Time as a Parameter in Quantum Mechanics

In ordinary nonrelativistic quantum mechanics, time is usually an external parameter. A state is written as $\lvert\psi(t)\rangle$, operators may have explicit dependence on $t$, and the Hamiltonian generates translations in that parameter. Time is not, in the default formulation, a universal self-adjoint observable analogous to the position operator.

This distinction prevents many common errors. A particle can have a position distribution at a fixed time, but standard quantum mechanics does not begin with a single operator whose spectral measure gives “the time of the particle” for every system and every question.

Time in Classical Mechanics

In elementary Hamiltonian mechanics, time is also normally a parameter. A trajectory is a map

$$t \longmapsto \bigl(q(t),p(t)\bigr),$$

and Hamilton’s equations are

$$\frac{dq^i}{dt} = \frac{\partial H}{\partial p_i}, \qquad \frac{dp_i}{dt} = - \frac{\partial H}{\partial q^i}.$$

The coordinates $q^i$ and momenta $p_i$ are dynamical variables. The parameter $t$ labels the succession of states. A Hamiltonian may have explicit time dependence,

$$H=H(q,p,t),$$

which means the rule generating the motion changes with the external time parameter. A driven oscillator is the standard classical example.

Time in Nonrelativistic Quantum Mechanics

The quantum analogue replaces phase-space points by state vectors or density operators. In the Schrödinger picture, a pure state satisfies

$$i\hbar\frac{d}{dt}\lvert\psi(t)\rangle = H(t)\lvert\psi(t)\rangle.$$

The parameter $t$ labels the state assignment. At each time, observables are represented by operators. If $A$ has no explicit time dependence, its expectation value is

$$\langle A\rangle_t = \langle\psi(t)\rvert A\lvert\psi(t)\rangle.$$

If the observable itself depends explicitly on time, write $A(t)$:

$$\langle A\rangle_t = \langle\psi(t)\rvert A(t)\lvert\psi(t)\rangle.$$

This distinction is important. The state can change with time, the measured quantity can change with time, or both can happen at once.

For a density operator,

$$\rho(t) = U(t,t_0)\rho(t_0)U^\dagger(t,t_0)$$

in closed-system unitary evolution. The same time parameter labels the state of the system and the external settings used to define the Hamiltonian.

Why Time Is Not Treated Like Position

Position is an observable. For a particle on the line, the position operator $X$ has a spectral measure that gives probabilities for position intervals at a specified time:

$$\operatorname{Prob}(x\in \Delta;t) = \langle\psi(t)\rvert P_X(\Delta)\lvert\psi(t)\rangle.$$

The phrase “at a specified time” is doing real work. The probability distribution is indexed by $t$; the operator $X$ is not a clock.

A common temptation is to postulate a time operator $T$ satisfying

$$[T,H]=i\hbar I.$$

For many ordinary Hamiltonians this cannot be a universal self-adjoint observable. The standard obstruction is spectral: if a self-adjoint $T$ generated arbitrary energy translations, then

$$e^{-iaT/\hbar}H e^{iaT/\hbar} = H+aI.$$

That would shift the spectrum of $H$ by any real $a$. A Hamiltonian with a lower-bounded spectrum, such as the harmonic oscillator or hydrogen atom, cannot have that spectrum invariant under all real shifts.

This argument is not a ban on all useful time observables. Arrival times, dwell times, clock readings, and detector times can be modeled operationally, often with POVMs or with an explicit quantum clock included in the Hilbert space. The point is narrower and more important: ordinary quantum mechanics does not contain one universal self-adjoint time operator that plays the same structural role as position for every Hamiltonian.

Time-Dependent Observables and Hamiltonians

The derivative of an expectation value separates dynamical evolution from explicit time dependence. For a closed system,

$$\frac{d}{dt}\langle A(t)\rangle = \frac{i}{\hbar} \langle[H(t),A(t)]\rangle + \left\langle \frac{\partial A(t)}{\partial t} \right\rangle,$$

with the expectation value evaluated in the current state. The commutator term comes from state evolution; the partial-derivative term comes from the observable’s explicit dependence on the time parameter.

A Hamiltonian may also depend explicitly on time:

$$H(t) = H_0+V(t).$$

Here $t$ labels an external drive, pulse, moving trap, time-dependent field, or changing control parameter. If $H(t)$ remains self-adjoint and well-defined, the closed-system evolution is still unitary. What changes is that energy need not be conserved, because the external drive can exchange energy with the system.

Energy-Time Uncertainty Preview

Because time is not a universal observable in the usual formalism, energy-time uncertainty is not the same kind of statement as position-momentum uncertainty. The relation

$$\Delta x\,\Delta p \ge \frac{\hbar}{2}$$

comes from two observables $X$ and $P$ with a canonical commutator. Energy-time statements instead require specifying what “time” means: a lifetime, a change timescale, an orthogonalization time, an arrival time, or a measurement duration.

