Skip to content
Quantum.world

The Signal

Study classifies Berry-phase estimation across three input settings

A paper published in Quantum presents a quantum algorithm for estimating Berry phases across their full range modulo 2π. It also classifies the problem’s complexity under three different assumptions about information available about the ground state.

Study classifies Berry-phase estimation across three input settings

A full-range estimation method

A quantum algorithm for estimating Berry phases across their full range modulo 2π appears alongside new complexity classifications in a paper published in Quantum on November 18, 2026. The authors, Ryu Hayakawa, Kazuki Sakamoto and Chusei Kiumi, study how the problem’s computational complexity changes with the information supplied about a system’s ground state.

A Berry phase is a geometric phase acquired by an eigenstate as a Hamiltonian moves around a closed path. Earlier approaches cited in the paper could be restricted to estimating phases in the interval from zero to π when the Hamiltonian lacked time-reversal symmetry. The new method removes that restriction, the authors report, and estimates the phase modulo a full 2π cycle.

The method compares adiabatic evolutions along the same closed Hamiltonian path but with different total runtimes. The paper explains that the geometric phase stays unchanged while the dynamical phase scales with runtime; combining results from quantum phase estimation can therefore separate the two contributions. Under the stated conditions, the algorithm’s running time is O(poly(n) · 1/εB · log(1/η)), for additive accuracy εB of at least 1/poly(n) and failure probability η between zero and one-third.

Complexity depends on the inputs

For the first setting, the problem includes a “guiding state” with at least inverse-polynomial overlap with the initial ground state. The authors prove that Berry-phase estimation in this setting is BQP-complete. The paper describes BQP as the class of decision problems that quantum computers can solve efficiently.

In a second setting, the initial ground state is not supplied, but an energy threshold separates its energy from that of the first excited state. Here, the authors prove completeness for UQMA ∩ co-UQMA, where UQMA is the unique-witness version of QMA. They say this is, to their knowledge, the first natural problem complete for that intersection class.

A third formulation supplies neither a guiding state nor an energy threshold. For this version, the paper establishes hardness for PUQMA∩co-UQMA[log] under polynomial-time truth-table reductions, and containment in PPGQMA[log].

Scope of the result

The authors’ hardness results use families of 5-local Hamiltonians on qubits with polynomially bounded interaction strengths. For the guiding-state case, the paper says that the BQP-completeness result supports a quantum-advantage conclusion under the standard assumption that BPP is not equal to BQP; its discussion describes the implication as a superpolynomial advantage.

The work is a theoretical study of an estimation algorithm and the complexity of different problem formulations. Its results concern the stated Hamiltonian settings and assumptions, rather than a reported hardware demonstration.

Sources

  1. Quantum
  2. Quantum

Related stories

Explore

More articles