Unleashing the Power of Global Gates for Preparing Long-Range Entangled Quantum States

In many quantum computing architectures, the standard gate set includes arbitrary single-qubit gates and two-qubit entangling gates like CNOT. However, this choice is not always aligned with the native interactions available in certain hardware. In ion trap quantum computers, for instance, the natural entangling operation is not restricted to two qubits but can act on multiple, or even all, qubits simultaneously. These global gates present a unique advantage by enabling multi-qubit interactions in a single operation.

We leverage this capability to design an ansatz composed of global gates and single-qubit unitaries, which can prepare highly entangled states in constant time. By simulating the ground states of the Toric code Hamiltonian and the Heisenberg model with next-nearest neighbor interactions, we demonstrate that our ansatz generates long-range entanglement and possesses high expressibility, comparable to other state-of-the-art ansatzes, which typically require time scaling linearly with system size. Furthermore, the ansatz remains trainable as the system size increases, avoiding the barren plateau problem, as evidenced by the slow decay of gradient variance as the qubit count grows, which ensures scalability while maintaining performance. Despite being barren plateau-free, our ansatz is not efficiently simulable classically, highlighting its potential to offer a quantum advantage.