Efficient Numerical Evaluation of a Two-Loop Contribution to the Dark-Matter Trispectrum

This paper presents a numerically efficient method for evaluating a two-loop contribution to the dark-matter trispectrum by expanding around a fixed reference cosmology, achieving sub-percent accuracy with only third-order terms and offering a practical pathway for faster computations of higher-order correlators in the Effective Field Theory of Large-Scale Structure.

The Gist

Here is an explanation of the paper, translated from complex physics jargon into everyday language using analogies.

The Big Picture: Mapping the Cosmic Web

Imagine the universe as a giant, invisible web made of dark matter. As the universe expands, this web gets denser in some places (forming galaxies) and emptier in others. Scientists want to understand exactly how this web is structured.

To do this, they use a "recipe" called the Effective Field Theory of Large-Scale Structure (EFTofLSS). This recipe helps them calculate how the web looks at different scales. However, as the universe gets more complex, the recipe requires adding more and more ingredients (mathematical terms called "loops").

The problem? Calculating these complex loops for every possible version of the universe is like trying to bake a cake for every single person on Earth, one by one. It takes too long and requires too much computer power.

The Problem: The "Baking" Bottleneck

In the past, if scientists wanted to test a new theory about the universe, they had to:

1. Pick a set of ingredients (cosmological parameters).
2. Run a massive, slow computer simulation to bake the cake (calculate the "trispectrum," which is a fancy way of measuring how four points in the universe are connected).
3. Repeat this thousands of times to find the best fit.

The old method was like trying to bake a new cake from scratch every time you changed the amount of sugar or flour. It was inefficient.

The Solution: The "Master Recipe" Trick

Andrea Favorito, the author of this paper, introduces a clever shortcut. Instead of baking a whole new cake from scratch for every variation, he suggests:

1. Pick a "Reference Cake": Bake one perfect, standard cake (a reference cosmology) that represents the "average" universe.
2. Measure the Differences: Instead of looking at the whole cake, just measure the tiny differences between your new cake and the reference cake. Is it slightly sweeter? A bit fluffier?
3. The Expansion: Mathematically, you can describe the new cake as:
   • (The Reference Cake) + (A little bit of extra sugar) + (A tiny bit of extra flour) + (A microscopic pinch of salt).

In the paper, the "Reference Cake" is a specific model of the universe. The "extra sugar and flour" are small mathematical differences ($\Delta P$) between the reference model and the new model you are testing.

The Magic Analogy: The "Zoom Lens"

Imagine you are looking at a landscape through a camera lens.

• The Old Way: To see a different landscape, you had to rebuild the entire camera and the scenery from the ground up.
• The New Way: You keep the camera and the scenery fixed. You just adjust the zoom and the focus slightly.

Favorito's method says: "We don't need to recalculate the whole universe for every new theory. We just need to calculate how the differences affect the result."

Because these differences are usually very small, you only need to calculate the first few "adjustments" (up to the third order) to get a result that is accurate to within 1%. You can ignore the tiny, microscopic adjustments because they don't change the picture much.

The "Infrared Safety" Fix: Cleaning the Lens

There was one technical snag. When doing these calculations, the math sometimes gets "messy" at the very edges (like trying to measure something that is infinitely small or infinitely large). This is called an "infrared singularity." It's like trying to take a photo of a foggy window; the image gets blurry and the computer crashes.

The author developed a special "lens cleaner" (an IR-safe integrand). This is a mathematical trick that rearranges the calculation so that the foggy parts cancel each other out. It ensures the computer can take a clear, sharp photo of the universe without getting stuck on the blurry edges.

Why This Matters

This paper is a breakthrough because it turns a "supercomputer-only" task into something that can be done much faster.

• Before: To test 1,000 different universe theories, you might need 1,000 years of computer time.
• After: With this new method, you bake the "Reference Cake" once, and then you just tweak the recipe for the other 999. It might take only a few days.

The Bottom Line

This paper provides a practical roadmap for scientists to analyze the universe's structure much faster. By treating the universe as a "standard model plus small tweaks," and by cleaning up the messy math at the edges, we can now explore complex theories about dark matter and the early universe with a speed and efficiency that was previously impossible.

It's the difference between hand-crafting every single brick of a cathedral versus having a master blueprint where you only need to adjust the color of the mortar.

Technical Summary

Here is a detailed technical summary of the paper "Efficient Numerical Evaluation of a Two-Loop Contribution to the Dark-Matter Trispectrum" by Andrea Favorito.

1. Problem Statement

The statistical precision of Large-Scale Structure (LSS) measurements is rapidly improving, necessitating theoretical predictions beyond tree-level or one-loop approximations within the Effective Field Theory of Large-Scale Structure (EFTofLSS).

