The Light Quark Connected Hadronic Vacuum Polarization Contribution to the muon anomaly via Sparsened Meson Fields

This paper presents an improved determination of the light-quark connected hadronic vacuum polarization contribution to the muon anomalous magnetic moment using a fine 2+1+1 HISQ lattice ensemble, achieved by implementing a sparsening strategy within the low-mode averaging framework to efficiently reduce the computational cost of the dominant low-low correlator component while preserving signal quality.

The Gist

The Big Picture: The "Muon Mystery"

Imagine the muon as a tiny, spinning top. In the world of physics, this top has a magnetic field, like a tiny bar magnet. Scientists have a very precise prediction for how strong this magnet should be based on the "Standard Model" (the rulebook of the universe).

However, when scientists actually measure the muon in real life, it spins slightly differently than the rulebook predicts. This tiny difference is called the muon anomaly ($a_\mu$).

Why does this matter?

• If the difference is real, it means our rulebook is missing a chapter. It could hint at new particles or forces we haven't discovered yet.
• To be sure, we need to check the rulebook's math with extreme precision. The biggest source of "math error" in the rulebook comes from the Strong Force (the glue that holds atoms together).

The Problem: The "Noisy Room"

To calculate the Strong Force's effect, scientists use Lattice QCD. Imagine the universe is a giant 3D grid (like a massive Rubik's cube made of invisible strings). To calculate the physics, they simulate particles moving on this grid.

The problem is that the grid is huge and the calculations are expensive.

• The "Low Modes" (The Heavy Lifters): Some parts of the calculation (called "low modes") are like the heavy, slow-moving trucks in a city. They carry the most important information but are very hard to move. They take up 90% of the computer time.
• The "High Modes" (The Fast Cars): Other parts are like fast sports cars. They are easier to calculate but contribute less to the final answer.

The Challenge: The researchers needed to simulate a much larger, finer grid (the 144c ensemble) to get a more accurate answer. But on this giant grid, the "heavy trucks" (low modes) became so expensive to move that the calculation was becoming impossible to finish in a reasonable time.

The Solution: "Sparsening" the Mess

The authors (led by Vaishakhi Moningi) came up with a clever trick called "Sparsening Meson Fields."

The Analogy: The Crowd Survey
Imagine you want to know the average opinion of a crowd of 10,000 people in a stadium.

• Old Way: You ask every single person. This takes forever and costs a fortune.
• The New Way (Sparsening): You realize that people sitting right next to each other usually have the same opinion. So, instead of asking everyone, you ask every 4th person in a regular pattern.
  • You skip 3 people, ask the 4th, skip 3, ask the 4th.
  • Because the people are so close together, the 4th person's opinion is a perfect stand-in for the 3 you skipped.
  • Result: You get the same answer, but you only did 25% of the work.

In the paper, they applied this to the "meson fields" (the mathematical representation of the particles). By skipping specific points on the grid in a regular pattern, they reduced the amount of data the computer had to process by a factor of 4 or more, without losing the accuracy of the result.

The Results: Sharper Eyes

By using this "Sparsening" trick, the team was able to:

1. Run the simulation on a finer grid: They used a grid with a spacing of 0.042 femtometers (that's 0.000000000000000042 meters!). This is like switching from a blurry photo to a 4K ultra-HD image.
2. Reduce the error: Their new calculation reduced the uncertainty in the muon anomaly by a factor of 1.4 compared to their previous work.

The Final Numbers:
They calculated the contribution of light quarks (the building blocks of protons and neutrons) to the muon anomaly.

• Old Result: Roughly 646 (with a big error bar).
• New Result: 661 (with a much smaller error bar).

Why This Matters

The paper compares their new, sharper number with two other ways of getting the answer:

1. Other Lattice Simulations: Their result agrees well with other teams doing similar grid calculations.
2. Experimental Data (The "Real World" measurements): Their result is getting closer to the experimental measurements, but there is still a tiny "tension" (a difference of about 2 sigma).

The Takeaway:
This paper is a victory for efficiency. The researchers didn't just throw more supercomputers at the problem; they invented a smarter way to organize the data (Sparsening). This allowed them to see the universe with higher resolution.

While they haven't solved the "Muon Mystery" completely yet, they have tightened the net. They are now ready to double their data collection to see if that tiny difference between the theory and reality is a fluke or a sign of New Physics waiting to be discovered.

Summary in One Sentence

The team used a clever "skip-a-step" trick to make a super-complex physics simulation run faster and clearer, giving us a more precise measurement of a tiny magnetic quirk in the muon that might reveal new secrets of the universe.

Technical Summary

1. Problem Statement

The paper addresses the calculation of the light-quark connected contribution to the Hadronic Vacuum Polarization (HVP) for the muon anomalous magnetic moment ($a_\mu$).

• Context: The muon $g-2$ anomaly is a critical test of the Standard Model. While experimental precision has improved significantly (Fermilab measurements), theoretical predictions from Lattice QCD must match this precision to resolve current tensions between lattice results and data-driven evaluations.
• Challenge: The HVP calculation requires evaluating a Euclidean time correlation function $C(t)$. The Low-Low (LL) component of this correlation (involving low eigenmodes of the Dirac operator) dominates the statistical uncertainty, particularly in the long-distance tail, but is also the most computationally expensive part to evaluate.
• Specific Goal: To improve the precision of the light-quark connected HVP contribution ($a_\mu^{\text{HVP,lqc}}$) by incorporating a new, finer lattice ensemble (144c) and developing an efficient computational strategy to handle the high cost of the LL contribution.

