Maximum average entanglement of four qubits

Split four qubits into two pairs in each of the three possible ways. For a pure state, the entropy of either pair measures its entanglement with the other. How large can the average of these three entropies be?

Why it matters and what counts as progress

Why it matters: A state can maximise entanglement across one split while doing poorly across another. The maximum average quantifies the best compromise when the three splits count equally.

What counts as progress: Prove a sharp upper bound, identify all states that attain it, or quantify how close a near-maximiser is to one of those states. Specify the entropy and the partitions being averaged.

How the problem developed

  1. 2000 · Candidate maximum [3]

    Atsushi Higuchi and Anthony Sudbery Showed that four qubits cannot be maximally entangled across every balanced split and proposed a state with high average entropy.

  2. 2007 · Local maximum [4]

    Stephen Brierley and Atsushi Higuchi Proved the proposed state locally maximises the average entropy.

  3. 3 Nov 2026 · Global maximum and equality cases [1]

    Antony Lewis with Claude and GPT Reported a sharp global bound and classified the two local-unitary orbits that attain it.

AI contribution

The most entangled states of four and seven qubits

Antony Lewis

For pure states of four qubits, the mean von Neumann entropy of the three balanced cuts is at most 1+12log⁡231+\tfrac12\log_2 3 bits. Equality holds exactly for the Higuchi–Sudbery state and its complex conjugate, up to independent unitary transformations on the four qubits.

AI involvementAI-led
ImportanceNotable
Technical verificationFormal (Lean 4)
Human understandingFew
Paper date:

Sources

  1. The most entangled states of four and seven qubits
  2. Lewis’s Lean formalization and theorem index
  3. Higuchi and Sudbery’s four-qubit state
  4. Brierley and Higuchi’s local maximum result
Publication history