Trial Wavefunctions
A trial wavefunction is a guessed state with one or more adjustable parameters. Choosing a good family is the art of the variational method; optimizing the parameters is the mechanics. This lesson walks through both for the cleanest possible example, the one-dimensional harmonic oscillator.
Choosing a family
A useful trial family should satisfy three things: it must be normalizable, it should respect the symmetries and boundary behaviour of the problem, and it should contain at least one knob to turn. For a smooth, symmetric, confining potential like the harmonic oscillator
a natural guess is a single Gaussian centred at the origin,
It is square-integrable, even in (matching the symmetric potential), and decays at infinity. Its one parameter controls the width: large is a narrow spike, small is a broad spread.
Computing the energy
With Gaussian integrals the kinetic and potential expectation values work out (in natural units , after normalizing) to
so the variational energy is
This captures the universal tension behind every confined ground state: shrinking the wavefunction (large ) lowers the potential energy but, by the uncertainty principle, raises the kinetic energy, and vice versa. The optimum balances the two.
Optimizing
Minimize over by setting the derivative to zero:
Evaluating there,
In these units the exact ground-state energy is . The Gaussian trial state nails it exactly — because the true oscillator ground state is a Gaussian, so the family contains the right answer. This is the best case: when the trial family includes the exact solution, the variational minimum is the exact energy.
Try it
Minimize over and return the minimum energy. Your
code should return a single number.
Sign in on the full site to ask questions and join the discussion.