Approximating Feynman Integrals Using Complete Monotonicity and Stieltjes Properties

This paper introduces two novel numerical approaches for computing Feynman integrals by leveraging their complete monotonicity and Stieltjes properties to impose strong constraints on differential equation solutions and enable efficient rational approximations via Padé approximants, respectively.

The Gist

Imagine you are trying to predict the weather in a specific city. You have a complex mathematical model (the "differential equations") that tells you how the temperature changes based on wind speed, humidity, and time. However, the model is so complicated that solving it exactly is like trying to calculate the path of every single raindrop in a storm. It's too hard.

Usually, physicists have to run massive, slow computer simulations to get a number for the temperature at a specific moment. But what if you could guess the answer with incredible accuracy just by knowing a few simple rules about how temperature behaves?

This paper introduces two new "super-guessing" methods for calculating Feynman integrals. In the world of particle physics, these integrals are the mathematical recipes used to predict what happens when particles smash together (like at the Large Hadron Collider). They are notoriously difficult to solve, often requiring supercomputers and weeks of calculation.

The authors, Sara Ditsch, Johannes Henn, and Prashanth Ramana, have found that these integrals follow two very strict, predictable patterns. By exploiting these patterns, they can calculate the answers much faster and more accurately than before.

Here is how they do it, explained through simple analogies:

1. The "Smooth Slide" Rule (Complete Monotonicity)

The Concept:
Imagine a ball rolling down a perfectly smooth, frictionless slide. You know it will always go down (it never goes up). You also know that as it rolls, it doesn't just go down; it slows down its descent in a very specific, predictable way. It never wobbles, never speeds up, and never reverses direction.

In math, this behavior is called Complete Monotonicity (CM). The authors discovered that Feynman integrals behave exactly like this ball on a slide when you look at them in a specific "safe zone" (called the Euclidean region).

The Method:
Instead of trying to solve the whole slide at once, the authors use a "bootstrap" method.

• The Setup: They have the rules of the slide (the differential equations).
• The Constraint: They know the ball must always roll down smoothly and never wobble.
• The Trick: They ask a computer: "Given the rules and the fact that the ball must roll smoothly, what are the only possible values the ball can have at this specific spot?"

Because the "smooth rolling" rule is so strict, it squeezes the possible answers into a tiny, narrow range. It's like if you knew a car was driving on a straight road and you knew its speed was decreasing, you could guess its position in 10 seconds with amazing precision, even without knowing the exact engine specs.

The Result:
They tested this on complex "banana-shaped" particle diagrams (integrals with many loops). The method worked incredibly fast, often finding the answer in seconds where other methods took minutes or hours.

2. The "Magic Recipe" (Stieltjes Functions)

The Concept:
The first method is great, but it only works well in the "safe zone." What if you want to know the answer in a "stormy" region (where particles are actually colliding)?

The authors realized that these integrals aren't just smooth rollers; they are Stieltjes functions. Think of a Stieltjes function as a dish made from a secret, high-quality recipe.

• The Recipe: The function is built from a mix of simple ingredients (positive numbers) that never spoil.
• The Magic: Because the ingredients are so pure and positive, you can reconstruct the entire dish just by tasting a tiny spoonful of it.

The Method:
This is where Padé Approximants come in.

• The Problem: Usually, if you try to guess a complex curve based on a few points (a Taylor series), your guess gets terrible the further you get from the starting point. It's like trying to guess the shape of a mountain just by looking at the grass at the bottom; you might think it's a hill, but it could be a volcano.
• The Solution: Because these integrals are Stieltjes functions, the "guessing" tool (Padé approximant) is magical. It doesn't just guess a curve; it builds a rational function (a fraction of polynomials) that is mathematically guaranteed to stay accurate even when you jump far away from the starting point.

The Analogy:
Imagine you have a map of a city, but it's torn and only shows the center.

• Normal Math: If you try to guess the rest of the city, you might draw a straight line that leads you off a cliff.
• Stieltjes Math: Because you know the city follows a specific "Stieltjes" layout (like a grid), you can use the center map to perfectly reconstruct the entire city, including the parts you can't see, even across the ocean.

