Dynamical Poles in Non-Hermitian Impurity Scattering

This paper demonstrates that in non-Hermitian impurity scattering, late-time dynamics are governed by "dynamical poles" derived from the analytic continuation of the Green's function rather than static bound states, revealing a fundamental breakdown of the Hermitian correspondence where isolated exponential signals uniquely identify bound states.

The Gist

The Big Picture: When the Map Doesn't Match the Territory

Imagine you are a detective trying to solve a mystery in a city. In the old days (what physicists call Hermitian systems), the city was very predictable. If you wanted to know where a criminal (an "impurity" or defect) was hiding, you could just look at the city's static map (the spectrum or list of all possible locations).

In this old world, the rule was simple: Every hiding spot on the map showed up as a distinct signal later. If the map said "Criminal is at the library," you would eventually hear a distinct sound coming from the library. The map and the reality were perfectly synced.

This paper says: That rule is broken in the new world of "Non-Hermitian" systems.

In modern physics (like lasers, open quantum systems, or biological networks), energy leaks out or gets pumped in. The city is no longer static; it's a living, breathing, leaking organism. The authors discovered that in this new world, the map is a liar. You can have a hiding spot on the map that makes absolutely no sound later, and you can hear a loud sound coming from a place that isn't even on the map.

The Core Concept: The "Ghost" vs. The "Echo"

To understand this, we need to look at how the authors solved the mystery. They introduced a new tool called Dynamical Poles (DPs).

Think of the system as a giant, complex drum.

1. Static Bound States (The Map): These are the places where a drumstick could theoretically get stuck. In the old world, if the stick got stuck, the drum would ring with a specific, pure tone.
2. Dynamical Poles (The Echo): These are the places where the drum actually rings loudly in the real world, after you hit it and let it decay.

The Shocking Discovery:

• The Invisible Ghost: Sometimes, the drumstick gets stuck in a "Static Bound State" (it's on the map), but because of the way energy leaks out of the system, the drum stays silent. The ghost is there, but it's dynamically dark. You can't hear it.
• The Phantom Echo: Sometimes, the drumstick never gets stuck (it's not on the map), but the drum still rings with a loud, pure tone. This is a Dynamical Pole. It's a "phantom" frequency that only exists because of how the sound travels through the leaking system, not because of a physical trap.

The Analogy: The Leaky Bathtub

Imagine a bathtub with a drain (non-Hermitian loss) and a rubber duck (the impurity).

• The Old Way (Hermitian): If you drop the duck in, it sinks to a specific depth (a bound state). If you listen to the water, you hear a specific "glug" sound corresponding to that depth. The depth and the sound match perfectly.
• The New Way (Non-Hermitian): The water is draining fast, and the tub is vibrating strangely.
  • Scenario A (The Dark Duck): You drop the duck, and it sinks to a specific depth. But because the water is draining so fast and swirling, the "glug" sound is completely drowned out by the rushing water. The duck is there, but it's invisible to your ears.
  • Scenario B (The Phantom Sound): You drop the duck, and it floats on the surface (no deep sink). But, due to the weird vibration of the draining water, the tub starts humming a specific, loud note that doesn't match the surface level. This note is the Dynamical Pole. It's a real sound, but it doesn't come from a "trapped" duck; it comes from the interaction between the duck and the draining water.

The "Analytic Continuation" Magic Trick

How did they find these phantom sounds?

In math, there's a trick called Analytic Continuation. Imagine you have a map of a city, but the map only shows the streets above ground. To find the "Dynamical Poles," the authors had to use a special X-ray vision to look underground (into the complex mathematical plane).

• The Static Map: Only shows the streets above ground.
• The Analytic Continuation: Reveals the underground tunnels.
• The Result: The "Dynamical Poles" are the exits to the underground tunnels. Even if the street above (the static bound state) is blocked or non-existent, the tunnel exit (the DP) might be wide open, letting a loud signal through.

Why Does This Matter?

This changes how we understand the universe, from lasers to quantum computers.

1. Spectroscopy is Tricky: If you try to measure a system by looking at its "map" (energy levels), you might miss the most important signals. You might think a system is quiet because you don't see a trap, but it's actually screaming with "phantom" frequencies.
2. Time is Key: The authors show that time is the ultimate truth-teller. If you wait long enough and listen to how the system decays, you will hear the Dynamical Poles. The "static map" is just a snapshot; the "time signal" is the movie.
3. The Background Noise: The rest of the signal (the part that isn't a pure tone) is like the background hiss of a radio. In this new world, that hiss comes from "complex edges" (mathematical boundaries) that act like the edge of a cliff in a foggy landscape.

Summary in One Sentence

In systems where energy leaks or flows, the "hiding spots" you see on a static map don't necessarily make a sound, and the sounds you hear don't necessarily come from a hiding spot; to understand the future, you must stop looking at the map and start listening to the echo.

Technical Summary

1. Problem Statement

In Hermitian quantum mechanics, there is a fundamental correspondence between the static energy spectrum and long-time scattering dynamics:

• Static Bound States correspond to isolated poles in the Green's function.
• Late-time Dynamics: Each bound state manifests as a discrete exponential decay (or oscillation) in the time-domain signal. The continuous spectrum contributes only as a subleading algebraic background.

The authors investigate whether this correspondence holds in non-Hermitian systems (systems with gain, loss, or particle decay). In non-Hermitian lattices, eigenvalues are complex, and dynamics are non-unitary. While recent work has identified static bound states and scattering states in these systems, it remains unclear if these static eigenstates govern the long-time scattering dynamics in the same way as in Hermitian systems. Specifically, the paper asks: What determines the late-time signal in non-Hermitian impurity scattering?

