Limits of the non-Hermitian description of decay models

This paper establishes that while non-Hermitian and Lindblad decay dynamics are equivalent in the highest particle subspace, the accuracy of non-Hermitian descriptions is strictly limited to weak-coupling and singular-coupling regimes, thereby questioning their validity for more complex systems and proving that exceptional points cannot occur in the weak-coupling limit for nondegenerate Hamiltonians.

The Gist

Imagine you are watching a leaky bucket. Water (representing energy or particles) is slowly draining out into a vast ocean (the "bath"). Physicists have two main ways to describe this leaking process, and this paper is essentially a detective story checking if those two ways tell the same story.

Here is the breakdown of the paper using simple analogies:

1. The Two Ways to Describe a Leak

When a quantum system (like an atom or a tiny circuit) loses energy to its surroundings, physicists usually use one of two mathematical "maps":

• Map A: The Lindblad Map (The Full Story). This is the most accurate, but also the most complicated map. It accounts for the water leaking out and the fact that sometimes, if you look closely, the water might splash back in or interact in complex ways. It's like tracking every single water droplet and every ripple in the ocean.
• Map B: The Non-Hermitian Map (The Shortcut). This is a popular, simplified map. It assumes the water only leaks out and never comes back. It treats the system as if it's slowly fading away, like a candle burning down. It's much easier to calculate, but the big question is: Is this shortcut actually accurate?

2. The Big Discovery: The Shortcut Only Works in Extreme Cases

The authors of this paper asked: "When is it safe to use the simple 'leaking candle' map instead of the complex 'full ocean' map?"

They tested this on a very simple model: a two-room house where water can leak out of either room into two different oceans. They ran the numbers to see if the simple map matched the complex reality.

The Result: The simple map (Non-Hermitian) is only accurate in two very specific, extreme situations:

1. The "Tiny Leak" Limit (Weak Coupling): When the hole in the bucket is so tiny that the water drips out incredibly slowly compared to how fast the water sloshes around inside the bucket.
2. The "Sudden Splash" Limit (Singular Coupling): When the connection to the ocean is so strong and fast that the water leaves instantly, behaving in a very specific, predictable way.

The Catch: In the "middle ground"—which is where most real-world experiments actually happen—the simple map is wrong. It fails to capture the true behavior of the system.

Analogy: Imagine trying to predict the path of a leaf falling in a river.

• If the river is a calm, slow-moving stream (Weak Coupling), you can just draw a straight line down.
• If the river is a massive waterfall (Singular Coupling), the leaf falls straight down instantly.
• But if the river has rapids and whirlpools (the "middle ground"), a straight line is a terrible prediction. The authors found that most real systems are like the river with rapids, so the simple "straight line" math often fails.

3. The "Ghost" States (Exceptional Points)

The paper also talks about something called Exceptional Points. In the world of complex math, these are like "ghost states" where two different behaviors of a system merge into one. It's like two different musical notes suddenly becoming the exact same pitch.

Scientists love hunting for these ghost states because they can lead to super-sensitive sensors or new types of lasers.

• The Finding: The authors proved that if you are in the "Tiny Leak" (Weak Coupling) zone, you will never find these ghost states. They simply cannot exist there.
• Why it matters: If you are an experimentalist trying to build a device to find these ghost points, don't bother trying to do it with very weak connections. You need to crank up the interaction to find them.

4. The Takeaway for Everyone

This paper is a reality check for the physics community.

For years, many scientists have been using the "Non-Hermitian" shortcut to describe open quantum systems because it's easy and elegant. This paper says: "Be careful."

While the shortcut works in two extreme corners of the universe, it is likely not a good description for the messy, complex systems we encounter in the real world. If you use this shortcut for a complicated system, you might be drawing a straight line through a river of rapids.

In summary:

• The Shortcut: Easy to use, but often wrong.
• The Reality: The shortcut only works when the system is either barely leaking or leaking instantly.
• The Lesson: For everything else, we need to use the harder, more complex math to get the right answer.

Technical Summary

1. Problem Statement

Open quantum systems are frequently modeled using two distinct theoretical frameworks:

1. Lindblad Master Equation: A rigorous approach describing the evolution of the density matrix $\rho_A$ including both coherent evolution and dissipative "quantum jumps" (gain/decay terms).
2. Non-Hermitian Hamiltonian ($H_{nh}$): A simplified approach where the system evolves under an effective non-Hermitian operator, $H_{nh} = H_A - \frac{i}{2}\sum \Gamma_j L_j^\dagger L_j$. This description is valid only if the "quantum jump" term (which moves probability out of the current subspace) can be ignored.

The Core Question: Under what conditions is the non-Hermitian description an accurate approximation of the full Lindblad dynamics? While it is known that the Lindblad equation is accurate in the weak-coupling and singular-coupling limits, it is unclear if the non-Hermitian reduction (ignoring jumps) holds in these same limits or elsewhere. Furthermore, the existence of Exceptional Points (EPs)—degeneracies where eigenvectors coalesce—is often studied using non-Hermitian models, but the validity of these models in specific coupling regimes is debated.

