Multi-product Zeno effect achieving higher order convergence rates

Technical Summary: Multi-product Zeno effect achieving higher order convergence rates

Problem Statement
The quantum Zeno effect is a fundamental mechanism for implementing effective dynamics in projected Hamiltonian and Lindbladian systems. It approximates a target projected evolution by interleaving system dynamics (governed by a generator $L$) with frequent quantum operations (represented by an operator $M$) associated with a desired subspace. While the standard Zeno limit converges to the projected dynamics, the known convergence rate is typically limited to order $O(1/n)$, where $n$ is the number of interleaving steps. This stands in contrast to the Trotter product formula, which has been improved to arbitrary high-order convergence $O(1/n^K)$ via multi-product formulas and Suzuki expansions. However, existing high-order techniques for Trotterization rely on the Baker-Campbell-Hausdorff formula, which is not directly applicable to the Zeno effect due to the non-commutative nature of the projection and the specific structure of the Zeno sequence. Consequently, there is a lack of efficient schemes to achieve high-order convergence for Zeno-like expectation values, limiting the precision of simulations for decoupling, error correction, and state engineering.

Methodology
The paper develops a multi-product scheme to improve the convergence rate of the quantum Zeno effect. The methodology proceeds in three main stages:

1. Higher-Order Error Expansion: The authors first establish a higher-order expansion for the error of the Zeno sequence.

   • For the projective Zeno effect (where $M=P$ is a projection), they utilize a modified Chernoff $\sqrt{n}$-Lemma and a modified Dunford-Segal approximation. These tools allow for an explicit integral representation of the error terms, proving that the difference between the discrete Zeno sequence $(Pe^{t/n L})^n$ and the continuous projected evolution $e^{tPLP}P$ can be expanded in powers of $1/n$.
   • For the generalized Zeno effect (where $M$ is a general quantum operation with peripheral eigenvalues on the unit circle), the authors employ holomorphic functional calculus. They analyze the perturbed eigenprojections of the operator $M e^{t/n L}$ using a spectral gap assumption. This allows them to extend the higher-order expansion to the general case, showing that the error admits an expansion of the form $\sum_{k=1}^K n^{-k} E_{k,j} + O(n^{-(K+1)})$.

2. Multi-Product Construction: Based on the derived error expansion, the authors construct a multi-product formula. This involves summing $K+1$ parallel Zeno sequences, each executed with a different number of steps $n_\ell = m_\ell n$ (where $m_\ell$ are integers) and weighted by coefficients $c_\ell$.

   • The coefficients $c_\ell$ and step multipliers $m_\ell$ are chosen to satisfy a system of linear equations (a Vandermonde-type system) such that the lower-order error terms ($1/n, \dots, 1/n^K$) cancel out.
   • The system requires $\sum c_\ell = 1$ and $\sum c_\ell m_\ell^{-k} = 0$ for $k=1, \dots, K$.

3. Conditioning and Stability: The paper addresses the issue of numerical stability. Using an equidistant grid ($m_\ell = \ell$) leads to coefficients that grow exponentially with $K$, limiting scalability. To overcome this, the authors adopt a strategy from Hamiltonian simulation literature, selecting $m_\ell$ based on an approximate Chebyshev grid. This approach ensures the sum of the absolute values of the coefficients grows only logarithmically with $K$ ($O(\log K)$), making the scheme numerically well-conditioned.

Key Contributions and Results

• Arbitrary High-Order Convergence: The primary result is the proof that the convergence rate of the quantum Zeno effect can be improved from $O(1/n)$ to $O(1/n^{K+1})$ for any $K \in \mathbb{N}$. This is achieved by post-processing the results of $K+1$ parallel Zeno sequences.
• General Applicability: The framework applies to both bounded and unbounded generators (under appropriate domain assumptions) and covers both projective and generalized Zeno effects (including systems with periodic peripheral eigenvalues).
• Efficient Post-Processing: The method yields an improved approximation for Zeno-like expectation values. Specifically, the error in the expectation value of an observable $O$ scales as $O(1/n^{K+1})$.
• Scalability: By utilizing Chebyshev-like grids, the authors demonstrate that the overhead (sum of absolute coefficients) scales logarithmically with the order $K$, avoiding the exponential blow-up associated with equidistant grids.
• Applications: The paper illustrates the method with two specific examples:
  1. Bang-Bang Decoupling: Improving the convergence of dynamical decoupling schemes used to isolate quantum systems from their environment.
  2. Bosonic Cat Code: Enhancing the engineering of the $X(\theta)$ gate in bosonic cat codes by improving the convergence of the Zeno dynamics used to generate the gate.

Significance and Claims
The paper claims that this work provides a fundamental improvement in the efficiency of simulating projected quantum dynamics. By bridging the gap between the low-order convergence of the Zeno effect and the high-order capabilities of multi-product Trotter formulas, the method offers a pathway to more accurate simulations of open quantum systems, error correction protocols, and state engineering without requiring a prohibitive increase in computational resources (the number of parallel sequences grows only linearly with $K$).

The authors note that while the method is general, it currently applies to single or few-mode systems; extending it to multi-mode dissipative systems remains an open problem due to challenges with information propagation bounds (Lieb-Robinson bounds) in such systems. Furthermore, the work leaves open the question of whether the time-dependence of the error bounds (currently quadratic in $t$) can be improved and suggests future directions involving time-dependent systems and tighter error bounds for generalized Zeno effects.