2. Methodology

The authors employ a rigorous analytic continuation approach to the Green's function to resolve the time-domain dynamics.

• Model Systems:
  • Hatano-Nelson Chain: A 1D non-Hermitian model with asymmetric hopping ($J \neq 1$) and uniform loss ($-i\kappa$), serving as a primary counterexample to static intuition.
  • General 1D Lattices: Models with finite-range hopping (e.g., next-nearest-neighbor).
  • Boundary Conditions: Analysis covers both Periodic Boundary Conditions (PBC) and Open Boundary Conditions (OBC).
• Theoretical Framework:
  • The time-evolution amplitude $M(t)$ is expressed as a frequency integral of the Green's function $G(\omega)$.
  • Instead of using the standard Green's function defined on the Brillouin Zone (BZ), the authors construct an analytically continued Green's function, denoted $\tilde{g}_0(\omega)$.
  • Contour Deformation: For $t > 0$, the integration contour is deformed into the lower half of the complex frequency plane ($\text{Im } \omega < 0$).
  • Root Selection Logic: The construction of $\tilde{g}_0(\omega)$ involves tracking the roots of the dispersion relation $\omega - h(z) = 0$ (where $z=e^{ik}$). The function is defined by summing residues of roots that lie inside the unit circle at a reference point in the upper half-plane and analytically continuing them.
  • Singularity Analysis: The authors identify two types of singularities in the continued function that dominate the late-time signal:
    1. Dynamical Poles (DPs): Zeros of the denominator $1 - \lambda \tilde{g}_0(\omega) = 0$.
    2. Branch Points: Points where roots coalesce (multiple roots) and exchange between "inside" and "outside" the unit circle, creating branch cuts.

3. Key Contributions and Results

A. The Concept of Dynamical Poles (DPs)

The authors introduce Dynamical Poles (DPs) as the true generators of late-time exponentials in non-Hermitian systems.

• Definition: DPs are poles of the analytically continued Green's function $\tilde{g}_0(\omega)$, not necessarily the static Green's function $g_0(\omega)$.
• Mechanism: A DP arises when $1 - \lambda \tilde{g}_0(\omega) = 0$. Because $\tilde{g}_0$ differs from $g_0$ in the lower half-plane (the "mismatch domain"), DPs can exist even where no static bound state exists.

B. Breakdown of Static-Dynamical Correspondence

The paper establishes a two-sided mismatch between static spectra and dynamical signals:

1. Dynamical Poles without Bound States: A DP can appear in the lower half-plane where the static Green's function has no pole (e.g., inside a point gap where $g_0(\omega) = 0$). This results in a late-time exponential signal with no spectral origin in the static Hamiltonian.
2. Dynamically Dark Bound States: A static bound state (a pole of $g_0(\omega)$) may lie in a region where $g_0(\omega) \neq \tilde{g}_0(\omega)$. If this state does not satisfy $1 - \lambda \tilde{g}_0(\omega) = 0$, it contributes to the eigenvalue spectrum but leaves no trace in the time-domain signal. These are termed dynamically dark states.

C. Complex Continuum Edges

The continuous spectrum in non-Hermitian systems is organized by branch points of $\tilde{g}_0(\omega)$.

• These branch points act as "non-Hermitian continuum edges."
• They generate an incoherent background signal that decays algebraically (e.g., $t^{-3/2}$ in 1D) multiplied by an exponential factor determined by the complex branch point energy.
• The authors provide a general algorithm to identify these branch points: they correspond to energies where a root collision causes an exchange between a root selected for the residue sum and one that is not.

D. Non-Perturbative Reshaping of Continuum Response

The presence of an impurity does not just add poles; it fundamentally alters the continuum edge behavior.

• Unperturbed: The continuum edge contribution scales as $t^{-1/2} e^{-i\omega_b t}$.
• Impurity-Dressed: The scattering amplitude acquires a factor $[1 - \lambda \tilde{g}_0(\omega)]^{-1}$. Near a branch point, this suppresses the singularity, changing the decay law to $t^{-3/2} e^{-i\omega_b t}$. This is a genuinely non-perturbative effect.

E. Generalization to Higher Dimensions and OBC

• The framework extends to higher dimensions ($d \geq 2$), where the singularity structure of $\tilde{g}_0$ follows standard band-edge scaling laws (e.g., logarithmic in $d=2$, square-root in $d=1$), but with complex energies.
• The results hold for Open Boundary Conditions (OBC), where the "mismatch domain" is defined by the OBC spectrum and the branch cuts of the continued Green's function.

4. Significance and Implications

• Redefining Scattering Theory: The paper overturns the assumption that static eigenvalues dictate long-time dynamics in non-Hermitian systems. It proves that the real-time analytic structure of the Green's function is the organizing principle.
• Experimental Relevance: This explains why spectroscopic measurements (probing static eigenvalues) and time-resolved measurements (probing late-time response) in non-Hermitian platforms (photonic lattices, cold atoms, quantum dots) may yield discrepant sets of discrete frequencies.
• New Physical Phenomena:
  • Dynamically Dark States: Bound states that are "invisible" to time-domain probes.
  • Spectral-less Dynamics: Exponential signals that appear without any corresponding static bound state.
• Connection to Saddle Points: The authors show that the branch points of the continued Green's function are equivalent to the Relevant Saddle Points (RSPs) used in non-Hermitian edge dynamics, providing an alternative, algebraic method to locate these critical points without explicit gradient flow analysis.

In conclusion, the paper establishes that in non-Hermitian impurity scattering, the late-time signal is governed by Dynamical Poles and Complex Branch Points derived from the analytically continued Green's function, leading to a generic decoupling between the static spectrum and the dynamical response.