The Result:
They used this to calculate a 20-loop banana integral. In the world of physics, 20 loops is like trying to solve a puzzle with 20 layers of complexity. Usually, this is impossible to do analytically. But by using the "Magic Recipe" (Stieltjes property) and tasting just a few points (Taylor expansion), they reconstructed the entire function with high precision, even in complex, physical regions where particles collide.

Why This Matters

1. Speed: They can calculate these values in seconds, whereas traditional methods might take hours or days.
2. Precision: They provide rigorous "upper and lower bounds." It's not just a guess; they can say, "The answer is definitely between 1.05 and 1.06."
3. No Boundary Needed: Traditional methods often need you to know the answer at a specific starting point to begin. These new methods can often start from almost nothing (just the shape of the equations) and bootstrap their way to the answer.
4. Universal: This works for complex, multi-loop diagrams that were previously too hard to handle.

The Bottom Line

The authors found that the chaotic world of particle collisions actually follows very strict, orderly rules (Complete Monotonicity and Stieltjes properties). By treating these integrals not as chaotic monsters, but as well-behaved mathematical objects, they built a "shortcut" that allows physicists to calculate the outcomes of particle collisions with unprecedented speed and accuracy.

It's like realizing that while a storm looks chaotic, the wind patterns actually follow a simple, predictable rhythm. Once you know the rhythm, you can predict the storm's path without needing to track every single drop of rain.

Technical Summary

1. Problem Statement

Feynman integrals are fundamental to computing scattering amplitudes in perturbative quantum field theory (QFT), particularly for collider physics and gravitational wave predictions. While analytic evaluation is possible for simple cases, multi-loop integrals with multiple scales remain a significant challenge.

• Current Limitations: Existing numerical methods (e.g., sector decomposition, contour deformation, tropical Monte Carlo) often suffer from high computational costs, memory constraints, or limited applicability to finite integrals. Methods based on differential equations require precise boundary conditions and often struggle with complex analytic continuations to physical scattering regions.
• Goal: The authors aim to develop efficient numerical methods to compute Feynman integrals by leveraging deep mathematical properties (Complete Monotonicity and Stieltjes properties) to constrain solutions without needing full analytic forms or complex boundary conditions.

2. Methodology

The paper introduces two distinct but related numerical approaches:

A. The Complete Monotonicity (CM) Bootstrap

This method combines the known linear differential equations satisfied by Feynman integrals with the property of Complete Monotonicity.

• Definition: A function $f(x)$ is completely monotonic in a region if $(-1)^n \frac{d^n}{dx^n} f(x) \geq 0$ for all $n \in \mathbb{N}_0$.
• Application: Scalar Feynman integrals in the Euclidean kinematic region are proven to be CM functions of kinematic invariants.
• Algorithm:
  1. Express the differential equation system as $\frac{d}{dx}\mathbf{f}(x) = A(x)\mathbf{f}(x)$.
  2. Derive recursive relations for the derivatives of the basis integrals using the matrix $A(x)$.
  3. Impose the CM inequality constraints ($Q_n(x)\mathbf{f}(x) \geq 0$) at a specific kinematic point $x_0$.
  4. Formulate a Linear Program (LP) to find the feasible region for the integral values. By normalizing one component, the LP maximizes/minimizes other components to generate tight upper and lower bounds.
  5. As the number of derivatives ($n_0$) increases, these bounds converge, uniquely determining the integral value.

B. Stieltjes Function Approximation

This method refines the CM approach by proving that Feynman integrals are Stieltjes functions under specific parameter ranges.

