VQE for H2 (2D)

Circuit diagram, measurement histograms and energy landscapes of a 2-qubit variational quantum eigensolver for the hydrogen molecule.

Switch to 3D

Readouts

Presets

Circuit and measurement

Top: Z-basis circuit (terms II, ZI, IZ, ZZ). Bottom: the same state with a Hadamard on each qubit to measure XX. Bars: exact outcome probabilities (outline) and the simulated shot counts (filled).

Independent numerical checks

The science

Honest scope: this page shows what a quantum computer is actually used for in chemistry research: estimating the energy of a small molecule with the variational quantum eigensolver (VQE). It does not model, and there is no evidence for, any "quantum healing" of the body. H2 is a textbook toy that a laptop solves exactly; VQE is of interest for larger molecules in drug and materials research and is still experimental.

The molecule. H2 at 0.735 angstrom in the minimal STO-3G basis, reduced to two qubits (parity mapping): H = -1.05237 II - 0.39794 ZI + 0.39794 IZ - 0.01128 ZZ + 0.18093 XX (Hartree; coefficients as published in the Qiskit chemistry tutorials, to be spot-checked), plus the nuclear repulsion 0.71997 Ha.

The ansatz. |ψ(θ)⟩ = cosθ|01⟩ + sinθ|10⟩; θ = 0 is the Hartree-Fock state. The energy is E(θ) = A + B cos 2θ + C sin 2θ, a sinusoid, which is why the parameter-shift rule gives the exact gradient and the landscape in the plot is a smooth single well.

Measuring energy. Z terms come from measuring both qubits in Z; XX needs a Hadamard on each qubit first. With N shots per setting the statistical error is about 1/√N (see the histogram). Depolarizing noise p only shrinks the non-identity part by (1 - p), so in the contour map the best angle is the same for every p but the well gets shallower.

Simplifications. Noise is a single global depolarizing channel, not a device model; the optimizer is plain gradient descent with fixed step 0.3; the bond length is fixed; the quoted coefficients and the FCI value are typical literature values. No medical or healing claim is made.

Molecule, qubit and gate sprites: Blender (Cycles) renders of the 3D page's model.

Energy landscape E(theta) (blue), Hartree-Fock (grey), exact ground state (green)
Optimizer: energy estimate per iteration (orange)
Energy map E(theta, noise p), bright = low energy; white dot = current state
Shot-noise histogram: repeated estimates (100-300) at the current settings
Made with 🟠 Blender