Realization of the SI Second Defined by Geometric Mean of Multiple Clock Transitions

This paper investigates practical methods for realizing a proposed SI second definition based on the geometric mean of multiple clock transitions by deriving uncertainty expressions, comparing geometric versus arithmetic combination routes, and developing time-weighted strategies to minimize errors from dead time and non-overlapping clock operations.

The Gist

The Big Picture: Redefining the "Second"

Imagine you are baking a cake, and the recipe calls for exactly one cup of flour. For decades, the world has agreed that "one cup" is defined by a specific, physical metal cup kept in a vault. This is how we currently define the SI second: it's based on a specific vibration of a Cesium atom (like a very precise, tiny pendulum).

However, scientists have recently discovered "super-pendulums" (Optical Clocks) that are 100 to 1,000 times more accurate than the old Cesium one. They are so precise that the old definition feels like using a ruler made of rubber to measure a microchip.

The scientific community is debating a new definition. Instead of picking just one of these super-accurate clocks to be the new "gold standard," they are proposing a composite definition. Imagine defining a "cup" not by one specific container, but by the average volume of five different, highly precise measuring cups. This new definition is called a Weighted Geometric Mean.

The Problem: Not Everyone Has the Same Tools

The paper tackles a very practical problem: What happens when you try to use this new definition in the real world?

1. Missing Tools: Not every lab has all five types of super-clocks. Some might only have two.
2. Different Skill Levels: One lab might have a clock that is perfect, while another has one that is slightly "wobbly."
3. Dead Time: These clocks don't run 24/7. They need maintenance, or they might be turned off to compare with a "flywheel" clock (like a Hydrogen Maser). When they stop, the flywheel drifts a little bit, creating "dead time" errors.

The authors ask: How do we combine these messy, imperfect, and incomplete measurements to get the most accurate "Second" possible?

The Two Main Strategies: The "Math" of Averaging

The paper compares two ways to combine the data, using a simple analogy: Measuring the height of a building.

Strategy A: The Arithmetic Mean (The "Add and Divide" Method)

Imagine you ask three people to measure a building.

• Person A uses a laser (very accurate).
• Person B uses a tape measure (okay).
• Person C uses a ruler (not great).

If you simply add their numbers and divide by three, the person with the bad ruler drags the average down. This method works well if the "bad" measurements are just a little off, but it struggles if one measurement is wildly inaccurate.

Strategy B: The Geometric Mean (The "Multiplicative" Method)

This is the method proposed for the new SI definition. Instead of adding the numbers, you multiply them and take the root.

• The Analogy: Think of this like mixing ingredients for a sauce. If you have a tiny bit of a very strong spice (a highly accurate clock) and a lot of a mild spice (a less accurate clock), the geometric mean balances them differently than simple averaging.
• The Result: The paper finds that if your measurements are generally good (low uncertainty), the Geometric Mean is usually the winner. It is more robust and keeps the final result closer to the true value. However, if one of your measurements is terrible (very high uncertainty), the Geometric Mean can get skewed, and the Arithmetic Mean might actually be safer.

The "Crossover" Point: The authors did the math to find the exact "tipping point." If your clocks are good enough, use the Geometric Mean. If one clock is really bad, switch to the Arithmetic Mean.

The "Dead Time" Problem: The Broken Stopwatch

Here is the trickiest part. The super-accurate optical clocks are so sensitive they can't run continuously. They have to pause to check themselves against a "flywheel" clock (a Hydrogen Maser).

• The Metaphor: Imagine you are trying to measure the speed of a race car, but your stopwatch stops every 10 minutes to reset. When the stopwatch is off, the car keeps moving, but you don't know exactly how far it went. When you turn the stopwatch back on, you have to guess the gap. This "guessing" introduces a huge error called Dead Time Uncertainty.

In the past, scientists would just average the whole day's data, which made the error huge.

The Paper's Solution: The "Time-Segmented" Approach
The authors propose breaking the day into small chunks (time segments).

• Instead of averaging the whole day, they look at 10-minute blocks.
• They use a complex Covariance Matrix (think of this as a giant spreadsheet that tracks how the errors in one block relate to the errors in the next).
• They give more weight to the times when the best clocks were running and less weight to the times when the clocks were "wobbly" or off.

This is like a conductor leading an orchestra where some musicians are playing perfectly and others are taking a break. Instead of silencing the whole orchestra, the conductor listens to the perfect players during their solo and fills in the gaps with the best possible estimate from the backup players, ensuring the final song (the definition of the second) is perfect.

The Takeaway

This paper is a user manual for the future of timekeeping.

1. It confirms that the new "Geometric Mean" definition is a great idea, but you have to be careful about how you calculate it.
2. It provides a rulebook for when to use simple averaging vs. complex geometric averaging, depending on how good your clocks are.
3. It solves the "Dead Time" headache by using advanced math to stitch together fragmented measurements, ensuring that even if the clocks stop and start, the final definition of the "Second" remains incredibly precise.

In short, the authors are building the bridge between the theoretical "perfect second" and the messy reality of running multiple clocks in different labs around the world.

Technical Summary

1. Problem Statement

The current definition of the SI second relies on the microwave hyperfine transition of the $^{133}$Cs atom, with uncertainties around $(1-2) \times 10^{-16}$. In contrast, state-of-the-art optical atomic clocks have achieved uncertainties in the $10^{-18}$ to $10^{-19}$ range, prompting discussions on redefining the SI second based on optical transitions.