2. Methodology

The authors employ a rigorous analytical and numerical approach to test the validity of the non-Hermitian approximation:

• Decay Model Definition: They consider systems initially containing $N$ particles connected to empty baths. The total particle number is conserved in the system+bath, but particles decay from the system into the bath.
• Subspace Analysis: They focus on the highest particle subspace (where the system still has $N$ particles).
  • Result R1 (Proof): They prove that for Lindblad dynamics with only decay terms (annihilation operators), the quantum jump term vanishes strictly within the highest particle subspace. Consequently, the dynamics in this subspace are exactly equivalent to evolution under a non-Hermitian Hamiltonian $H_{nh}$.
• Diagnostic Metric ($R_{\theta, \phi}$): To determine if the dynamics outside the highest particle subspace (or the full evolution) can be described by a non-Hermitian operator, they propose an unbiased test.
  • They scan all possible initial states $|v\rangle$ in the $N$-particle subspace.
  • They calculate $R_{\theta, \phi} = \max_t |r_v(t)|$, where $r_v(t)$ is the probability of the system transitioning to an orthogonal state within the same particle number sector.
  • If $R = 0$, the state does not mix, implying the existence of eigenstates of $H_{nh}$ and validating the non-Hermitian description. If $R > 0$, mixing occurs, and the non-Hermitian description fails.
• Specific Model: They apply this to a two-site system ($H_A$) coupled to two semi-infinite chain baths ($H_B$) via coupling terms ($H_C$). They use Green's function methods to calculate exact time evolutions and compare them against the non-Hermitian approximation.

3. Key Contributions and Results

Result R1: Exact Equivalence in the Highest Subspace

The authors provide a formal proof that for decay-only Lindblad models, the dynamics restricted to the subspace with the maximum number of particles are governed exactly by a non-Hermitian Hamiltonian. This establishes a rigorous baseline: if eigenstates exist in this subspace, they are robust against mixing within that subspace.

Result R2: Limited Validity of Non-Hermitian Dynamics

By analyzing the two-site model, the authors demonstrate that the non-Hermitian description (where $R \to 0$) is only accurate in two specific asymptotic limits:

1. Weak Coupling Limit: Where the system-bath coupling $C$ is much smaller than the system and bath energy scales ($C/A \ll 1$).
2. Singular Coupling Limit: Where $C/A \to \infty$ and $B/A \to \infty$ such that $C^2/B$ remains constant.

Crucial Finding: In the intermediate parameter space (away from these two limits), the mixing parameter $R$ is finite. This means the exact dynamics involve mixing of states that cannot be captured by a simple non-Hermitian Hamiltonian. The authors argue that since even this simple model fails the non-Hermitian test outside these narrow limits, the general applicability of non-Hermitian descriptions for complex open systems is highly questionable.

Result R3: Constraints on Exceptional Points (EPs)

The authors investigate the conditions under which Exceptional Points (EPs) can occur.

• Weak Coupling: They prove that for a system with a non-degenerate Hamiltonian $H_A$, EPs cannot occur in the weak coupling limit. Perturbative arguments show that the eigenvectors of the non-Hermitian operator remain close to the orthogonal eigenvectors of $H_A$; small corrections cannot cause them to coalesce.
• Singular Coupling: Conversely, EPs can occur in the singular coupling limit. The authors derive a condition (Eq. 13 in the paper) where the relaxation time of the system becomes comparable to the intrinsic system timescale ($\tau_A \sim \tau_R$), allowing for EPs.

4. Significance and Implications

• Validity of Approximations: The paper challenges the widespread use of non-Hermitian Hamiltonians to model open quantum systems. It demonstrates that such descriptions are not generally valid but are restricted to specific asymptotic regimes (weak and singular coupling). For intermediate couplings, the full Lindblad dynamics (including quantum jumps) are necessary.
• Experimental Design: The findings regarding Exceptional Points are critical for experimentalists. The paper suggests that searching for EPs in the weak coupling regime is theoretically futile for systems with non-degenerate spectra. Experiments aiming to observe EPs must target the singular coupling regime or other specific parameter regions where the relaxation time matches the system's intrinsic timescale.
• Theoretical Framework: By establishing the exact equivalence between decay-only Lindblad dynamics and non-Hermitian dynamics in the highest particle subspace, the authors provide a clear criterion for when the "quantum jump" term can be safely ignored and when it is essential for the system's evolution.

Conclusion

Monkman and Berciu provide a rigorous proof that non-Hermitian descriptions of decay are exact only in the highest particle subspace and accurate for the full system only in the weak and singular coupling limits. Their work serves as a cautionary guide, indicating that non-Hermitian models may yield incorrect predictions for intermediate coupling strengths and that the search for Exceptional Points must be carefully targeted to regimes where system and bath timescales are comparable.