Variational Principle

The variational principle is a fundamental result in quantum mechanics that establishes a lower bound: the true ground state energy of a system is the minimum possible expected value of the Hamiltonian over all normalized quantum states. For any trial state |psi>, the expectation value <psi|H|psi> is always greater than or equal to the ground state energy E_0.

Mathematical Statement

For a Hamiltonian H with ground state energy E_0 and ground state |E_0>:

<psi|H|psi> >= E_0    for all normalized |psi>

with equality if and only if |psi> is the ground state. This means you can find the ground state energy by minimizing the expected energy over a family of trial states: the minimum of <psi(theta)|H|psi(theta)> over the parameter theta provides an upper bound on E_0, and the minimizing state is the best approximation to the ground state within the trial family.

Why It Matters for Quantum Computing

The variational principle is the mathematical foundation for the Variational Quantum Eigensolver (VQE) and all other variational quantum algorithms. VQE parameterizes a quantum state |psi(theta)> using a parameterized quantum circuit (called an ansatz), evaluates <psi(theta)|H|psi(theta)> by measuring the circuit on quantum hardware, and then uses a classical optimizer to minimize this quantity over the parameters theta.

The guarantee that the energy can only decrease (or stay the same) as optimization improves the ansatz is what makes VQE a trustworthy algorithm. Any energy estimate from VQE is always an upper bound on the true ground state energy, which is a useful consistency check.

The Ansatz Choice

The quality of a variational calculation depends critically on the choice of ansatz. If the true ground state cannot be well approximated by any state in the ansatz family, no amount of optimization will find a good answer. Common ansatz strategies include:

• Hardware-efficient ansatz: circuits that match the native gate set and connectivity of the quantum device, prioritizing trainability on real hardware
• Chemically inspired ansatz: unitary coupled cluster (UCC) circuits derived from classical chemistry approximations, which tend to capture the relevant physics for molecular ground states
• MERA-inspired circuits: circuits with a hierarchical structure that can capture long-range entanglement in many-body systems

Variational Principle Beyond Ground States

The variational principle extends to excited states through orthogonality constraints. By requiring trial states to be orthogonal to already-found lower-energy states, you can sequentially find excited state energies. Quantum algorithms like Variational Quantum Deflation (VQD) implement this strategy.

The variational principle also applies to time evolution and open quantum systems. Variational time evolution algorithms minimize the difference between the parameterized circuit evolution and the exact Schrodinger equation evolution, enabling classical simulation of quantum dynamics and near-term quantum simulation of time-dependent problems.

Limitations

The variational principle guarantees an upper bound, not a lower bound. You cannot verify from the energy estimate alone whether you have converged to the true ground state or are stuck at a local minimum of the energy landscape. Barren plateaus, regions of the energy landscape where gradients vanish exponentially with system size, make variational optimization increasingly difficult for deep circuits and large systems.