Approximation and Semiclassical Methods

Most quantum systems cannot be solved exactly. Approximation methods let us predict their energies, states, and dynamics by exploiting something simpler: a weak coupling, a useful trial state, a large action, or well-separated scales. This volume explains both how to calculate an answer and how to decide where it is reliable.

Required background. States and operators supply the language of the calculations; solvable wave-mechanics models provide examples to compare with; quantum dynamics explains the time-evolution pictures used for driven systems.

Start with the question you want to answer

A method is useful only in relation to a particular observable and regime. The same potential may invite perturbation theory for a low-lying level, a variational estimate for its ground state, and WKB for highly excited states.

Your question	A useful starting point
How does a weak static interaction change an energy or state?	Time-independent perturbation theory
How does a weak drive cause transitions?	Time-dependent perturbation theory
Can I bound a ground-state energy without solving the equation?	Variational and bound methods
What survives when actions are large compared with $\hbar$?	WKB and semiclassical methods
How small is a tunneling splitting or decay rate?	Tunneling and nonperturbative effects
Can distant energy levels or rapid motion be simplified?	Effective Hamiltonians and scale separation

For collision probabilities, begin with Scattering Theory, which defines amplitudes, incident flux, and cross sections before introducing approximations to them.

Learn to recognize a small parameter

Calling an interaction “weak” is not enough. A coupling with the dimensions of energy must be compared with an energy scale. For two unperturbed levels separated by $\Delta$, coupled by a matrix element $v$, the ratio $\lvert v\rvert/\Delta$ measures their mixing. Even a modest coupling can mix levels strongly when their separation is small.

This is why ordinary perturbation theory can fail near a degeneracy, and why keeping both levels and diagonalizing their small matrix often succeeds. Validity and error control introduces observable-dependent accuracy and limiting regimes; small parameters and error estimates distinguishes a bound, an estimated error, and a useful consistency check.

Not every approximation uses a small coupling. A variational energy has a one-sided bound under suitable assumptions. A semiclassical expansion uses a limiting hierarchy in action or wavelength. Learning what each result actually promises is as important as learning its formula.

Choose a reading path

For a first pass, compare perturbative, variational, and asymptotic reasoning, then study nondegenerate perturbation theory, the variational principle, and WKB in that order. Each introduces a different way of justifying an approximation. Use the decision tree for a particular problem and the worked problems for calculations to attempt alongside those chapters.

For driven systems, continue from perturbation theory to transition amplitudes and Fermi’s golden rule. Pay particular attention to the observation time: a weak drive can accumulate a large effect. For low-energy models, move from degenerate perturbation theory to projection methods and the Schrieffer–Wolff transformation.

For semiclassical physics, study turning-point matching before applying tunneling or quantization formulas. The distinction between transmission, a bound-state splitting, and metastable decay determines the boundary conditions and prefactor.

Use the computational notebooks to compare approximations with numerical solutions. The reference pages collect formulas and method comparisons after their meaning is familiar. Shankar and Zwiebach’s MIT notes, listed below, offer complementary textbook and lecture-based treatments.

Exercises

The same coupling, different behavior. Two systems have the same off-diagonal coupling $\lvert v\rvert=0.01\,\mathrm{eV}$. Their uncoupled level separations are $1\,\mathrm{eV}$ and $0.001\,\mathrm{eV}$. In which system is nondegenerate perturbation theory a reasonable starting point?

Solution

The ratios are $0.01$ and $10$. Weak mixing is plausible in the first system. In the second, the two nearby levels should be treated together. The numerical value of $v$ alone did not decide the answer.

An energy bound is not an exact state. Why can a good variational ground-state energy leave a weak tail of the wavefunction inaccurate?

Solution

The energy is an integrated expectation value. A part of the wavefunction that contributes little to this integral can still matter greatly for a tail-sensitive observable, such as a tunneling amplitude. The variational theorem bounds the energy; a separate comparison is needed for that observable.

References

• R. Shankar, Principles of Quantum Mechanics, 2nd ed., Springer, 1994. Chapters on variational, WKB, and perturbative methods.
• B. Zwiebach, Quantum Physics III, MIT OpenCourseWare 8.06, 2018. Lecture notes on perturbation theory, semiclassics, and transitions.