arXiv Science⌕ Search

arXiv subjects

Matthew D. Schwartz

Publications and source records attributed to Matthew D. Schwartz.

At least 19 recordsLinked to original sources
arXiv2026

Elliptic master integrals and the one-loop cosmological wavefunction

Loop corrections to the cosmological wavefunction involve similar integrals to those required for scattering amplitudes. For the one-loop triangle of a conformally coupled scalar, the master integrals are elliptic, and so the triangle wavefunction coefficient was expected to be elliptic as well. Using a bootstrap approach, we compute this triangle coefficient exactly and show that it is not: the elliptic periods cancel, and the flat-space wavefunction coefficient is a combination of dilogarithms. The cancellation occurs because the residue of the loop integrand on the elliptic sector is a flat-space amplitude, rational in the loop momentum. The one-loop box, by contrast, is not polylogarithmic, even with a single external leg at each site. Its threshold discontinuity involves elliptic integrals, which we also give in closed form. For regular-tetrahedron momenta the periods of a K3 surface enter its loop integrand.

hep-th↗
arXiv2026

Landau singularity of the K3 maximal cut in black-hole scattering at fifth post-Minkowskian order

Modeling the gravitational waves from merging black holes requires precise calculations of the two-body problem. The post-Minkowskian expansion of black-hole scattering, in powers of Newton's constant, now stands at fifth order, four loops. At this order one part of the deflection diverges when the black holes pass at about 94 percent of the speed of light. The divergence is known to come from periods of a K3 surface, which has a node at this speed. We show that the node lies on the integration contour of the K3 maximal cut, which makes this speed a Landau singularity of the cut. Requiring the divergence to cancel, without introducing new infinities, has been shown to fix the memory contribution, named for the permanent displacement that the waves leave in test masses, up to one constant. We derive the leading coefficients of its boundary integrals in closed form. We obtain both results from maximal cuts and boundary limits, without an integration-by-parts reduction of the full four-loop integrand. We also continue the K3 and Calabi--Yau periods of these integrals to the energies of bound orbits, as in the inspiral before a merger.

hep-th↗
arXiv2026

The hidden sunrise in the energy-energy correlator

The energy-energy correlator (EEC) is one of a handful of collider observables that can be computed analytically to high orders. As an energy-weighted cross-section, it exposes features of quantum field theory that scattering amplitudes do not, and it can be compared directly to data. As with scattering amplitudes, the analytic expressions require special functions beyond polylogarithms. In Henn et al., the EEC was computed to next-to-next-to-leading order (NNLO) for $\mathcal{N}=4$ super Yang-Mills (SYM) theory, and the result is expressed in terms of both harmonic polylogarithms (HPLs) and one two-fold integral, which contains elliptic curves. In this work, we report a complete result including the remaining elliptic sector, which is closely related to the sunrise Feynman integrals. We find the $j$-invariants of the EEC and the sunrise agree identically under a Möbius map, and thus the elliptic sector in the EEC lives on the $Γ_1(6)$ modular curve of the sunrise integral. We then express the answer in terms of iterated Eisenstein integrals, which can be evaluated to high precision quickly. The analytic form also allows a first study of the EEC Landau bootstrap, and the understanding of its function space in $\mathcal{N}=4$ SYM offers a concrete handle on the elliptic sector of the EEC in QCD. Many of the technical results in this paper were completed with AI under human supervision.

hep-th↗
arXiv2026

BootLoops: an LLM-driven toolkit for exact quantitative science

Much of quantitative science involves difficult mathematical computations: a Feynman integral, evidence for an evolutionary tree, or a regulatory statistic. These are often estimated by Monte Carlo or heuristic search, or computed in floating point without tracking rounding error. Exact methods often exist, but in a field far from the problem. The reduction methods of collider physics also work on a Bayesian evidence, and a regulatory threshold can be settled with interval arithmetic from numerical analysis, but few people know both fields. A large language model knows both, and every other field, and with working programs in hand it can carry a method from one to another. BootLoops is a pool of such programs: integral reduction, differential equations and high-precision evaluation from particle physics, integer-relation fitting from experimental mathematics, exact enumeration, and ball arithmetic, in one toolkit that any agentic large language model can operate and extend. These tools can compute multi-loop Feynman integrals by bootstrap methods rather than direct integration, across the function classes that arise, including polylogarithms, elliptic functions, and periods of K3 surfaces and Calabi-Yau manifolds. They can also evaluate Bayesian evidence integrals as exact rational numbers, enumerate finite configuration spaces with proof of completeness, show when a sought closed form cannot exist, and redo floating-point calculations with guaranteed digits. In practice this brings the tools of mathematics, computer science and physics to problems in genomics, statistics, ecology and other fields. The toolkit and its documentation are at https://bootloops.ai.

hep-ph↗
arXiv2026

Renormalon Saddles in the OPE

The cancellation of renormalon ambiguities in the operator product expansion is usually formulated as a matching between perturbative Borel ambiguities and prescription dependence in non-perturbative condensates. We give this cancellation a contour realization in the two-dimensional Gross--Neveu model at large $N$. The exact order-$1/N$ fermion self-energy in the massive vacuum is a finite real integral; ambiguities arise only after its large-momentum behavior is decomposed into products of perturbative Wilson coefficients and the matrix elements of the associated operators. A reduced integral isolates the leading renormalon and reveals a two-dimensional contour geometry in which the perturbative series is generated at the boundary corner where the exchanged momentum reaches the external hard scale, while its Stokes discontinuity is controlled by a distinct boundary critical point at zero exchanged momentum. After renormalizing the ultraviolet divergences of the condensate scale integral, we derive its compensating lateral ambiguity without fixing it by matching to the perturbative sector, distinguish the renormalon saddle from the apparent IR Landau pole, and obtain the reduced contour variables from the large-$N$ auxiliary-field path integral. The renormalon saddle is a boundary critical point of this reduced mode-space integral. The perturbative Wilson coefficient and the renormalized condensate thus define complementary relative cycles of the same reduced integrand whose complex tails cancel, reconstructing the original real integration cycle.

hep-th↗
arXiv2026

The Double Well Done Doubly-Well

The symmetric double-well potential is one of the simplest quantum-mechanical systems in which perturbative and non-perturbative physics are deeply entangled. Its energy levels have non-analytic expansions in inverse powers of the inter-well separation, with factorially growing coefficients, while the parity splitting is exponentially small and invisible to perturbation theory. Resurgence ties the two features together, organizing the exact spectrum into a single tightly-constrained trans-series. This paper gives a self-contained account of this trans-series from two complementary approaches: exact WKB and the Euclidean path integral, developed in a common notation with explicit calculations through the four-instanton level and three-loop order. In exact WKB, Stokes phenomena encoded in the Delabaere--Dillinger--Pham relations control the analytic continuation of the wavefunction past turning points. The quantization condition expressed in terms of Voros symbols then determines the full trans-series. The DDP relations are local and do not require knowing the global topology of the energy surface, but that surface is an elliptic curve. In the path integral, elliptic curves enter differently: the classical saddle points are doubly-periodic elliptic functions of Euclidean time, and Stokes phenomena play out within the finite-dimensional manifold of quasi-zero modes rather than through analytic continuation of the wavefunction. A Lefschetz thimble decomposition determines which saddles contribute, and the resulting partition function trans-series is much simpler than the energy trans-series: at each instanton order the $T$-dependence is a polynomial fixed by the quasi-zero-mode thimble integrals. Together, the two approaches deploy a shared mathematical infrastructure in complementary ways, showing that the double well is an ideal setting to explore resurgence.

hep-th↗
arXiv2026

A Precise Determination of $α_s$ from the Heavy Jet Mass Distribution

A global fit for $α_s(m_Z)$ is performed on available $e^+e^-$ data for the heavy jet mass distribution. The state-of-the-art theory prediction includes $\mathcal{O}(α_s^3)$ fixed-order results, N$^3$LL$^\prime$ dijet resummation, N$^2$LL Sudakov shoulder resummation, and a first-principles treatment of power corrections in the dijet region. Theoretical correlations are incorporated through a flat random-scan covariance matrix. The global fit results in $0.1148^{+ 0.0015}_{-0.0022}$, compatible with similar determinations from thrust and $C$-parameter. Dijet resummation is essential for a robust fit, as it engenders insensitivity to the fit-range lower cutoff; without resummation the fit-range sensitivity is overwhelming. In addition, we find evidence for a negative power correction in the trijet region if and only if Sudakov shoulder resummation is included.

hep-ph↗
arXiv2026

Renormalons as Saddle Points

Instantons and renormalons play important roles at the interface between perturbative and non-perturbative quantum field theory. They are both associated with branch points in the Borel transform of asymptotic series, and as such can be detected in perturbation theory. However, while instantons are associated with non-perturbative saddle points of the path integral, renormalons have mostly been understood in terms of Feynman diagrams and operator product expansions. We suggest a non-perturbative path integral explanation of how both instantons and renormalons produce singularities in the Borel plane using representative finite-dimensional integrals. In particular, we build evidence that renormalons can be understood as saddle points of the 1-loop effective action, enabled by a crucial contribution from the quantum scale anomaly. These results are illustrated in simple toy models and indicate a possible route toward studying renormalons within realistic asymptotically-free field theories.

hep-th↗
arXiv2026

Beyond the Dilute Instanton Gas: Resurgence with Exact Saddles in the Double Well

The path-integral approach to the double well has long been limited by the dilute instanton gas approximation. We show that if the finite Euclidean-time structure is taken seriously by using exact saddles, the dilute gas can be sidestepped, allowing the partition function and energy levels to be computed systematically. At each instanton order, the full resurgent structure -- which saddles contribute, what asymptotic growth is expected and how ambiguities cancel -- is encoded in a finite-dimensional Picard--Lefschetz contour integral over the quasi-zero modes with a clear geometric interpretation. Working at finite $T$ is essential: the dilute instanton gas can only access the ground-state splitting, whereas the exact finite-$T$ computation systematically produces the non-perturbative energy splittings for all excited states, including their full dependence on the level number. The key ingredients -- Weierstrass elliptic functions for the saddles, Lamé operators for the fluctuations and Picard--Fuchs equations for the periods -- form a coherent mathematical framework that both overlaps and complements that of Exact WKB.

hep-th↗
arXiv2026

Observable Optimization for Precision Theory: Machine Learning Energy Correlators

The practice of collider physics typically involves the marginalization of multi-dimensional collider data to uni-dimensional observables relevant for some physics task. In any cases, such as classification or anomaly detection, the observable can be arbitrarily complicated, such as the output of a neural network. However, for precision measurements, the observable must correspond to something computable systematically beyond the level of current simulation tools. In this work, we demonstrate that precision-theory-compatible observable space exploration can be systematized by using neural simulation-based inference techniques from machine learning. We illustrate this approach by exploring the space of marginalizations of the energy 3-point correlator to optimize sensitivity to the the top quark mass. We first learn the energy-weighted probability density from simulation, then search in the space of marginalizations for an optimal triangle shape. Although simulations and machine learning are used in the process of observable optimization, the output is an observable definition which can be then computed to high precision and compared directly to data without any memory of the computations which produced it. We find that the optimal marginalization is isosceles triangles on the sphere with a side ratio approximately $1:1:\sqrt{2}$ (i.e. right triangles) within the set of marginalizations we consider.

hep-ph↗
arXiv2026

Resummation of the C-Parameter Sudakov Shoulder Using Effective Field Theory

The C-parameter distribution in $e^+e^-$ annihilation exhibits a kinematic shoulder at $C = 3/4$, where three-parton final states reach their maximum and a fourth parton is required to exceed it. This boundary generates large logarithms that must be resummed. Using soft-collinear effective theory, we derive a factorization theorem involving new jet and soft functions specific to the C-parameter measurement, in which soft radiation contributes quadratically in transverse momentum. This quadratic structure explains the step discontinuity at leading order. We compute all ingredients at one loop, validate against Monte Carlo, and present matched NLL+NLO results. Unlike thrust and heavy jet mass, the C-parameter has no Sudakov--Landau pole, making momentum-space resummation straightforward. All calculations, numerical analysis, and manuscript preparation were performed by Claude, an AI assistant developed by Anthropic, working under physicist supervision.

