Watching the Horizon Break Entanglement: Measurement-Induced Collapse on a 156-Qubit Quantum Processor

We investigate measurement-induced effects on trans-horizon correlations in dynamic quantum circuits executed on IBM's 156-qubit superconducting processor (ibm_fez, Heron r2 architecture). We engineer a 40-qubit spin chain with a spatially-varying XX + YY coupling profile that creates an analog horizon at position x _h = 20, where coupling strength reaches a minimum. We define a correlation contrast metric C = ⟨σ ^x iσ ^x j⟩ _horizon − ⟨σ ^x σ ^x ⟩ _exterior and compare three conditions: (i) unitary Trotter evolution (baseline), (ii) mid-circuit measurement with reset on qubits near the horizon, and (iii) control measurement on distant qubits. Each circuit was executed with 16,384 shots. The near-horizon measurement produces a (56.5 ± 1.9)% reduction in correlation contrast (95% CI from n = 5000 bootstrap resamples), while the control preserves (87.8 ± 2.1)% of baseline. Two ablation experiments constrain interpretation: removing the reset operation yields only a (2.6 ± 1.2) percentage point change, indicating that projective measurement—not state reset—drives the effect. Crucially, moving the measured subset 4 links away from the horizon collapses the reduction to (6.5 ± 2.3)%, yielding an 8.2 ± 1.4 ratio that demonstrates strong spatial localization. We additionally performed a flat-spacetime control with uniform coupling J(x) = J _max on ibm_torino (Heron r1, 133 qubits), obtaining a baseline contrast that collapses to a small residual near zero (C = − 0.015, 95% CI [− 0.029, − 0.002]), confirming that the spatial structure arises from the engineered coupling profile rather than generic measurement artifacts. These results establish that mid-circuit measurement near an engineered coupling minimum preferentially disrupts correlations across that boundary, a signature qualitatively consistent with information-theoretic models of horizon physics.