SubTropica

The paper introduces SubTropica, a Mathematica package that utilizes tropical geometry to symbolically integrate multi-polylogarithmic integrals like Feynman diagrams, alongside its independent HyperIntica engine and an AI-driven online library for cataloging and retrieving physics results.

The Gist

Imagine you are a chef trying to bake a very complex cake. This cake represents a Feynman integral, a mathematical recipe used by physicists to predict how subatomic particles crash into each other.

The problem? The recipe calls for ingredients that are "infinite" or "undefined" in certain spots. If you try to bake it directly, the kitchen explodes. In physics, these explosions are called divergences. For decades, physicists have had to manually fix these explosions by adding "counter-ingredients" (mathematical tricks) to cancel out the infinities before they could bake the cake. It was slow, tedious, and required a PhD in baking just to follow the instructions.

Enter SubTropica.

The Magic Kitchen Assistant

SubTropica is a new software tool (a "package" for the computer program Mathematica) that acts like a super-smart kitchen assistant. Its job is to take those messy, exploding recipes and automatically fix them so they can be baked into a perfect, finite cake.

Here is how it works, broken down into simple concepts:

1. The "Tropical" Map (Finding the Danger Zones)

Imagine the ingredients of your cake are laid out on a giant map. Some areas of the map are safe; others are "danger zones" where the recipe breaks down (the infinities).

• Old Way: The chef had to walk the whole map, step-by-step, guessing where the danger zones were.
• SubTropica's Way: It uses a branch of math called Tropical Geometry. Think of this as a high-tech drone that flies over the map and instantly draws a "polytope" (a multi-dimensional shape) around the danger zones. It doesn't just guess; it knows exactly where the infinities are hiding based on the shape of the ingredients.

2. The "Subtraction" Trick (The Safety Net)

Once the drone spots the danger zones, SubTropica uses a technique called Tropical Subtraction.

• Imagine you have a bucket of water (the integral) that is overflowing.
• Instead of trying to stop the water from flowing in, SubTropica instantly adds a second bucket of negative water (a "counter-term") that perfectly cancels out the overflow.
• The result? The water level is now stable and manageable. The recipe is now "locally finite," meaning it's safe to cook.

3. The "Hyperlogarithm" Assembly Line (The Baking)

Now that the recipe is safe, it needs to be cooked. The cooking process involves integrating (adding up) the ingredients one by one.

• SubTropica uses a sub-engine called HyperIntica. Think of this as a robotic assembly line.
• It takes the safe ingredients and processes them using a special language called Hyperlogarithms. These are like advanced, multi-layered logarithms that can describe the complex flavors of the particle collisions.
• The robot checks the order of operations carefully. If it tries to mix the ingredients in the wrong order, the cake collapses. SubTropica uses a smart algorithm to find the perfect order to mix everything so the cake rises perfectly.

Why is this a Big Deal?

Before this paper, doing these calculations was like trying to solve a Rubik's cube while blindfolded, using only one hand, and with the cube on fire.

• Speed: SubTropica can solve problems that would take humans months or years in a matter of hours (or even minutes for simpler ones).
• Complexity: It can handle "cutting-edge" problems, like a four-loop Feynman diagram (a very complex particle collision) that no one had ever successfully calculated before.
• Accessibility: It comes with a Graphical User Interface (GUI). You don't need to be a coding wizard. You can literally draw the particle collision diagram on the screen, click "Integrate," and get the answer.

The Library of Recipes

The authors also built a Library (available at subtropi.ca).

• Think of this as a public cookbook.
• If a physicist has calculated a specific particle collision before, they can upload the "recipe" and the "finished cake" to this library.
• Other scientists can search the library to see if someone else has already baked that specific cake, saving them from reinventing the wheel.

The "AI" Chef's Helper

The paper also mentions that the developers used AI (specifically an AI model) to help write the code.

• The AI acted like a junior sous-chef: it helped check the math, find better ways to organize the ingredients, and even helped write the code that runs the kitchen. This shows that AI is becoming a powerful partner in high-level scientific discovery, not just for writing emails, but for solving the universe's hardest math puzzles.

In Summary

SubTropica is a revolutionary tool that turns the impossible task of calculating particle collisions into a manageable, automated process. It uses geometric maps to find trouble spots, subtracts the infinities to make things safe, and then uses a robotic assembly line to bake the final result. It's like giving every physicist a super-powered, automated kitchen that never burns the cake.

Technical Summary

1. Problem Statement

High-energy physics calculations, particularly in perturbative Quantum Field Theory (QFT), rely heavily on the evaluation of multi-dimensional integrals. These include:

• Feynman integrals: Multi-loop scattering amplitudes.
• Phase-space integrals: Used in cross-section calculations.
• Correlators: Energy-energy correlators and cosmological correlators.

These integrals generally belong to the class of Euler integrals, defined as:
$$I(\varepsilon, s) = \int_{\mathbb{R}_{\ge 0}^n} \prod_i dx_i \prod_j P_j(x, s)^{a_j \varepsilon + b_j}$$
where $P_j$ are polynomials, $s$ represents kinematic variables, and $\varepsilon$ is a dimensional regulator (e.g., $D = 4 - 2\varepsilon$).

Key Challenges:

1. Divergences: Direct expansion of the integrand in $\varepsilon$ often leads to meaningless expressions because the integral is divergent at specific boundaries of the integration domain.
2. Linear Reducibility: To integrate these analytically using hyperlogarithms (iterated integrals), the integrand must be "linearly reducible." This means that at every step of iterated integration, the singularities (poles) of the integrand must be linear in the current integration variable. If this condition fails, standard algorithms (like those in Maple's HyperInt or FORM's HyperFORM) fail or produce non-closed forms.
3. Workflow Fragmentation: While powerful tools exist in Maple and FORM, the high-energy physics community predominantly uses Mathematica. Existing tools often require switching between platforms for different stages of calculation (generation, reduction, integration, numerical verification).

