Quantum Algorithms

VQE Hardware-Efficient Ansatz

Shallow parameterized circuits built from a device's native gates, chosen for trainability on noisy hardware rather than physical fidelity.

Design philosophy

A hardware-efficient ansatz builds the trial state from gates the device runs natively and from the entangling connectivity it actually has. Instead of encoding physical structure, it stacks alternating layers of single-qubit rotations and fixed entangling gates. The goal is to keep circuits shallow enough to survive decoherence while remaining expressive enough to reach a useful state.

Typical structure

Kronos motion — fidelity

A layer consists of a rotation block (e.g. RY and RZ on every qubit, each with its own angle) followed by an entangling block (e.g. a ladder or ring of CNOT or native two-qubit gates). Stacking L layers gives a circuit with O(L n) parameters for n qubits. The final state is U(theta)|0> fed to the VQE loop.

python

# schematic: L layers of RY-RZ rotations + CNOT ring on n qubits
def hea(circuit, thetas, n, layers):
    p = 0
    for _ in range(layers):
        for q in range(n):
            circuit.ry(thetas[p], q); p += 1
            circuit.rz(thetas[p], q); p += 1
        for q in range(n):
            circuit.cx(q, (q+1) % n)   # entangling ring
    return circuit

Advantages

Uses native gates, minimizing compilation overhead and error.
Shallow depth suits short coherence times.
Flexible: works for any Hamiltonian without problem-specific derivation.

Risks

Expressive, unstructured ansaetze are prone to barren plateaus, where gradients vanish exponentially with qubit count, making optimization intractable at scale. They also lack physical symmetries, so the optimizer may converge to an unphysical state or a poor local minimum. Excessive entangling layers worsen both noise accumulation and plateau severity.

Mitigations

Keep depth modest, initialize near identity or with small angles, use local cost functions, and enforce symmetries where possible. Compared with the physically motivated UCCSD ansatz, hardware-efficient circuits trade guaranteed physical structure for shorter depth, a bargain that pays off only when the problem and device are well matched.