The 3D simulator numerically diagonalizes an N-spin Ising Hamiltonian (up to 16×16) to track a many-body ground state through an adiabatic sweep. This 2D companion isolates the single mechanism underneath every avoided crossing in that bigger problem — the exactly-solvable Landau-Zener two-level model — and checks the live numeric integration against its closed-form textbook result.
H(t) = [ v·t C ]
[ C −v·t ] (ħ = 1)
E±(t) = ± sqrt((v·t)² + C²) adiabatic energies
Δ(t) = E+(t) − E−(t) = 2·sqrt((v·t)² + C²)
The state ψ(t) is evolved from t = −T (deep in the lower diabatic branch) to t = +T by real-time RK4 integration of i·dψ/dt = H(t)ψ — a genuinely independent 2×2 complex solve, not a projection of the 3D engine. The probability the system is still in its original diabatic branch at t = +T (a "diabatic transition", i.e. it failed to adiabatically follow the ground state through the crossing) has an exact closed form:
P_LZ = exp(−π C² / v)
- Slow sweep (small v) → P_LZ → 0: the adiabatic theorem holds, the state smoothly follows the ground state across the gap and ends up in the other diabatic branch — exactly the "success" regime of the 3D annealer.
- Fast sweep (large v) → P_LZ → 1: the sweep is too quick for the state to adjust (sudden approximation), so it is left behind in the original branch — the "diabatic transition" failure mode.
- Coupling C sets the minimum adiabatic gap (2C) at t = 0 — a bigger gap makes the crossing easier to follow adiabatically at any given sweep rate, mirroring how a larger Δmin helps a real annealer.
This exact formula is the v→0 / single-avoided-crossing limit of the same adiabatic theorem the 3D simulator's many-body Δmin obeys only approximately (T ≫ 1/Δmin²) — this page verifies that limit numerically rather than assuming it.