This view keeps a real array of N signed amplitudes, one per basis state, starting uniform at 1/√N. It is plotted as a radar/polygon chart instead of a bar chart: each of the N spokes is one state, and the vertex distance from the centre is that state's current amplitude magnitude.
Oracle: a[target] = -a[target]
Diffusion: mean = average(a)
a[i] = 2*mean - a[i] for every i
- Every "Grover step" button press really recomputes the mean of the current array and reflects every entry about it — the same inversion-about-the-mean operator used in the 3D bar-chart version, just rendered as a growing polygon spike.
- The dashed circle marks the mean amplitude magnitude each iteration is computed from, so you can see why the diffusion step pushes the below-mean marked state up and the above-mean others down.
- The classical scan track below runs an actual linear search over a freshly shuffled order of the N items, incrementing a real comparison counter box by box until it reaches the target — not a static "O(N)" label.
Grover needs about π/4·√N iterations regardless of N; classical linear search needs N/2 comparisons on average and up to N in the worst case. Reset re-shuffles the classical order and re-uniformises the quantum register.