contributions

1. Cubic Simplification of the Inverse KLT Kernel:
   The paper proves that the inverse KLT kernel, despite having an infinite tower of Feynman vertices, is structurally equivalent to a cubic theory when expressed via the new Berends-Giele recursion. This reduces the computational complexity from exponential (in the number of diagrams) to linear ($O(n)$ terms).

2. Definition of the Inverse KLT Integrand:
   The authors define the planar inverse KLT integrand $m_{L,n}^{\alpha'}$ for arbitrary loop order $L$ and multiplicity $n$. This is the first constructive definition of this object at the loop level.

3. Discovery of Loop-Level Contact Terms:
   A major finding is the existence of an "infinity of infinite towers" of contact terms at the integrand level.

   • While the integrated matrix elements match the field theory limit ($\alpha' \to 0$), the integrand contains additional scaleless contact terms (e.g., $\kappa_{L,n}$) that vanish upon integration but are essential for the integrand's consistency.
   • These terms are generated by the cubic recursion and are necessary to satisfy properties like the Adler zero for pions after the $\alpha'$-shift.

4. All-Loop Lagrangian:
   The authors derive a compact, all-loop Lagrangian (Eq. 4.54) that generates these contact terms. The derivative of this Lagrangian is surprisingly simple, involving elliptic integrals and Catalan numbers, despite the complexity of the individual terms.

5. Connection to NLSM Pions:
   The paper demonstrates that the inverse KLT integrand serves as a "perfect" stringy parent for NLSM pion integrands. The $\alpha'$-shift maps the cubic inverse KLT integrand directly to the known "surfacehedron" integrand for pions, unifying the scattering of cubic scalars and pions.

4. Results

• Recursion Formula: The $L$-loop integrand is constructed via:
  $$m_{L,n}^{\alpha'} = \sum_{\text{partitions}} \left( \frac{1 + t_L t_R}{t_L t_R} \right) \times (\text{Symmetric Product of Sub-integrands})$$
  where the sum runs over partitions of external legs and loop momenta.
• Contact Term Structure: The contact terms $\kappa_{L,n}$ vanish if $L+n$ is even and are non-zero if $L+n$ is odd. They follow a pattern involving Catalan numbers and polynomials in $n$:
  $$\kappa_{L,n} \propto P_L(n) C_{\lfloor (n-1)/2 \rfloor}$$
• Computational Efficiency: The recursion allows for the calculation of high-multiplicity, high-loop integrands (e.g., 4-loop, 8-point) in seconds on a standard desktop, a task previously intractable with Feynman diagrams.
• Verification: The authors verified the consistency of the integrand by checking:
  • Factorization: Correct pole structure on single cuts.
  • Adler Zero: The resulting pion integrands vanish in the soft limit.
  • Field Theory Limit: The $\alpha' \to 0$ limit correctly reproduces the $\text{tr}\phi^3$ surface integrand.

5. Significance

• Toy Model for String Amplitudes: The inverse KLT integrand is proposed as the simplest toy model for stringy amplitudes. It retains key string features (infinite resonances, monodromy) while being computationally as simple as a cubic field theory.
• Unification: It provides a unified framework for understanding the relationship between bi-adjoint scalars, NLSM pions, and string theory, mediated by the kinematic surface and $\alpha'$-shifts.
• Geometric Insight: The work deepens the connection between scattering amplitudes and positive geometry (associahedra, halohedra). The integrand is naturally defined on the kinematic surface, suggesting a geometric origin for the "infinity of contact terms."
• Future Applications: The framework opens the door to studying the double copy of string loop integrands directly in kinematic space and exploring positive geometries for higher-loop moduli spaces, which are known to be non-polytopal.

In summary, the paper revolutionizes the calculation of inverse KLT kernels by replacing an intractable infinite tower of Feynman diagrams with a simple, cubic, surface-based recursion, revealing a profound structural simplicity in string theory amplitudes.