|q⟩ Bad Qubits

intermediate · Physics · The Variational Method & WKB

Estimating Ground-State Energy

The previous lesson chose a trial family that happened to contain the exact answer. The variational method earns its keep when the trial family does not contain the true ground state — then it still delivers a rigorous upper bound, and we learn how good (or poor) the guess is.

A deliberately imperfect guess

Take the hydrogen atom, in atomic units (=me=e=1\hbar = m_e = e = 1, energies in Hartree):

H^=1221r,E0=12 Ha.\hat{H} = -\frac{1}{2}\nabla^2 - \frac{1}{r}, \qquad E_0 = -\frac{1}{2} \text{ Ha}.

The true ground state decays exponentially, ψ0er\psi_0 \propto e^{-r}. Suppose we did not know that and guessed a Gaussian instead,

ψα(r)=eαr2.\psi_\alpha(r) = e^{-\alpha r^2}.

A Gaussian falls off too fast at large rr and is too flat at the origin (it has zero slope there, whereas the real wavefunction has a cusp). It is the wrong shape — exactly the situation where the variational principle proves its worth.

The variational energy

Carrying out the (standard Gaussian) integrals for the kinetic and Coulomb terms and normalizing gives

E(α)=32α22απ.E(\alpha) = \frac{3}{2}\alpha - 2\sqrt{\frac{2\alpha}{\pi}}.

The first term is the kinetic cost of confinement; the second is the (negative) Coulomb attraction. Minimizing,

dEdα=322π1α=0        α=89π.\frac{dE}{d\alpha} = \frac{3}{2} - \sqrt{\frac{2}{\pi}}\,\frac{1}{\sqrt{\alpha}} = 0 \;\;\Longrightarrow\;\; \alpha^* = \frac{8}{9\pi}.

Substituting back,

E(α)=43π0.4244 Ha.E(\alpha^*) = -\frac{4}{3\pi} \approx -0.4244 \text{ Ha}.

Reading the result

Two facts make this a textbook illustration of the method:

Try it

Minimize E(α)=32α22α/πE(\alpha) = \tfrac{3}{2}\alpha - 2\sqrt{2\alpha/\pi} over α>0\alpha > 0 and return the minimum energy (a negative Hartree number). Confirm for yourself that it stays above 0.5-0.5.

Run your code to see the quantum state.

Sign in on the full site to ask questions and join the discussion.