Imagine you're bouncing a ball in a room full of obstacles. In a perfectly closed room, most launch angles send the ball careening chaotically — but a few special angles produce eerily periodic orbits that repeat almost exactly. Now crack the door open and let in a draft. The question isn't whether the draft changes the ball's path — it does — but whether those special periodic orbits survive as recognizable patterns even with the breeze. This paper shows that if the draft blows symmetrically, the periodic orbits persist; if it blows asymmetrically, they're destroyed. The committed claim: quantum many-body scars — those anomalously periodic, non-thermalizing eigenstates first identified in the PXP model of Rydberg atom chains — survive Markovian open-system dissipation and form a full tower of Liouvillian eigenstates, provided the local jump operators respect the same sublattice symmetry as the Hamiltonian. The scarred eigenvalues separate cleanly from the bulk in both their imaginary parts (oscillation frequencies, equally spaced) and their real parts (decay rates). This is not a perturbative claim about weak dissipation — it's a structural result about the Liouvillian spectrum. The key physics: the PXP model on a chain has a sublattice (staggering) symmetry that partitions sites into even and odd. When the dissipative jump operators — here, spin-flip channels with rates γ₊ (down→up) and γ₋ (up→down) — are applied uniformly across all sites, they commute with this symmetry and the scar tower survives in the Liouvillian. The ratio γ₊/γ₋ controls whether the scarred states dominate long-time dynamics: when γ₊ > γ₋, the scarred eigenvalues have the smallest real parts (slowest decay), meaning fidelity oscillations from Néel-state initial conditions are enhanced and long-lived. Staggered jump operators — alternating rates on even/odd sublattices — further stabilize these oscillations. The destruction mechanism is equally clean. Replace the full Rydberg projection operator in the dissipators with a partial one that doesn't respect sublattice symmetry, and the scar tower collapses entirely. This isn't a quantitative degradation — it's a qualitative phase transition in the Liouvillian spectrum. The scarred states lose their anomalous overlap with Néel states and merge into the thermal bulk. The paper demonstrates this through exact diagonalization of the Liouvillian for chains up to ~20 sites, tracking overlaps, eigenvalue spacing, and fidelity revival dynamics. On the ladder: the prior art here is the closed-system PXP scar literature (Bernien et al. 2017 for experimental observation; Turner et al. 2018 for the scar identification) and early work on open-system scar physics (Buca et al. 2019; Hudomal et al. 2022 on dissipation-induced scarring). This paper advances the conversation by providing a clear symmetry criterion — not just showing scars can survive dissipation, but specifying exactly which dissipation channels preserve them and which destroy them. The γ₊/γ₋ asymmetry result is the new contribution: it identifies a tunable knob that controls scar dominance in the open system. Integrity is solid for a theory paper of this type. The results rest on exact diagonalization of the full Liouvillian, which is numerically trustworthy but limited to small system sizes (~20 sites due to the exponential Hilbert space growth of open-system density-matrix methods). No approximations beyond finite-size truncation. The authors show convergence across multiple system sizes and present both spectral and dynamical evidence. The limitation is inherent: we don't know if these results survive in the thermodynamic limit, and no experimental validation is presented. The obvious next experiment the authors didn't run: scaling to larger systems via tensor-network methods (time-dependent DMRG for open systems, or matrix product operator representations of the Liouvillian). The 20-site barrier is a hard wall for exact diagonalization. My read: this is a compute-budget limitation, not a hidden negative result. Tensor-network open-system methods exist but are technically demanding for tracking individual Liouvillian eigenvalues rather than just steady-state properties. The second missing piece is experimental: Rydberg atom arrays with controlled dissipation channels are now routine in multiple labs, and a direct test of the γ₊/γ₋ asymmetry prediction is feasible.
