Precision asymptotics of string amplitudes

Marco Maria Baccianti, Lorenz Eberhardt, Sebastian MizeraarXiv 2026 · DOI · arXiv

TL;DR

This paper resolves a long-standing discrepancy between the classic 1987 Gross–Mende analysis of high-energy string scattering and modern numerical data. By identifying an infinite family of previously overlooked complex "saddle points" and using a "bootstrap" method to determine their contributions, the authors provide a new, precise formula that accounts for the complex oscillatory patterns seen in one-loop string amplitudes.

What Problem Does This Paper Address?

In the late 1980s, David Gross and Paul Mende published a landmark analysis of string scattering at high energies and fixed angles. They argued that the string path integral is dominated by a specific "saddle point"—a geometry where the string effectively "folds" on itself, leading to a characteristic exponential suppression.

For decades, this was the textbook result. However, recent advancements in numerical techniques (specifically the Rademacher method) revealed a "tension": the actual numerical data for one-loop amplitudes didn't quite match the Gross–Mende (GM) prediction. While the overall exponential "envelope" was roughly correct, the data showed complex oscillations and a significant real part that the GM analysis—which predicted a purely imaginary result—could not explain.

The gap exists because the original GM analysis was performed in unphysical kinematics (imaginary scattering angles). When moving to physical, Lorentzian kinematics, the landscape of "saddles" (points where the action is stationary) becomes much more crowded and complex.

Methods

The authors combine classical semi-classical analysis with modern "bootstrap" philosophy:

  1. Saddle Hunting: They performed a massive numerical search, identifying approximately 2,000 saddle points in the complexified moduli space of the torus (the worldsheet for one-loop scattering).
  2. Classification via Modular Symmetry: They discovered that the dominant saddles form infinite families labeled by the congruence subgroup $\Gamma(2)$ of the modular group. They identified a subset of "elementary saddles" that appear to be the only ones contributing to the final amplitude.
  3. The Bootstrap Problem: Determining exactly how much each saddle contributes (its "multiplicity") is a notoriously difficult problem involving 8-dimensional contour deformations. Instead of solving this directly, the authors formulated a bootstrap: they wrote an ansatz for the multiplicities and constrained them using physical principles:
  4. Crossing Symmetry: The result must be invariant under the exchange of scattering channels.
  5. Reality: Saddles must pair up to ensure the final amplitude respects complex conjugation properties.
  6. Boundedness: Using the Phragmén–Lindelöf principle, they showed the amplitude must remain bounded in certain regions of the complex energy plane, which severely restricts the possible forms of the multiplicities.
  7. Numerical Matching: They used high-precision numerical data to fix the remaining free parameters in their bootstrap ansatz.

Datasets

The study relies on a comprehensive numerical dataset published on Zenodo [32]. This includes:

  • Imaginary Parts: Computed via unitarity cuts and the Baikov representation for energies up to $s=70$.
  • Real Parts: Computed using the recently developed Rademacher method, which allows for the evaluation of string integrals that were previously considered numerically unreachable due to massive cancellations.
  • Decay Widths: Analytical results for the coefficients of double poles at integer energy levels.

Key Insights and Ideas

  • The "Tail" of Saddles: The high-energy limit isn't dominated by one geometry, but by an infinite "tail" of complex saddles. These saddles interfere with each other, creating the oscillatory behavior seen in the data.
  • Analytic Continuation Matters: The discrepancy in the original GM analysis stems from the fact that the "steepest descent" contours change when you move from unphysical to physical scattering angles, picking up new complex contributions.
  • Emergent Non-Analyticities: The sum over the infinite tail of saddles produces new, logarithmic singularities in the high-energy limit at specific energy thresholds ($s \in 2\mathbb{Z}$), a feature not present in individual saddle contributions.
  • Simplicity of Multiplicities: Despite the complexity of the integral, the multiplicities of these saddles turn out to be simple integers or simple trigonometric functions (like $\cot^2(\pi s)$), suggesting a deep underlying structure.

Key Results

The primary result is a new asymptotic expansion for the one-loop four-graviton amplitude (Equation 4.38).

  • Quantitative Match: The new formula (shown in red in the paper's figures) perfectly tracks the numerical data (blue), capturing both the exponential decay and the high-frequency oscillations that the original GM prediction (black) missed entirely.
  • Resolution of the "Real Part" Puzzle: By including the full family of complex saddles, the authors explain why the real and imaginary parts of the amplitude are of the same order of magnitude.
  • The Master Formula: They provide a precise expression for the "tail sum," showing that the high-energy behavior is governed by the Lerch transcendent function.

Limitations and Future Work

  • The $\alpha$ Parameter: One constant in the multiplicity bootstrap (denoted as $\alpha$) could not be determined conclusively from the current numerical data, though the data suggests $\alpha=0$.
  • Higher Genus: The authors speculate that for $g$-loop amplitudes, the dominant saddles are $(g+1)$-fold covers of the tree-level saddle. However, verifying this for $g > 1$ remains a computational challenge.
  • The Chaos Question: While the one-loop result shows regular oscillations (not chaos), the authors suggest that "chaos" might emerge only after summing over all loop orders (all genera), which would be necessary to describe black hole physics.
  • Non-Perturbative Breakdown: The paper notes that the perturbative expansion itself eventually breaks down at extremely high energies, where the string coupling effectively becomes strong regardless of its initial value.

Summary updated

Abstract

Recent work revealed a tension between the Gross-Mende analysis of the high-energy fixed-angle behavior of string amplitudes and the explicit numerical data. Motivated by this puzzle, we revisit the problem of classifying saddle-point geometries for the one-loop amplitude. We find an infinite family of complex saddles that dominate the high-energy regime. Using general constraints and matching to numerical data, we formulate a bootstrap problem that determines their multiplicities. This procedure yields a precise asymptotic expansion of the one-loop amplitude at high energies. The resulting oscillatory contributions lead to a much richer high-energy behavior than that predicted by the original Gross-Mende analysis.
BibTeX
@misc{baccianti2026precision,
  author = {Marco Maria Baccianti and Lorenz Eberhardt and Sebastian Mizera},
  title = {{Precision asymptotics of string amplitudes}},
  year = {2026},
  eprint = {2601.09707},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2601.09707},
}

Loading related content...

Figure