2. Methodology

A. Lattice Setup

• Ensembles: The study utilizes 2+1+1 flavor Highly Improved Staggered Quark (HISQ) gauge configurations generated by the MILC Collaboration.
• New Data: A key addition is the 144c ensemble, featuring:
  • Physical pion mass ($m_\pi \approx 134.5$ MeV).
  • Fine lattice spacing ($a \approx 0.0426$ fm).
  • Large volume ($144^3 \times 288$ sites).
  • 4000 low modes of the preconditioned Dirac operator computed per configuration.
• Previous Data: Results from five other ensembles (32c to 96c) from previous work are included for continuum extrapolation.

B. Theoretical Framework: Low-Mode Averaging (LMA)

The quark propagator $S$ is spectrally decomposed into low modes ($S_L$) and high modes ($S_H$). The correlation function $C(t)$ is split into four components:

1. LL (Low-Low): Dominant statistical uncertainty; computationally expensive.
2. HL (High-Low) & LH (Low-High): Mixed terms.
3. HH (High-High): High-mode contribution.

The authors use Low-Mode Averaging (LMA) and All-Mode Averaging (AMA) to separate these terms, allowing for efficient computation of the high-mode parts via deflated Dirac operators.

C. Innovation: Sparsened Meson Fields

To address the prohibitive cost of calculating the LL contribution on the large 144c lattice, the authors implement a sparsening strategy:

• Concept: The meson fields (constructed from eigenvectors) are computed on a subset of lattice sites rather than the full lattice.
• Implementation:
  • Sites are omitted in a regular pattern (e.g., omitting $s$ consecutive points).
  • To preserve the spin-taste structure of staggered fermions and ensure projection to zero spatial momentum, the sparsening is applied by randomly selecting a "hypercube" on a time-slice and then omitting every $s$-th hypercube in spatial directions.
  • Constraints: Sparsening is limited to a factor of 4 in each dimension to prevent excessive noise.
• Impact: This reduces the memory footprint and computational cost of the meson field calculation from $N_S^3 \times N_T$ to $(N_S/s)^3 \times (N_T/t)$, enabling the inclusion of the 144c ensemble without prohibitive resource costs.

D. Analysis and Continuum Extrapolation

• Integration: The $a_\mu$ value is obtained by integrating the kernel-weighted correlation function. A bounding method is used to handle the noisy long-distance tail without summing over all time slices.
• Corrections: Finite-volume (FV), taste-breaking (TB), and pion-mass retuning effects are applied using:
  • NLO/NNLO Staggered Chiral Perturbation Theory (SChPT).
  • Chiral Vector Model (ChVM) (specifically developed for staggered fermions).
• Extrapolation: Results are extrapolated to the continuum limit ($a \to 0$) using an $a^2$ ansatz across six ensembles.

3. Key Contributions

1. Sparsening Technique: Successfully demonstrated that sparsening meson fields allows for the efficient calculation of the dominant LL term on very large, fine lattices (144c) while preserving signal quality.
2. New High-Precision Ensemble: Inclusion of the 144c ensemble (physical pion mass, fine spacing) significantly reduces discretization errors and improves control over the long-distance behavior of the correlator.
3. Improved Precision: Achieved a reduction in total uncertainty by a factor of 1.4 compared to previous analyses (based on ensembles up to 96c).

4. Results

The paper presents the light-quark connected HVP contribution to $a_\mu$ ($a_\mu^{\text{HVP,lqc}}$) in units of $10^{-10}$:

Scheme	With Taste Breaking (TB)	Without TB	Final Value (with TB)
NNLO SChPT	$664.4 \pm 9.2$	$656.5 \pm 9.2$	$661(10)$
ChVM	$657.3 \pm 9.2$	$647.5 \pm 9.2$	$652(10)$

• Total Uncertainty: Reduced to $\approx 10 \times 10^{-10}$ (from $\approx 14 \times 10^{-10}$ in previous work).
• Window Quantities:
  • Short Window (0.4–1.0 fm): $206.07 \pm 0.38$ (ChVM). Note: Continuum extrapolation here is model-dependent due to the short distance.
  • Long Window (1.5–1.9 fm): $101.6 \pm 1.5$ (NNLO with TB).

Comparison with Other Results:

• The result is consistent with other lattice calculations (BMW 21, RBC/UKQCD 24, Mainz/CLS 24) within $1\sigma$ to $2\sigma$.
• A tension remains with data-driven evaluations (e.g., Benton 24, KNT, DHMZ), showing a difference of roughly 1.5$\sigma$ to 2.35$\sigma$.

5. Significance and Future Work

• Resolution of Tension: The updated result narrows the gap between lattice QCD and data-driven methods but does not fully resolve the tension with data-driven evaluations, highlighting the need for even higher precision.
• Methodological Advancement: The sparsening technique provides a scalable path for future calculations on even finer lattices where full eigenvector computation is impossible.
• Future Plans: The authors plan to double the statistics on the fine lattices to achieve a sub-percent level of precision. They also propose exploring multi-grid methods or machine learning frameworks to further optimize the calculation of the High-Low (HL) contribution, which currently develops large uncertainties at long Euclidean times.

In summary, this work represents a significant step forward in Lattice QCD precision for the muon $g-2$ by combining a new fine lattice ensemble with a novel computational optimization (sparsening) to tackle the most expensive part of the calculation.