• Definition: A Stieltjes function $f(z)$ admits an integral representation $f(z) = \int_0^{1/R} \frac{\rho(u)}{1+uz} du$ with $\rho(u) \geq 0$. Stieltjes functions are a subset of CM functions.
• Key Advantage: Stieltjes functions can be approximated with high precision by Padé approximants (rational functions) that converge in the cut complex plane.
• Algorithm:
  1. Prove that a class of Feynman integrals (specifically those where the second Symanzik polynomial $F$ can be written as $Ax+B$ with $A,B \geq 0$) are Stieltjes functions.
  2. Compute a Taylor expansion of the integral around a point $x_0$ (either via the CM bootstrap or direct expansion of known representations like Bessel integrals).
  3. Construct Padé approximants $[N, M]$ from the Taylor coefficients.
  4. Use the rigorous convergence theorems of Stieltjes functions to analytically continue the result to physical regions (complex kinematics) and evaluate it efficiently.

3. Key Contributions

1. Proof of Stieltjes Property: The authors provide a rigorous proof that scalar Feynman integrals (with specific constraints on propagator powers and dimension) are Stieltjes functions in the Euclidean region. This generalizes previous results on Complete Monotonicity.
2. CM Bootstrap Framework: They developed a novel "bootstrap" technique that uses differential equations and CM inequalities to determine integral values numerically without explicit boundary conditions, solving a linear program to bound the solution space.
3. Efficient Analytic Continuation: By combining the CM bootstrap with Padé approximants, they demonstrated a method to transport numerical results from the Euclidean region to physical scattering regions (complex plane) with guaranteed error bounds.
4. Independence from Differential Equations: The Stieltjes approach allows for high-precision evaluation using only a Taylor expansion (e.g., from a soft limit or Bessel representation), bypassing the need to solve differential equations entirely.

4. Results and Performance

The methods were tested on several benchmark integrals:

• Massive Bubble Integral (Pedagogical Example):

  • The CM bootstrap successfully determined the integral value in the Euclidean region.
  • In regions with two-sided bounds, the method converged rapidly (e.g., achieving 15-digit precision with $n_0=30$ derivatives).
  • Performance varied by kinematic point: points near the upper bound of the convergence region were significantly faster than those near the lower bound.

• Multi-loop "Banana" Integrals (2 to 4 loops):

  • The CM bootstrap was applied to equal-mass banana integrals.
  • Comparison: The method was compared against AMFlow (a standard numerical integration tool).
    • For points near the optimal convergence region, the CM bootstrap was orders of magnitude faster (e.g., 0.2s vs 23s for 3-loop).
    • The method is highly efficient for evaluating multiple phase-space points once the differential equation and Padé approximants are pre-computed.

• 20-Loop Banana Integral:

  • Demonstrated the power of the Stieltjes approach without differential equations.
  • Using the Bessel integral representation to generate the first 20 Taylor moments (50-digit precision), the authors constructed Padé approximants.
  • Result: The Padé approximants provided high-precision numerical values across the complex kinematic plane, matching the Bessel integral representation to high accuracy. This proves the method works for extremely high-loop orders (20 loops) where analytic solutions are intractable.

• One-Loop Five-Particle Scattering:

  • Showed that specific combinations of transcendental functions appearing in 5-particle amplitudes are Stieltjes functions, allowing for efficient Padé-based evaluation.

5. Significance and Outlook

• Computational Efficiency: The methods offer a fast, high-precision alternative to traditional numerical integration, particularly for evaluating integrals at many phase-space points. The "pre-compute once, evaluate many times" nature of the Padé approach is a major advantage.
• Analytic Continuation: The rigorous convergence properties of Padé approximants for Stieltjes functions provide a mathematically sound way to move from Euclidean (safe) regions to physical (complex) scattering regions, a notoriously difficult step in QFT calculations.
• Broad Applicability: The authors argue these properties extend beyond particle physics to cosmology (cosmological correlators) and gravitational physics.
• Future Directions:
  • Extending the framework to multivariate Padé approximants to handle multiple kinematic variables simultaneously.
  • Applying the method to dimensional regularization ($\epsilon$-expansions) to constrain Laurent series coefficients.
  • Using Stieltjes properties to construct stable bases for special functions in QFT calculations.

In conclusion, this paper establishes a powerful new paradigm for computing Feynman integrals by exploiting the mathematical structure of Complete Monotonicity and Stieltjes functions, enabling efficient, high-precision numerical evaluation and analytic continuation for complex multi-loop systems.