Chapter 19 Key Takeaways: The Variational Principle

Core Message

The variational principle provides rigorous upper bounds on ground state energies without requiring you to solve the Schrödinger equation. By encoding physical intuition into a trial wavefunction and minimizing the energy expectation value over adjustable parameters, you can achieve remarkable accuracy for systems that are analytically intractable. The variational method is the foundation of all modern computational quantum mechanics.

Key Concepts

1. The Variational Theorem

For any normalized trial state $|\psi\rangle$ in the Hilbert space of a quantum system with Hamiltonian $\hat{H}$ and ground state energy $E_0$:

$$\langle \psi | \hat{H} | \psi \rangle \geq E_0$$

Equality holds if and only if $|\psi\rangle$ is the exact ground state. The proof follows from expanding $|\psi\rangle$ in the energy eigenbasis and using $E_n \geq E_0$ for all $n$. This means every trial function gives a legitimate upper bound — you can never accidentally undershoot the true ground state energy.

2. Trial Wavefunctions and Physical Intuition

The quality of a variational calculation depends entirely on the choice of trial function. Good trial functions: (a) satisfy the correct boundary conditions, (b) capture the qualitative shape of the ground state, (c) incorporate known physics (cusp conditions, screening, symmetry), and (d) contain enough adjustable parameters to accommodate the unknown features. More parameters always improve (or maintain) the optimized energy.

3. Screening and the Helium Atom

The helium ground state cannot be solved exactly due to electron-electron repulsion. A one-parameter trial function $\psi \propto e^{-Z'(r_1+r_2)}$ with effective nuclear charge $Z'$ gives $Z'_{\text{opt}} = 27/16 \approx 1.69$ and energy $E = -2.848$ hartree — beating first-order perturbation theory ($-2.750$ hartree) and achieving 1.9% accuracy vs. experiment ($-2.904$ hartree).

4. Chemical Bonding in H₂⁺

The LCAO trial function $\psi = c_A\phi_A + c_B\phi_B$ for the hydrogen molecule ion reveals chemical bonding as a quantum interference effect. The exchange integral $K$ — with no classical analogue — creates the energy splitting between bonding (attractive) and antibonding (repulsive) molecular orbitals.

5. The Ritz Method (Linear Variational Method)

When the trial function is a linear combination of fixed basis functions, $|\psi\rangle = \sum c_i |\phi_i\rangle$, the variational optimization reduces to the generalized eigenvalue problem $\mathbf{H}\mathbf{c} = E\mathbf{S}\mathbf{c}$. This gives upper bounds on all eigenvalues simultaneously (Hylleraas-Undheim-MacDonald theorem) and is the foundation of computational quantum chemistry.

6. Variational Monte Carlo

For many-electron systems, energy integrals are high-dimensional and cannot be evaluated by standard quadrature. VMC uses Metropolis sampling from $|\psi|^2$ to estimate the average local energy $E_L = \hat{H}\psi/\psi$. The statistical error scales as $1/\sqrt{M}$ regardless of dimensionality. The zero-variance principle ensures that the statistical noise vanishes as the trial function approaches an eigenstate.

Key Equations

Key Techniques

Critical Distinctions

Variational Method vs. Perturbation Theory

What the Variational Method Does and Does NOT Tell You

It tells you: - An upper bound on $E_0$ (guaranteed by the theorem) - An approximate ground state wavefunction - How the energy depends on physical parameters (screening, bond length, etc.)

It does NOT tell you: - How close your bound is to the true $E_0$ - A lower bound (you need additional methods for that) - Guaranteed accuracy for the wavefunction (energy is second-order insensitive to wavefunction errors)

Common Pitfalls

1. Confusing upper bound with underestimate. The variational method overestimates $E_0$, not underestimates. Students frequently get the direction of the inequality backward.

2. Assuming good energy means good wavefunction. The energy is a quadratic functional of $|\psi\rangle$, so $O(\epsilon)$ wavefunction errors produce $O(\epsilon^2)$ energy errors. A 0.1% energy error can mask a 3% wavefunction error.

3. Using trial functions that violate boundary conditions. While the theorem still holds, violating boundary conditions wastes variational freedom and gives unnecessarily poor bounds.

4. Forgetting that $\hat{H}$ in the expectation value is the TRUE Hamiltonian. In the helium calculation, the variational parameter $Z'$ appears in the trial function, but the Hamiltonian always has the physical nuclear charge $Z = 2$.

5. Neglecting the cusp conditions. Trial functions that do not satisfy the Kato cusp condition at nuclear or electron-electron coalescence points converge slowly.

Connections to Other Chapters

One-Sentence Summary

The variational principle turns unsolvable quantum problems into optimization problems — any guess for the wavefunction gives a guaranteed upper bound on the ground state energy, and the better your physical intuition, the tighter the bound.