The Time-Evolution Operator
How the unitary e^(-iHt) propagates quantum states, and why its structure sets the shape of every simulation algorithm.
From Schrodinger to the propagator
The time-dependent Schrodinger equation i d/dt |psi> = H|psi> has, for a constant H, the formal solution |psi(t)> = e^(-iHt)|psi(0)>. The operator U(t) = e^(-iHt) is the propagator. It is unitary because H is Hermitian, which conserves probability:
Properties that matter
- Unitarity: U(t) U(t)^dagger = I, so norms are preserved.
- Group law: U(t1) U(t2) = U(t1 + t2), evolution composes additively in time.
- Eigenstructure: if H|E> = E|E>, then U(t)|E> = e^(-iEt)|E>, a pure phase.
- Reversibility: U(-t) = U(t)^dagger undoes the evolution.
The eigenphase view
Because eigenstates pick up phases e^(-iEt), the propagator encodes the spectrum of H in the phases it applies. This is the entire basis of quantum phase estimation: run controlled evolution, then read the accumulated phase to recover an eigenvalue E.
Time-dependent Hamiltonians
When H = H(t) varies in time, the simple exponential fails because H at different times need not commute. The correct object is the time-ordered exponential U(t) = T exp(-i integral_0^t H(s) ds). Simulation algorithms handle this by slicing time into short intervals over which H is nearly constant.
Why algorithms decompose it
Directly implementing e^(-iHt) as one gate is not possible on a universal machine; the propagator must be built from a fixed gate set. Every simulation method is a recipe for approximating U(t) to accuracy epsilon using elementary operations. The difficulty is that when H = A + B with A and B non-commuting, e^(-i(A+B)t) is not e^(-iAt) e^(-iBt); the mismatch is the source of Trotter error and motivates higher-order and post-Trotter techniques.
Understanding U(t) as a phase-applying, composable, reversible unitary is the conceptual anchor for dynamics simulation, spectroscopy, and eigenvalue estimation alike.