Experimental Results Summary

​

Abstract

I report extended experimental observations from a quantum circuit simulator using neural network backends. Three independently trained models continue to produce identical results across standard quantum algorithms. A phase coherence test suite passed 22 of 22 tests. A variational quantum eigensolver recovered 100% of correlation energy for molecular hydrogen. New experiments extend these findings significantly. Stark effect calculations yield electric polarizability of 2.750 a₀³, matching exact diagonalization with zero error to measurable precision. The energy response shows perfect symmetry |E(+F) - E(-F)| at machine precision across all field values. QED corrections approximate Lamb shift and anomalous magnetic moment with expected perturbative accuracy. Polyatomic molecules including H2O, NH3, and CH4 are processed through the same pipeline. A visualization framework renders quantum state evolution in publication-quality figures that exactly match numerical logs. These results suggest that neural physics backends preserve not only quantum mechanical constraints but also the mathematical structure needed for response properties and extended physical calculations.

​

Reproducibility Confirmation

The baseline quantum algorithm suite continues to produce identical results across all three backends.

​

Standard Quantum Algorithms

​

Phase Coherence Tests

​

Molecular Hydrogen VQE

​

H₂ Ground State Calculation

============================================================
  VQE Result: H2  [openfermion_jw]
============================================================
  Qubits: 4
  Parameters: 5
────────────────────────────────────────────────────────────
  HF energy  : -1.11699900 Ha
  VQE energy : -1.13730604 Ha
  FCI energy : -1.13730604 Ha
────────────────────────────────────────────────────────────
  |VQE-FCI|  : 1.31e-11 Ha
  Correlation: 100.0%
============================================================

​

Electric Polarizability (Stark Effect)

​

Field Sweep Results

​

Symmetry Verification

​

Polarizability

​

QED Corrections

​

Lamb Shift

​

Anomalous Magnetic Moment

• Calculated: 2.002326393574
• Experimental: 2.002319304363

​

Polyatomic Molecules

​

H₂O (Water)

​

NH₃ and CH₄

• Electrons: 10, Orbitals: 8, Qubits: 16
• Processed successfully

• Electrons: 10, Orbitals: 9, Qubits: 18
• Processed successfully

​

Visualization Framework

Generated figures:

• bell_state.png: Probability bars, Bloch spheres, phase space, entropy curve
• ghz_3q.png: Same components for 3-qubit GHZ state
• qft_3q.png: Uniform distribution verification with all visualizations
• grover_3q_m5.png: Amplification pattern with 94.53% success probability

All visualizations match numerical logs exactly. Backend comparison plots show all three backends producing identical results.

​

Discussion

The polarizability result is the most significant new finding. A calculation that couples VQE to an external field and fits a quadratic response should be sensitive to any noise or asymmetry in the underlying simulation. The neural backends are not explicitly constrained to preserve field-response properties. They are not told that positive and negative fields should produce symmetric energy shifts. They are not told that the polarizability should match exact diagonalization. Yet they do all of this correctly. The perfect symmetry |E(+F) - E(-F)| at machine precision tells me that the backends are not introducing spurious field-dependent artifacts. They are not breaking parity. They are not leaking information between positive and negative field directions. The zero-error polarizability tells me that the entire pipeline is internally consistent. The dipole operator construction, the Jordan-Wigner mapping, the VQE optimization, and the energy differencing all work together without introducing bias.

​

Limitations

Polarizability was tested only for H2 with the STO-3G basis. I do not know if larger molecules or better basis sets preserve this accuracy. The QED calculations are approximate and do not use the neural backends directly. Polyatomic molecules were processed through PySCF but not optimized with VQE due to classical FCI cost scaling. The visualization system produces static figures; real-time animation remains unexplored. GPU execution was not tested.

​

Conclusion

I have presented extended experimental observations from a neural-network-based quantum simulator. Three independently trained backends produce identical results across standard algorithms, recover 100% correlation energy for H2, and correctly handle external field coupling to yield exact polarizability. The energy response shows perfect symmetry at machine precision. The framework supports QED corrections, polyatomic molecules, and direct visualization of quantum state evolution. The significance depends on whether these observations generalize. I offer this as a data point: neural physics backends can preserve not only quantum mechanical constraints but also the mathematical structure needed for response properties. That is the new boundary.

​

Related Documentation

• Constraint Preservation
• Phase Coherence
• Polarizability
• Extended Capabilities
• Benchmarks