The Bravyi-Kitaev Transformation
A fermion-to-qubit encoding that stores occupation and parity information in a tree structure, cutting operator locality to logarithmic scale.
The motivation
Jordan-Wigner stores occupation numbers directly but must read parity by traversing a long chain of qubits, giving operator strings of length O(N). The Bravyi-Kitaev (BK) transformation balances two needs, storing occupation and storing partial parities, so that any fermionic operator acts on only O(log N) qubits.
The idea of partial sums
BK arranges qubits so that each stores a partial-sum (parity) over a subset of orbitals, following a binary-tree structure (the Fenwick tree). To determine the sign for an operator on orbital p, one needs only the qubits on the path to the root, of which there are O(log N), rather than all qubits below p.
The trade compared to JW
- Operator locality: O(log N) for BK versus O(N) for JW.
- Occupation readout: direct in JW, requires combining a few qubits in BK.
- