⚛️ phenomenology

Canonical differential equations beyond polylogs

This paper presents a method for systematically constructing canonical differential equations to compute Feynman integrals associated with complex geometries like elliptic curves and Calabi-Yau varieties, using the two-loop sunrise integral as a primary example to demonstrate the approach beyond multiple polylogarithms.

Original authors: Claude Duhr, Sara Maggio, Christoph Nega, Benjamin Sauer, Lorenzo Tancredi, Fabian J. Wagner

Published 2026-02-05

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Claude Duhr, Sara Maggio, Christoph Nega, Benjamin Sauer, Lorenzo Tancredi, Fabian J. Wagner

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: Solving the Universe's Math Puzzles

Imagine physicists are trying to predict exactly how particles smash into each other in giant machines like the Large Hadron Collider. To do this, they have to solve incredibly complex math problems called Feynman integrals. You can think of these integrals as the "receipts" for particle interactions—they tell us the probability of specific events happening.

For a long time, these receipts were relatively simple. They could be written down using a standard set of mathematical tools called polylogarithms (think of these as the "Lego bricks" of standard physics math). When the math was this simple, the authors of the paper (and others before them) found a special way to organize the equations, called a canonical form.

The Analogy: Imagine you are packing a suitcase. If you just throw clothes in randomly, it's a mess. But if you use a specific folding technique (the "canonical form"), everything fits perfectly, you can find anything instantly, and you know exactly how much space you have left. This technique made calculating particle collisions much faster and cleaner.

The Problem: The Math Got Too Complex

The paper explains that as physicists look at more complex scenarios (like two particles interacting over two "loops" of time, or involving heavy particles), the math stops being simple Lego bricks.

Instead of a flat surface (like a sheet of paper), the geometry of these problems starts looking like a doughnut (a mathematical shape called an elliptic curve) or even more complex, multi-holed shapes.

The Old Tools: The standard "Lego bricks" (polylogarithms) only work well on flat surfaces. They break down when the shape gets curved and twisted like a doughnut.
The Result: The old "packing technique" (canonical form) stopped working. The equations became messy, and the elegant properties that made them easy to solve disappeared.

The Solution: A New Packing Technique for Doughnuts

The authors of this paper present a new method to fix this. They figured out how to create a "canonical form" even when the math involves these complex doughnut shapes.

Here is how their method works, broken down into steps:

1. Finding the "Skeleton" (Leading Singularities)

First, they look at the "skeleton" of the problem. In the old days, the skeleton was easy to see. Now, with the doughnut shapes, the skeleton is hidden.

The Analogy: Imagine trying to understand the structure of a complex, knotted rope. You can't just look at the whole thing; you have to pull on the ends to see how the knots are tied. The authors use a technique called "maximal cuts" to pull on the ends of the math problem and reveal its underlying shape (the elliptic curve).

2. Identifying the "Pure" Ingredients

Once they see the shape, they need to find the right "ingredients" (mathematical functions) to build their solution.

The Challenge: On a doughnut, there are two types of paths you can take: ones that go around the hole and ones that go through the hole. The old math only knew about the "through" paths.
The Fix: The authors identify new mathematical functions that correspond to these new paths. They realize that to keep the math "pure" (clean and organized), they need to mix these new functions in a very specific way, almost like balancing weights on a scale.

3. The "Magic Rotation" (The Algorithm)

This is the core of their discovery. They developed an algorithm (a step-by-step recipe) to rotate and rearrange the equations until they snap into the perfect "canonical form."

The Analogy: Imagine you have a jumbled Rubik's Cube where the colors are mixed up in a way that no standard algorithm can solve. The authors invented a new sequence of moves.
1. Step 1: They look at the "flat" version of the problem (where the doughnut is squashed) to get a starting point.
2. Step 2: They apply a "semi-simple" rotation. This is like separating the "solid" parts of the math from the "wobbly" parts.
3. Step 3: They apply a "unipotent" rotation. This is like tightening the screws to make sure the pieces fit together perfectly without any gaps.
4. Step 4: They adjust the "epsilon" (a tiny number used to fix math errors). This ensures that every layer of the solution is perfectly uniform, just like a well-layered cake.

Why This Matters (According to the Paper)

The paper claims that by using this new method:

Simplicity Returns: Even though the underlying geometry is a complex doughnut (or even more complex shapes like Calabi-Yau varieties), the final equations look just as clean and organized as the simple ones from the past.
Universality: This method isn't just for doughnuts. The authors say it works for any shape, including the most exotic geometries found in modern physics.
Specific Examples: They tested this specifically on the "two-loop sunrise integral" (a specific type of particle interaction diagram that looks like a sunrise). They showed that whether the particles are massless (simple) or massive (complex doughnut geometry), their method produces a clean, solvable equation.

Summary

Think of this paper as a new instruction manual for assembling the most complex furniture in the universe. Previously, if the furniture had a weird, curved shape (like a doughnut), the instructions were a mess, and the pieces didn't fit. The authors have written a new set of instructions that tells you exactly how to fold, rotate, and align the pieces so that even the weirdest, most complex shapes snap together perfectly, making the math easy to read and solve.

Technical Summary: Canonical Differential Equations Beyond Polylogs

Problem Statement
Modern precision calculations in Quantum Field Theory (QFT), ranging from collider cross-sections to post-Minkowskian gravitational-wave scattering, increasingly involve Feynman integrals associated with geometries extending beyond the Riemann sphere. While one-loop and massless amplitudes are well-understood and expressible via multiple polylogarithms (MPLs) characterized by logarithmic singularities, higher-loop integrals (starting at two loops) often evaluate to functions associated with elliptic curves and Calabi–Yau varieties.