The Consultative Committee for Time and Frequency (CCTF) is considering a new definition (Option 2) where the second is defined by a constant $N$, representing the weighted geometric mean of a set of specific atomic clock transition frequencies. However, realizing this definition in practice presents significant challenges:

• Incomplete Data: Not all defining transitions may be available or measured in a single laboratory.
• Heterogeneous Performance: Different optical clocks have varying performance levels (uncertainties) and operational uptimes.
• Dead Time: When using a Hydrogen Maser (H-maser) as a flywheel reference, "dead time" (gaps in measurement) introduces significant uncertainty, often dominating the total error budget.
• Correlations: Measurements may be correlated due to shared environmental factors (e.g., temperature, magnetic fields) or reference standards.

The core problem is determining the optimal strategy to reconstruct the constant $N$ with minimum total uncertainty under these practical constraints.

2. Methodology

The authors propose and analyze two complementary realization routes: Geometric-Mean Combinations and Arithmetic-Mean Combinations. They derive consistent uncertainty expressions for both methods, incorporating measurement uncertainties, recommended frequency uncertainties, and frequency ratio uncertainties.

Key Analytical Approaches:

• Full Ensemble vs. Subset: The study covers scenarios where all defining transitions are measured directly versus scenarios where only a subset is measured (requiring the use of recommended values for missing transitions).
• Uncertainty Propagation:
  • Geometric Mean: Derived directly from measured frequencies. If all clocks are referenced to the same timekeeping clock, the uncertainty depends solely on measured frequency uncertainties. If different references are used, frequency ratio uncertainties are included.
  • Arithmetic Mean: Derived from individual clock estimates. This method inherently includes the uncertainty of the recommended frequencies for the transitions, which can degrade performance if measurement uncertainties are low.
• Handling Dead Time & Asynchrony: For campaigns where multiple clocks operate asynchronously with an H-maser flywheel, the authors introduce a time-segmented, time-weighted combination. This method divides the campaign into intervals based on clock availability and uses coefficient and covariance matrices to account for:
  • Overlapping operation periods.
  • Correlations between measurements (e.g., shared environmental shifts).
  • Dead-time-induced uncertainty propagation.

3. Key Contributions

1. Comparative Uncertainty Analysis: The paper provides explicit mathematical conditions and crossover points determining when the Geometric Mean method yields lower total uncertainty compared to the Arithmetic Mean method.
2. Hybrid Realization Strategy: It outlines how to reconstruct the constant $N$ when only a subset of transitions is measured, utilizing a hybrid approach that combines measured data with recommended frequency ratios.
3. Dead-Time Mitigation Framework: A novel mathematical framework using coefficient and covariance matrices is introduced to minimize the impact of H-maser dead time in multi-clock, asynchronous campaigns.
4. Case Studies: Analytic three-transition case studies are used to visualize parameter regimes (ratios of measurement uncertainty to recommended uncertainty) where one method outperforms the other.

4. Results

• Geometric vs. Arithmetic Mean:
  • The Geometric Mean method generally yields lower total uncertainty when the measurement uncertainties of the clocks are relatively low (specifically, when the ratio of measurement uncertainty to recommended uncertainty is less than 2).
  • The Arithmetic Mean method becomes advantageous when one clock has a significantly larger measurement uncertainty than the others, or when the ratio of uncertainties is high.
  • The crossover point depends on the ratio $k$ (uncertainty of the "worst" clock relative to others) and the ratio of measurement uncertainty to recommended uncertainty.
• Impact of Recommended Values: When using the Geometric Mean with a subset of transitions, the uncertainty of the recommended frequencies for unmeasured transitions becomes a limiting factor. The paper derives the optimal weighting to balance measured data against recommended values.
• Dead Time Reduction: The time-segmented approach significantly reduces the total uncertainty compared to treating the campaign as a single block. By weighting intervals based on the number of operational clocks and their correlations, the effective uncertainty is minimized, preventing the H-maser dead time from dominating the error budget.
• Correlation Effects: The study confirms that ignoring correlations (e.g., shared blackbody radiation shifts) leads to underestimating total uncertainty. The proposed covariance matrix formalism correctly accounts for these effects.

5. Significance

This work provides critical practical guidance for the metrology community regarding the potential redefinition of the SI second:

• Operational Feasibility: It demonstrates that the "Option 2" definition (weighted geometric mean) is practically realizable even without a single laboratory housing all optical clocks, provided a rigorous statistical combination strategy is used.
• TAI Calibration: The proposed methods directly support the calibration of International Atomic Time (TAI) and the steering of local time scales (UTC(k)) with lower uncertainties.
• Future-Proofing: The framework is scalable to arbitrary numbers of clocks and measurement segments, making it applicable to future global optical clock networks.
• Standardization: By defining how to handle dead time and correlations, the paper helps establish a standardized protocol for evaluating the uncertainty of the new SI second, ensuring consistency across different national metrology institutes.

In summary, the paper bridges the gap between the theoretical definition of a new SI second and its practical realization, offering a robust mathematical toolkit to minimize uncertainty in a multi-clock, asynchronous environment.