• Computational Bottleneck: Evaluating higher-loop corrections (e.g., two-loop) and higher-multiplicity correlators (e.g., the trispectrum, or 4-point function) involves complex, high-dimensional momentum integrals.
• Likelihood Analysis Challenge: Cosmological parameter inference requires repeated evaluations of these integrals across a vast parameter space. Direct numerical integration at every likelihood call is computationally unfeasible.
• Limitations of Current Methods: Standard acceleration techniques (e.g., FFTLog, COBRA) rely on expanding the linear matter power spectrum ($P_{\text{lin}}$) into a basis of $N$ functions. For a diagram with $M$ internal power spectrum insertions, the number of precomputed cosmology-independent "building blocks" scales as $O(N^M)$. For a two-loop trispectrum contribution with five $P_{\text{lin}}$ insertions, this scales as $O(N^5)$, which becomes prohibitive for high-precision cosmology.

2. Methodology

The author proposes a reorganization of the calculation that expands the cosmology dependence around a fixed reference model rather than decomposing the power spectrum into a global basis.

A. Perturbative Expansion Around a Reference Cosmology

Instead of expanding $P_{\text{lin}}$ directly, the linear power spectrum for a target cosmology $j$ is written as:
$$P^{[j]}_{\text{lin}}(k) = N^{[j]} P^{[0]}_{\text{lin}}(k) + \Delta P^{[j]}_{\text{lin}}(k)$$
where:

• $P^{[0]}_{\text{lin}}$ is the power spectrum of a fixed reference cosmology.
• $N^{[j]}$ is a rescaling factor defined by the ratio of the maximum power of the target to the reference.
• $\Delta P^{[j]}_{\text{lin}}$ is the small difference between the target and reference spectra.

The trispectrum is then expanded perturbatively in terms of $\Delta P_{\text{lin}}$. For a diagram with five $P_{\text{lin}}$ factors, this generates $2^5 = 32$terms, organized by the number of$\Delta P_{\text{lin}}$insertions ($m=0$to$5$).

B. Infrared (IR) Safe Integrand Construction

The specific topology studied is a two-loop "sunset-like" diagram (labeled "2233") contributing to the dark-matter trispectrum.

• Challenge: While the full integral is IR-finite, the raw integrand contains local IR singularities (poles) when internal momenta become soft. These singularities degrade numerical stability.
• Solution: The author adapts a technique from Ref. [10] to construct an IR-safe integrand.
  1. Power Counting: The integrand is analyzed in soft limits ($q \to 0$).
  2. Partition of Unity: An IR measure $\mu$ (vanishing in soft limits) is introduced via a partition of unity. The identity $1 = \frac{1}{D} \sum \mu_i$ is inserted into the integral.
  3. Rearrangement: The integral is split into 8 terms. Each term involves momentum shifts that ensure the integrand is manifestly finite in all IR and UV regions. This allows for stable numerical evaluation using standard Monte Carlo integration (Vegas).

C. Truncation Strategy

Since $\Delta P_{\text{lin}}$ is small for cosmologies close to the reference, the expansion is truncated at a low order $m$.
$$T^{[j]} \approx (N^{[j]})^5 T^{[0]} + \sum_{i=1}^{m} (N^{[j]})^{5-i} \Delta_i T^{[j]}$$
The goal is to determine the minimum $m$ required to achieve sub-percent accuracy.

3. Key Contributions

1. IR-Safe Numerical Implementation: The paper provides a concrete, stable numerical implementation for a complex two-loop trispectrum topology that would be analytically intractable due to its complexity (7 propagators, non-degenerate masses).
2. Efficiency via Expansion: It demonstrates that expanding around a reference cosmology significantly reduces the computational cost compared to direct basis decomposition.
3. Validation of Truncation: The study validates that truncating the $\Delta P_{\text{lin}}$ expansion at the third order ($m=3$) is sufficient for high-precision cosmology.

4. Results

The method was tested on a "square" momentum configuration ($k_1, k_2, k_3, k_4$ forming a square) over a range of scales $k \in [10^{-3}, 0.8] \, h/\text{Mpc}$.

• Convergence: The expansion converges rapidly.
  • $m=2$ (2nd order): Shows percent-level deviations, reaching a few percent at the upper end of the scale range.
  • $m=3$ (3rd order): Reproduces the full numerical result with sub-percent accuracy across the entire range.
• Suppression of Higher Orders: The $m=3$ term contributes only a $O(1\%)–O(2\%)$ correction, and higher-order terms are strongly suppressed.
• Computational Scaling:
  • Direct Basis: Scales as $O(N^5)$.
  • Proposed Method: By using a smaller basis $N_m$ for the higher-order $\Delta P$ terms (since they are smaller and smoother), the cost scales roughly as $O(N_3^3)$ for the leading higher-order term. This represents a drastic reduction in the number of required cosmology-independent building blocks.

5. Significance

This work provides a practical and efficient pathway for evaluating higher-loop and higher-multiplicity correlators in the EFTofLSS.

• Feasibility: It makes the inclusion of two-loop trispectrum contributions in cosmological likelihood analyses computationally feasible.
• Scalability: The reorganization of the calculation reduces the "curse of dimensionality" associated with the number of power spectrum insertions. This is crucial for future surveys (e.g., Euclid, LSST) where the constraining power of higher-order statistics will be fully exploited.
• Generalizability: The IR-safe integrand construction and the expansion strategy are applicable to other EFTofLSS integrands, offering a general framework for accelerating cosmological inference.