The standard approach to solving these integrals relies on the method of differential equations, specifically seeking a "canonical form" where the system satisfies $d\mathbf{J} = \varepsilon \mathbf{A}(\mathbf{s}) \mathbf{J}$ . In this form, the connection matrix $\mathbf{A}(\mathbf{s})$ contains only simple poles (d-log forms), and the solution admits a uniform transcendental weight (UT) expansion. However, for integrals involving elliptic curves, the standard d-log forms are insufficient to span the cohomology of the underlying geometry. The presence of higher-order poles and the lack of a clear "leading singularity" structure in the elliptic regime make it unclear how to construct a basis of master integrals that preserves the desirable properties of canonical equations (UT weight and $\varepsilon$ -factorization) beyond the polylogarithmic case.

Methodology
The authors propose a systematic algorithm to construct canonical differential equations for integrals evaluating beyond polylogarithms, using the two-loop sunrise integral as a primary case study. The method proceeds in two main stages:

Initial Basis Selection via Geometric Analysis:
The authors utilize the Baikov representation to compute maximal cuts and leading singularities in integer dimensions ( $d=2$ ).
- For the polylogarithmic case ( $M^2=0$ ), the geometry degenerates to a Riemann sphere. The leading singularities are identified as rational functions with simple poles. A basis of pure master integrals is constructed by normalizing the integrals by the inverse of their leading singularities.
- For the elliptic case ( $M^2 \neq 0$ ), the geometry is defined by a quartic elliptic curve. The maximal cuts correspond to periods of the curve. The authors identify that the standard leading singularity concept must be generalized. Instead of a single scalar, they utilize the period matrix $\mathbf{W}$ , which pairs master integrands with independent homology cycles.
Basis Rotation and Canonicalization:
To achieve the canonical form for the elliptic case, the authors perform a sequence of basis rotations on the initial set of master integrals $\mathbf{I}$ :
- Decomposition of the Period Matrix: The period matrix $\mathbf{W}$ is decomposed into a semi-simple part ( $\mathbf{W}_{ss}$ ) and a unipotent part ( $\mathbf{W}_u$ ). The semi-simple part corresponds to the generalization of algebraic leading singularities (transcendental weight zero), while the unipotent part encodes the pure functions of higher transcendental weight.
- Normalization: The initial basis is rotated by the inverse of the semi-simple part of the period matrix ( $\mathbf{U}_{ss}$ ) to ensure the integrals have the correct transcendental weight at $\varepsilon=0$ .
- Weight Re-alignment: A diagonal rescaling matrix ( $\mathbf{U}_\varepsilon$ ) involving powers of $\varepsilon$ is applied to align the transcendental weights of the $\varepsilon$ -expansion coefficients, addressing the mixing of $\varepsilon$ -powers and higher-pole contributions.
- $\varepsilon$ -Factorization: Finally, a unipotent rotation ( $\mathbf{U}_t$ ) is applied to remove non-constant terms in the differential equation that are not proportional to $\varepsilon$ . This transforms the system into the canonical form $d\mathbf{J} = \varepsilon \mathbf{A}(\mathbf{s}) \mathbf{J}$ , where $\mathbf{A}(\mathbf{s})$ has simple poles.

Key Results

Generalization of Leading Singularities: The paper demonstrates that for elliptic integrals, the concept of a "leading singularity" generalizes from a scalar algebraic function to the semi-simple part of the period matrix. This allows for the identification of a basis where the singularity structure is manifest.
Construction of Pure Integrals: The authors successfully construct a basis of pure master integrals for the two-loop sunrise integral with non-zero masses. These integrals satisfy the conditions of uniform transcendental weight and possess differential equations with entries that are d-log forms (or their elliptic generalizations).
Algorithmic Robustness: The proposed algorithm is shown to be independent of the specific geometry. It successfully handles the transition from the polylogarithmic limit ( $M^2 \to 0$ ) to the elliptic regime, recovering the standard d-log forms in the limit.
Application Scope: The method is validated on the sunrise integral but is presented as applicable to a broader class of problems, including higher-genus curves and Calabi–Yau geometries.

Significance and Claims
The paper claims to provide a systematic method to extend the powerful framework of canonical differential equations to Feynman integrals that evaluate to functions beyond multiple polylogarithms. By establishing a clear procedure to identify pure integrals and construct $\varepsilon$ -factorized differential equations for elliptic and potentially higher-genus geometries, the work addresses a critical bottleneck in modern precision calculations.

The authors modestly frame the significance as a step toward a unified understanding of the analytic structure of scattering amplitudes in complex geometries. They note that while the algorithm yields the desired canonical form, the nature of the resulting " $\varepsilon$ -functions" (the solutions to the differential equations) requires further study. The paper does not claim to have solved the evaluation of these functions numerically or analytically in all cases, but rather to have provided the correct differential equation framework that makes such evaluation feasible and manifestly preserves the singularity structure of the theory.

Future Directions Mentioned
The authors suggest future work should focus on:

Gaining a deeper understanding of the $\varepsilon$ -functions generated by this method.
Rigorously proving that the method always yields a canonical form for arbitrary geometries.
Comparing this approach with other existing methods in the literature.
Assessing the numerical performance of the special functions appearing in the resulting differential equations.