Has Quantum Advantage Been Achieved? Decoding the "Weak-Noise" Frontier

Executive Summary

TL;DR: Despite skepticism within the physics community, this paper argues that quantum advantage is no longer a theoretical "maybe"—it is a demonstrable fact. By analyzing the phase transition between "weak" and "strong" noise, the author shows that current quantum processors (70+ qubits) operate in a regime where classical simulation remains exponentially costly, even with imperfect fidelity.

Positioning: This is a landmark perspective piece that bridges the gap between Experimental Physics (building the hardware) and Theoretical Computer Science (proving the hardness). It moves the goalposts from simply "beating a supercomputer" to "building a verifiable, fault-tolerant advantage."

The "Spoofing" Crisis: Why Skepticism Persists

The skepticism surrounding quantum advantage often stems from the 2019 Google "Sycamore" experiment. Shortly after the claim of quantum supremacy, classical algorithms were developed that could simulate the 53-qubit circuit by exploiting the low fidelity of the noisy hardware. This led many to ask: Is the task only hard because we haven't found the right classical trick yet?

The author identifies the core tension: Fidelity vs. Verifiability. In the NISQ era, we cannot run error correction, so we must settle for "finite-fidelity RCS." The question is whether a tiny signal (e.g., 0.1% fidelity) is enough to maintain a scaling advantage over classical machines.

Methodology: The Phase Transition in Noise

The most significant technical insight here is the identification of a sharp phase transition in the Linear Cross-Entropy Benchmark (XEB).

1. The Weak-Noise Regime

The author proposes a scaling limit where local noise $ϵ$ scales as $c_A/n$. In this regime:

• XEB is a reliable proxy for Fidelity: The noise acts as "global white noise," preserving the structure of the quantum state's interference pattern.
• Classical Hardness Persists: Even with a fidelity of $0.1\%$, the classical cost to reproduce the samples remains exponential.

2. The Strong-Noise Regime

If noise exceeds the threshold $c_A/n$, the XEB "decouples" from fidelity. Here, classical "spoofers" can achieve high XEB scores by breaking the circuit into smaller, simulatable chunks. This is where classical computers win.

Figure 1: The decay rate of XEB signifies the transition. Below the horizontal line (weak noise), the quantum signal is preserved across the entire system.

Evidence: The State of the Art

The paper summarizes the progress from Google (superconducting), USTC (superconducting/photonic), and Quantinuum (trapped-ion).

Figure 2: Plot of achieved XEB/fidelities since 2019. Note how systems have scaled to 80+ qubits while maintaining signal.

The key takeaway from Figure 2 is that while 0.1% fidelity sounds low, it is vastly higher than the trivial baseline of $2^{-n}$ (which is $\approx 10^{-14}\%$ for these systems). This gap is the "Quantum Advantage" territory.

The Road to 100 Logical Qubits

The most forward-looking part of the paper defines the next milestone: Classically Verifiable Quantum Advantage. Currently, verifying a 70-qubit circuit requires a supercomputer. For a 100-qubit circuit, verification becomes impossible.

The author proposes three strategies for the next era:

1. Fault-Tolerant RCS: Running random circuits on encoded logical qubits to reach $>90\%$ fidelity.
2. Symmetry-Protected Verification: Using "Random Circuits with Symmetries" (like Graph States) where single-qubit measurements can efficiently confirm the global state's correctness.
3. Planted Secrets: Using IQP (Instantaneous Quantum Polynomial) circuits with hidden "peaks" in the distribution that a classical user can check without needing to simulate the entire state.

Critical Insights & Conclusion

The author makes a compelling philosophical point: If you believe in the discovery of the Higgs Boson—which required massive data post-processing and theoretical modeling—you should believe in Quantum Advantage. Both rely on "proxies of proxies" to reveal fundamental truths.

Limitations: The tasks achieved so far (RCS) are "useless" for practical applications like drug discovery. The transition to useful advantage will require moving from random sampling to structured algorithms, likely utilizing the IQP frameworks discussed in Part 3.

The Future: The "100 logical qubit" regime is the sweet spot. It is small enough to build in the near term but large enough to host cryptographic proofs of quantumness that will finally silence the skeptics.