Bravyi-Kitaev Transformation
A fermion-to-qubit mapping that reduces Pauli operator weight from linear to logarithmic in the number of orbitals.
The locality problem
The Jordan-Wigner map stores occupation numbers directly, so determining the parity needed for fermionic signs requires reading all lower-indexed qubits, giving operators of weight up to O(n). The Bravyi-Kitaev transformation reorganizes the stored information so that both occupation and parity can be read from only O(log n) qubits.
Balancing occupation and parity
Jordan-Wigner stores occupation locally but parity nonlocally. An alternative parity basis stores parity locally but occupation nonlocally. Bravyi-Kitaev sits between them: it uses a binary tree structure where each qubit stores partial sums of occupations. Three sets of qubits become relevant for any operator: the update set, the parity set, and the flip set, each of size O(log n).
Consequences
- Creation and annihilation operators map to Pauli strings of weight O(log n).
- Time-evolution and ansatz circuits become shallower for large systems.
- The encoding is less intuitive than direct occupation storage.
Trade-offs
The logarithmic operator weight is a clear asymptotic win, but for small molecules the constant factors can make Bravyi-Kitaev comparable to or only modestly better than Jordan-Wigner. The bookkeeping (computing update, parity, and flip sets) is more involved, and the resulting Pauli strings are less transparent physically. Software libraries automate the transformation, so users rarely derive it by hand.
Where it fits
Bravyi-Kitaev is one option in the quantum chemistry workflow, chosen when circuit depth from long Pauli strings would otherwise dominate. Combined with symmetry-based qubit tapering, it helps fit larger active spaces onto limited hardware. The choice among Jordan-Wigner, parity, and Bravyi-Kitaev is problem- and device-dependent, and benchmarking all three is common practice.