Speaker
Description
Variational quantum eigensolvers are typically formulated in the circuit model, where parameterized unitaries prepare Ansatz states whose parameters are optimized to approximate the ground state of a given Hamiltonian. In measurement-based quantum computation (MBQC), however, variational state preparation is realized using an entangled graph state, adaptive single-qubit measurements in parameterized bases, and corrections conditioned on measurement outcomes. This creates two primary difficulties for MBQC-native Ansatz design: the realized transformation is not automatically deterministic, and its variational action is not directly apparent from the graphical description of the computation. We introduce a framework for constructing and analyzing deterministic MBQC-native Ansatz. Our central object is an MBQC-native construction, called a gadget, that admits both deterministic execution and a transparent variational interpretation. We show that the transformation realized by such a gadget has a structured form that allows a supplied warm-start approximation to be incorporated into the parameter optimization and reveals how the geometry of the gadget influences the optimization landscape. These properties bring MBQC-native Ansatz design closer to the compositionality and interpretability available for parameterized gates in the circuit model. To enlarge the available design space, we present an efficient algorithm for sampling random gadget instances and show how gadgets can be composed to construct larger Ansatz with additional variational freedom. Numerical investigations illustrate the practical role of the framework in diagnosing and improving the resulting optimization behaviour.
| Topic | Quantum variational circuits and their applications |
|---|---|
| Additional Topic | Hybrid quantum-classical approaches for learning and optimization |