2. Methodology

The paper introduces SubTropica, a Mathematica package that automates the evaluation of these integrals by combining two main theoretical pillars:

A. Tropical Subtraction and Nilsson–Passare Continuation

To handle divergences, the package employs tropical geometry.

• Newton Polytope Analysis: The singularities of the integral are encoded in the geometry of the Newton polytope of the integrand's polynomials. The package uses the external tool polymake to compute the polytope and identify "divergent rays" (directions in parameter space where the integral diverges).
• Tropical Subtraction: Instead of expanding the integrand directly, SubTropica constructs counter-terms based on the combinatorial structure of the polytope's facets. It subtracts these counter-terms from the original integrand to create a "renormalized" integrand that is locally finite (convergent at the boundaries).
• Nilsson–Passare Continuation: For cases where the integrand does not satisfy specific geometric properties (e.g., non-logarithmic divergences), the package performs an analytical continuation along specific rays to convert the integral into a sum of locally finite terms.

B. Hyperlogarithm Integration (HyperIntica)

Once the integrand is rendered locally finite, it is expanded in $\varepsilon$. The resulting terms are integrated using HyperIntica, a native Mathematica reimplementation of the HyperInt algorithm (originally by Panzer).

• Algorithm: The algorithm integrates variables one by one. It relies on the shuffle algebra of hyperlogarithms to reorganize products of iterated integrals.
• Linear Reducibility Check: Before integration, the package uses a Fubini reduction algorithm to find an integration order where all polynomials remain linear in the integration variable. If no such order exists naturally, it can introduce algebraic roots (square roots) to restore linear reducibility.
• Regularization: The package implements shuffle regularization to handle divergences at integration endpoints ($0, 1, \infty$), extracting finite parts while preserving algebraic relations.

3. Key Contributions

1. SubTropica Package: A comprehensive Mathematica tool that automates the entire pipeline from Feynman diagram input to the final Laurent expansion in $\varepsilon$.

   • Input Formats: Accepts Feynman diagrams (graph topology), lists of propagators (with tensor numerators and linearized/eikonal propagators), and generic Euler integrands.
   • GUI: Features a graphical user interface for drawing diagrams and configuring parameters, lowering the barrier to entry.
   • HyperIntica: A standalone, self-contained Mathematica implementation of the hyperlogarithm integration engine, decoupled from the tropical subtraction logic.

2. AI-Driven Library: The authors released an extensive, open-source library of Feynman integrals found in the literature.

   • Database: Contains diagrams, references, and computed results (hyperlogarithms).
   • Interface: Accessible via a web GUI (subtropi.ca) and the Mathematica package, allowing users to search, visualize, and retrieve results.
   • AI Integration: The development and library construction were assisted by AI (Claude Opus) for algebraic manipulation, rationalization of square roots, and numerical verification.

3. Advanced Capabilities:

   • Algebraic Letters: Handles integrals with square-root letters by automatically rationalizing them.
   • Tensor Numerators: Supports linearized propagators and tensor numerators (e.g., $v \cdot k$).
   • Beyond Scattering: Demonstrated applicability to non-standard integrals like gravitational energy-energy correlators and small-$x$ physics Fourier transforms.

4. Results and Performance

The paper validates the package through a series of benchmarks and examples:

• Cutting-Edge Benchmark: The package successfully evaluated a four-loop, nine-propagator massive Feynman diagram (Fig. 1 in the paper) for the first time, obtaining the result through $O(\varepsilon^0)$.
  • Performance: The computation took approximately 38 hours on a 32-core cluster, with the tropical subtraction phase taking ~4.7s, finding the linear order taking ~1.2 hours, and integration taking ~36.6 hours.
• Standard Examples: Successfully computed one-loop boxes with algebraic letters and two-loop triangle-box integrals, matching known results in the literature.
• Non-Feynman Applications:
  • Calculated Gravitational Energy-Energy Correlators in $N=8$ supergravity and pure Einstein gravity.
  • Evaluated Fourier transforms in small-$x$ QCD physics involving logarithmic numerators and multiple regulators.
• Numerical Verification: The package includes STVerify, which cross-checks symbolic results against numerical evaluations using backends like pySecDec, FIESTA, and AMFlow, achieving relative errors of $\sim 10^{-5}$.

5. Significance

• Unification of Workflow: SubTropica keeps the entire calculation workflow (diagram definition, subtraction, integration, and verification) within a single environment (Mathematica), addressing a major fragmentation issue in the field.
• Accessibility: The GUI and automated handling of complex steps (like finding linear orders and tropical subtraction) make high-loop calculations accessible to non-experts.
• Scalability: By successfully tackling a 4-loop massive diagram, the package demonstrates that tropical subtraction combined with hyperlogarithm integration is a viable strategy for next-to-next-to-next-to-leading order (N3LO) and beyond calculations.
• Open Science: The release of the code (MIT license) and the community-driven library (CC BY-NC-SA) fosters reproducibility and collaborative expansion of the known integral landscape.
• Future Outlook: The authors identify future directions, including extending the method to elliptic integrals, improving the handling of non-positive integrands (crown diagrams), and implementing "expansion by regions."

In summary, SubTropica represents a significant leap forward in the automation of perturbative QFT calculations, leveraging tropical geometry to tame divergences and providing a robust, user-friendly platform for symbolic integration.