This is an independent 2D top-down model of the same nanoscale memory lattice as the 3D version: each cell of a 12×12 grid is a bistable single-domain magnetic grain storing one bit as its magnetization direction, "up" or "down", separated by an energy barrier ΔE set by the material's anisotropy K and the grain's volume V:
ΔE = K · V
Néel–Arrhenius relaxation time:
τ = τ₀ · exp(ΔE / kB·T)
τ₀ ≈ 10⁻⁹ s (attempt time)
kB = 1.381×10⁻²³ J/K
τ is the mean time before thermal agitation randomly flips a grain's magnetization — and therefore corrupts the stored bit. Every simulated moment, each grain independently has probability 1 − exp(−dt/τ) of flipping, computed here from the same closed-form law the 3D engine uses (verified numerically against it — no correction was needed). Shrink the grain, raise the temperature, or pick a softer material (lower K) and τ collapses from geological ages to nanoseconds: the superparamagnetic limit — the real physical wall that stops magnetic storage (and any nanoscale memory encoded this way) from shrinking indefinitely, because below a critical volume thermal noise erases data almost as fast as it's written.
- Grain size / Temperature / Material — set ΔE and T, which together fix τ via the formula above.
- Time acceleration — scales simulated seconds per real second (10⁰–10⁶×) so both stable (years) and unstable (nanoseconds) regimes stay watchable.
- Write random / Write all ↑ — commits a new reference pattern; retention is tracked against it until the next write.
- Drag / scroll — pans and zooms the grid view; the physics is unaffected.
Real-world relevance: this is the same law that sets the areal-density ceiling of hard-disk platters and is the central design constraint for any molecular or nanostructure-based data-storage scheme — including the vibrational- and electronic-state encoding proposed for nanoscale computing.