hep-ph↗
arXiv2025

Asymptotic Behavior of Diagram Classes

The asymptotic nature of perturbative expansions in quantum field theory can arise from the factorial growth in the number of Feynman diagrams with loop order, as with instantons, or from a series of individual diagrams whose values grow factorially, as with renormalon chains in QED. Other classes of diagrams are known also to grow factorially, such as the Hopf series of graphs in $ϕ^3$ theory. This Hopf series was studied using Schwinger-Dyson equations and the Connes-Kreimer Hopf algebra of decorated rooted trees. We review the Hopf algebra approach and show that the same results can be obtained using analytic QFT techniques as with Hopf-algebraic ones. We present an efficient method to extract the asymptotic behavior and thereby generalize the analysis of the Hopf series to other classes of diagrams in other theories. We confront the question of whether these classes correspond to new types of asymptotic growth beyond instantons and renormalons, and find that they appear to be incomplete calculations of what would be renormalons in these theories if all diagrams were included. Although the Hopf algebra approach is not essential to deriving asymptotic behavior from Schwinger-Dyson equations, it does provide some other insights into quantum field theory. We therefore attempt also to provide a map between some relevant aspects of the Hopf algebra and quantum field theory.

hep-th↗
arXiv2025

Precision $e^+e^-$ Hemisphere Masses in the Dijet Region with Power Corrections

We derive high-precision results for the $e^+e^-$ heavy jet mass (HJM) $d σ/d ρ$ and dihemisphere mass (DHM) $d^2σ/(d s_1 d s_2)$ distributions, for $s_1\sim s_2$, in the dijet region. New results include: i) the N$^3$LL resummation for HJM of large logarithms $\ln^n(ρ)$ at small $ρ$ including the exact two-loop non-global hemisphere soft function, the 4-loop cusp anomalous dimension and the 3-loop hard and jet functions, ii) N$^3$LL results for DHM with resummation of logarithms $\ln(s_{1,2}/Q^2)$ when there is no large separation between $s_1$ and $s_2$, iii) profile functions for HJM to give results simultaneously valid in the peak and tail regions, iv) a complete two-dimensional basis of non-perturbative functions which can be used for double differential observables, that are needed for both HJM and DHM in the peak region, and v) an implementation of renormalon subtractions for large-angle soft radiation to ${\cal O}(α_s^3)$ together with a resummation of the additional large $\ln(Qρ/Λ_{QCD})$ logarithms. Here $Q$ is the $e^+e^-$ center-of-mass energy. Our resummation results are combined with known fixed-order ${\cal O}(α_s^3)$ results and we discuss the convergence and remaining perturbative uncertainty in the cross section. We also prove that, at order $1/Q$, the first moment of the HJM distribution involves an additional non-perturbative parameter compared to the power correction that shifts the tail of the spectrum (where $1\gg ρ\gg Λ_{QCD}/Q$). This differs from thrust where a single non-perturbative parameter at order $1/Q$ describes both the first moment and the tail, and it disfavors models of power corrections employing a single non-perturbative parameter, such as the low-scale effective coupling model. In this paper we focus only on the dijet region, not the far-tail distribution for $ρ\gtrsim 0.2$.

hep-ph↗
arXiv2025

Analytic Regression of Feynman Integrals from High-Precision Numerical Sampling

In mathematics or theoretical physics one is often interested in obtaining an exact analytic description of some data which can be produced, in principle, to arbitrary accuracy. For example, one might like to know the exact analytical form of a definite integral. Such problems are not well-suited to numerical symbolic regression, since typical numerical methods lead only to approximations. However, if one has some sense of the function space in which the analytic result should lie, it is possible to deduce the exact answer by judiciously sampling the data at a sufficient number of points with sufficient precision. We demonstrate how this can be done for the computation of Feynman integrals. We show that by combining high-precision numerical integration with analytic knowledge of the function space one can often deduce the exact answer using lattice reduction. A number of examples are given as well as an exploration of the trade-offs between number of datapoints, number of functional predicates, precision of the data, and compute. This method provides a bottom-up approach that neatly complements the top-down Landau-bootstrap approach of trying to constrain the exact answer using the analytic structure alone. Although we focus on the application to Feynman integrals, the techniques presented here are more general and could apply to a wide range of problems where an exact answer is needed and the function space is sufficiently well understood.

