File size: 3,332 Bytes
d2486ac | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 | """
Mitigation (ZNE) panel -- Phase 3 of the dashboard refactor.
Visualizes run_mitigation_sweep's result: how fidelity degrades across
noise scales, how the zero-noise-extrapolated estimate compares to the
raw noisy sample and the ideal circuit, and -- when predictive healing is
enabled -- makes the delta_preemp correction visible instead of leaving
it a silent internal number.
"""
import numpy as np
import matplotlib.pyplot as plt
from .plot_theme import C, MONO, _ax_style, _badge
def build_panel_mitigation(mitigation_res: dict, base_res: dict) -> plt.Figure:
"""mitigation_res: run_mitigation_sweep's return dict.
base_res: the plain run_simulation() result at the base noise scale
(only used for n_qubits/dominant-state formatting fallback -- the
sweep's own prob_per_scale[0] is the actual base-noise sample)."""
fig, (ax_fid, ax_prob) = plt.subplots(2, 1, figsize=(16, 12), facecolor=C['bg'])
noise_factors = mitigation_res['noise_factors']
fidelity_per_scale = mitigation_res['fidelity_per_scale']
fidelity_zne = mitigation_res['fidelity_zne']
# ROW 1 — fidelity per noise scale, with ZNE-corrected / ideal reference lines
_ax_style(ax_fid, 'Zero-Noise Extrapolation — Fidelity vs Noise Scale',
'Noise scale factor (× base p)', 'Fidelity')
x = np.arange(len(noise_factors))
ax_fid.bar(x, fidelity_per_scale, color=C['noise'], alpha=0.75, width=0.5,
label='raw noisy (measured)')
ax_fid.set_xticks(x)
ax_fid.set_xticklabels([f'{f:g}×' for f in noise_factors])
ax_fid.axhline(1.0, color=C['accent'], lw=1.0, ls='--', alpha=0.7,
label='ideal (F=1)')
ax_fid.axhline(fidelity_zne, color=C['title'], lw=1.4, ls='-', alpha=0.9,
label='ZNE-corrected')
ax_fid.legend(loc='lower left', fontsize=8, framealpha=0.15,
labelcolor=C['label'])
ax_fid.set_ylim(0.0, max(1.05, fidelity_zne * 1.1, max(fidelity_per_scale) * 1.1))
healing_txt = (
f"healing: ON (Δpre_emp={mitigation_res['delta_preemp']:.4f})"
if mitigation_res['healing_enabled'] else "healing: off (plain ZNE)"
)
_badge(ax_fid, f"F_ZNE={fidelity_zne:.6f} · {healing_txt}", C['title'])
# ROW 2 — top-K basis-state probability overlay: ideal / raw noisy (base
# scale) / ZNE-corrected (same dim_vis truncation idea as build_panel_mosaico)
prob_ideal = np.asarray(mitigation_res['prob_ideal'])
prob_noisy_base = np.asarray(mitigation_res['prob_per_scale'][0])
prob_zne = np.asarray(mitigation_res['prob_zne'])
dim_vis = min(1024, len(prob_ideal))
_ax_style(ax_prob, 'Probability Distribution — Ideal vs Raw vs ZNE-corrected',
'|n⟩ computational basis (truncated)', 'P(|n⟩)')
xs = np.arange(dim_vis)
ax_prob.plot(xs, prob_ideal[:dim_vis], color=C['accent'], lw=1.2, alpha=0.85,
label='ideal')
ax_prob.plot(xs, prob_noisy_base[:dim_vis], color=C['noise'], lw=1.0, alpha=0.6,
label='raw noisy (1×)')
ax_prob.plot(xs, prob_zne[:dim_vis], color=C['title'], lw=1.2, alpha=0.9,
ls='--', label='ZNE-corrected')
ax_prob.legend(loc='upper right', fontsize=8, framealpha=0.15,
labelcolor=C['label'])
fig.tight_layout()
return fig
|