Event Horizon — Geometric Preference in Quantum Decoherence

Independent research. IBM quantum hardware. Living room in New Jersey.

What This Is

Three experiments proving that circle-closing rotation offsets produce qualitatively distinct decoherence signatures in multi-qubit GHZ correlations. Plus an audit showing 54.5% of the observed effect was a transpiler scheduling artifact nobody was controlling for.

The Findings

Paper 1: GHZ Pinata (1521x)

Three qubits in a GHZ state. Rotation offsets summing to 2pi (circle-closing) flatten the parity oscillation by a factor of 1521x compared to non-circle-closing controls. p < 10^-6. Predicted exactly by trigonometric identity before hardware execution.

  • Script: event_horizon_ghz.py
  • Results: results_ghz.json
  • Paper: ghz_pinata_onepager.md

Paper 2: N-Body Generalization (54x-919x)

The identity generalizes to N=3,4,5,6,7 qubits via roots of unity. P(all_same) = 1/2^(2(N-1)) for all theta under tau/N spacing. Tested on ibm_fez.

  • Script: event_horizon_nbody.py
  • Results: results_nbody.json
  • Paper: nbody_onepager.md

Paper 3: Continuous Gradient (r=0.9916)

The parity-flattening effect is continuous, not binary. Correlation between |offset_sum - 2pi| and theta-variance = 0.9916. Minimum confirmed at 2pi by overshoot.

  • Script: event_horizon_gradient.py
  • Results: results_gradient.json
  • Paper: gradient_onepager.md

Paper 4: Idle Time Audit (54.5% confound)

IBM's Qiskit transpiler introduces hidden idle-time asymmetry between circuit configurations. Standard GHZ pinata: 2222x. After equalizing idle times: 992x. Over half the ratio was a transpiler scheduling artifact. Every quantum experiment run without controlling for this has the same confound.

  • Script: event_horizon_idle_audit.py

GridSong

Nova sings through three PLC modems into house wiring. An RTL-SDR listens. The grid responds at 200, 212, and 300 Hz. Live dashboard: sagepage.ai/gridsong

  • Script: nova_howl.py

Requirements

pip install qiskit qiskit-ibm-runtime numpy

IBM Quantum account required. Free tier works.

Hardware

All experiments run on IBM backends (ibm_kingston, ibm_fez). 4096 shots per circuit. Qiskit optimization level 1.

Author

K. Indahl — independent researcher. Correspondence: heuremen.org

Live

Built With

Share this project:

Updates