hep-th↗
arXiv2024

Learning the Simplicity of Scattering Amplitudes

The simplification and reorganization of complex expressions lies at the core of scientific progress, particularly in theoretical high-energy physics. This work explores the application of machine learning to a particular facet of this challenge: the task of simplifying scattering amplitudes expressed in terms of spinor-helicity variables. We demonstrate that an encoder-decoder transformer architecture achieves impressive simplification capabilities for expressions composed of handfuls of terms. Lengthier expressions are implemented in an additional embedding network, trained using contrastive learning, which isolates subexpressions that are more likely to simplify. The resulting framework is capable of reducing expressions with hundreds of terms - a regular occurrence in quantum field theory calculations - to vastly simpler equivalent expressions. Starting from lengthy input expressions, our networks can generate the Parke-Taylor formula for five-point gluon scattering, as well as new compact expressions for five-point amplitudes involving scalars and gravitons. An interactive demonstration can be found at https://spinorhelicity.streamlit.app .

hep-th↗
arXiv2024

The Landau Bootstrap

We advocate a strategy of bootstrapping Feynman integrals from just knowledge of their singular behavior. This approach is complementary to other bootstrap programs, which exploit non-perturbative constraints such as unitarity, or amplitude-level constraints such as gauge invariance. We begin by studying where a Feynman integral can become singular, and the behavior it exhibits near these singularities. We then characterize the space of functions that we expect the integral to evaluate to, in order to formulate an appropriate ansatz. Finally, we derive constraints on where each singularity can appear in this ansatz, and use information about the expansion of the integral around singular points in order to determine the value of all remaining free coefficients. Throughout, we highlight how constraints that have previously only been derived for integrals with generic masses can be extended to integrals involving particles of equal or vanishing mass. We illustrate the effectiveness of this approach by bootstrapping a number of examples, including the four-point double box with a massive internal loop.

hep-ph↗
arXiv2024

The Collective Coordinate Fix

Collective coordinates are frequently employed in path integrals to manage divergences caused by fluctuations around saddle points that align with classical symmetries. These coordinates parameterize a manifold of zero modes and more broadly provide judicious coordinates on the space of fields. However, changing from local coordinates around a saddle point to more global collective coordinates is remarkably subtle. The main complication is that the mapping from local coordinates to collective coordinates is generically multi-valued. Consequently one is forced to either restrict the domain of path integral in a delicate way, or otherwise correct for the multi-valuedness by dividing the path integral by certain intersection numbers. We provide a careful treatment of how to fix collective coordinates while accounting for these intersection numbers, and then demonstrate the importance of the fix for free theories. We also provide a detailed study of the fix for interacting theories and show that the contributions of higher intersections to the path integral can be non-perturbatively suppressed. Using a variety of examples ranging from single-particle quantum mechanics to quantum field theory, we explain and resolve various pitfalls in the implementation of collective coordinates.

hep-th↗
arXiv2023

Neural Network Field Theories: Non-Gaussianity, Actions, and Locality

Both the path integral measure in field theory and ensembles of neural networks describe distributions over functions. When the central limit theorem can be applied in the infinite-width (infinite-$N$) limit, the ensemble of networks corresponds to a free field theory. Although an expansion in $1/N$ corresponds to interactions in the field theory, others, such as in a small breaking of the statistical independence of network parameters, can also lead to interacting theories. These other expansions can be advantageous over the $1/N$-expansion, for example by improved behavior with respect to the universal approximation theorem. Given the connected correlators of a field theory, one can systematically reconstruct the action order-by-order in the expansion parameter, using a new Feynman diagram prescription whose vertices are the connected correlators. This method is motivated by the Edgeworth expansion and allows one to derive actions for neural network field theories. Conversely, the correspondence allows one to engineer architectures realizing a given field theory by representing action deformations as deformations of neural network parameter densities. As an example, $ϕ^4$ theory is realized as an infinite-$N$ neural network field theory.

hep-th↗