For example, a Mandelstam–Tamm type bound relates energy spread to the timescale over which an observable changes. It is a dynamical statement, not permission to say that energy conservation can be violated for a short time. The careful treatment belongs to Energy-Time Uncertainty.

Relativistic and Quantum-Gravity Caveats

The external-time structure is part of the nonrelativistic framework. Relativistic quantum theory treats space and time as spacetime coordinates, and quantum field theory uses fields, local operators, propagators, and correlation functions rather than a fixed-particle wavefunction as the fundamental language. Even there, practical calculations often choose a time coordinate or a foliation to define states and time-ordered products.

Quantum gravity raises deeper questions because the spacetime geometry that supplies the time parameter may itself be dynamical. Those questions are not settled by the nonrelativistic formalism. They should be treated as frontier or foundations issues, not as corrections to the basic Schrödinger equation inside its normal domain of validity.

Common Mistakes

• Treating time as an ordinary observable without specifying a clock, detector, or operational model.
• Reading $\Delta E\,\Delta t\ge\hbar/2$ as a universal Robertson uncertainty relation.
• Saying energy conservation is “temporarily violated” because a process is short-lived.
• Confusing a time-dependent state $\lvert\psi(t)\rangle$ with a time-dependent Hamiltonian $H(t)$.
• Forgetting the explicit term $\langle\partial A/\partial t\rangle$ for observables whose definition changes with time.
• Using relativistic or quantum-gravity slogans to modify nonrelativistic formulas outside their stated assumptions.

Cross-Links

• Foundations of Time Evolution
• Time-Dependent Schrödinger Equation
• Hamiltonians as Generators
• Time-Dependent Hamiltonians
• Energy-Time Uncertainty
• Assumptions and Scope
• Relationship to QFT

References

• P. A. M. Dirac, The Principles of Quantum Mechanics, 4th ed., Oxford University Press, 1958.
• W. Pauli, General Principles of Quantum Mechanics, Springer, 1980.
• A. Messiah, Quantum Mechanics, Dover, 1999.
• R. Shankar, Principles of Quantum Mechanics, 2nd ed., Springer, 1994.
• L. E. Ballentine, Quantum Mechanics: A Modern Development, 2nd ed., World Scientific, 2014.
• Y. Aharonov and D. Bohm, “Time in the Quantum Theory and the Uncertainty Relation for Time and Energy,” Physical Review 122, 1649-1658, 1961.
• P. Busch, M. Grabowski, and P. J. Lahti, Operational Quantum Physics, Springer, 1995.

Exercises

Expectation-value derivative

Let $\lvert\psi(t)\rangle$ satisfy the time-dependent Schrödinger equation and let $A(t)$ be an explicitly time-dependent observable. Derive

$$\frac{d}{dt}\langle A(t)\rangle = \frac{i}{\hbar} \langle[H(t),A(t)]\rangle + \left\langle \frac{\partial A(t)}{\partial t} \right\rangle.$$

Solution

Differentiate

$$\langle A(t)\rangle = \langle\psi(t)\rvert A(t)\lvert\psi(t)\rangle.$$

Using

$$\frac{d}{dt}\lvert\psi(t)\rangle = -\frac{i}{\hbar}H(t)\lvert\psi(t)\rangle$$

and the adjoint equation,

$$\frac{d}{dt}\langle\psi(t)\rvert = \frac{i}{\hbar}\langle\psi(t)\rvert H(t),$$

gives

$$\frac{d}{dt}\langle A(t)\rangle = \frac{i}{\hbar}\langle H(t)A(t)\rangle - \frac{i}{\hbar}\langle A(t)H(t)\rangle + \left\langle \frac{\partial A(t)}{\partial t} \right\rangle.$$

Combining the first two terms gives the commutator form.

Spectral obstruction

Suppose there were a self-adjoint operator $T$ with $[T,H]=i\hbar I$. Show why this is incompatible with a Hamiltonian whose spectrum is bounded below.

Solution

Formally applying the Baker-Campbell-Hausdorff identity gives

$$e^{-iaT/\hbar}H e^{iaT/\hbar} = H+aI.$$

The left side is unitarily equivalent to $H$, so it has the same spectrum as $H$. The right side has the spectrum of $H$ shifted by $a$. If this were true for every real $a$, the spectrum could not have a lowest energy. Therefore a lower-bounded Hamiltonian cannot have such a universal self-adjoint conjugate time operator.

This is a formal obstruction, not a statement that clocks, arrival-time distributions, or detector time records cannot be modeled.