Energy–Time Uncertainty: Foundational Meaning

Energy–time uncertainty is a family of statements involving different energy and time quantities. It is not a universal rule allowing energy to be borrowed from nowhere. The canonical Energy–Time Uncertainty page supplies the derivations; this page separates the meanings needed in foundational arguments.

Dynamical time is not a random time operator

For a time-independent Hamiltonian and an observable $A$ with no explicit time dependence, Robertson’s inequality and the equation of motion yield

$$\Delta H\,\Delta A\ge \frac{\hbar}{2}\left|\frac{d\langle A\rangle}{dt}\right|.$$

When the derivative is nonzero, define the local change timescale $\tau_A=\Delta A/|d\langle A\rangle/dt|$. Then $\Delta H\,\tau_A\ge\hbar/2$. Here $\tau_A$ is a derived state-and-observable-dependent scale, not the standard deviation of a universal time observable. At a turning point it can diverge without implying that all evolution has ceased.

For a pure state evolving under a time-independent Hamiltonian, the Mandelstam–Tamm orthogonalization bound is

$$t_\perp\ge\frac{\pi\hbar}{2\Delta H}.$$

This is a bound on distinguishable state change. It should not be applied indiscriminately to driven or open dynamics without the corresponding generalized hypotheses.

A conserved energy distribution can drive rapid change

Prepare $(\lvert E_0\rangle+\lvert E_1\rangle)/\sqrt2$, with gap $\delta E=E_1-E_0>0$. The energy probabilities remain exactly $1/2,1/2$ at all times, so the mean and variance are conserved. Yet

$$|\langle\psi(0)|\psi(t)\rangle|^2 =\cos^2\!\left(\frac{\delta E\,t}{2\hbar}\right).$$

The first orthogonal state occurs at $\pi\hbar/\delta E$. Since $\Delta H=\delta E/2$, the bound is saturated. The state changes through a relative phase; no temporary violation of energy conservation occurs.

An energy eigenstate has $\Delta H=0$ and acquires only a global phase under its own Hamiltonian. That statement concerns its ray, not an impossibility of measuring its energy.

Linewidth, observation duration, and a clock

An approximately exponential decay amplitude $e^{-t/(2\tau)}$ produces an approximately Lorentzian line with energy full width $\Gamma=\hbar/\tau$. The exponential approximation and line shape have a regime of validity; the Lorentzian full width is not the standard deviation of energy, which an ideal unbounded Lorentzian does not possess.

A finite observation window also broadens a frequency estimate through Fourier analysis. That instrumental bandwidth differs from intrinsic lifetime broadening. Clock accuracy, in turn, depends on a physical clock’s distinguishable states, resources, and readout. One cannot substitute any of these durations for $\tau_A$ without establishing the relevant theorem.

These distinctions also block the claim that virtual particles are directly observed violations of energy conservation. Perturbative intermediate states are elements of an amplitude calculation, not short-lived exceptions licensed by one uncertainty slogan.

Exercise

A two-level superposition has gap $\delta E$ and survives with probability $1/2$ at its first half-contrast time. Find that time and compare it with $t_\perp$.

Solution

The first solution is $\delta E\,t/(2\hbar)=\pi/4$, hence $t=\pi\hbar/(2\delta E)=t_\perp/2$. The energy distribution is unchanged at this time, as at all others. A loss of overlap is not energy loss.

References

• Mandelstam, L., and Tamm, I., “The uncertainty relation between energy and time in non-relativistic quantum mechanics,” Journal of Physics (USSR) 9, 249–254 (1945) — dynamical timescale inequality.
• Page, D. N., and Wootters, W. K., “Evolution without evolution: Dynamics described by stationary observables,” Physical Review D 27, 2885–2892 (1983), doi:10.1103/PhysRevD.27.2885 — clock time as a distinct relational construction.