Hopfield Network
Ising model → Associative Memory · N=1024 neurons · M=7 stored patterns
steps: 0
Stored Patterns
Click to recall →
Select a pattern to begin recall
Overlap m_μ with each stored pattern
m_μ = (1/N) Σ ξ_μi · s_i ∈ [-1, +1]
Noise Level
Flip fraction
20%
Update Speed
Neurons/frame
102
⚡ Random Initial State
⏭ Single Step (N/10)
⏸ Pause
Statistics
Energy E
—
ΔE/step
—
Best match
—
m_best
—
Total flips
0
Energy E(t)
—
Algorithm
h_i = Σ_μ ξ_μi · m_μ
s_i → sign(h_i)
E = -½ Σ_ij W_ij s_i s_j
O(M·N) per sweep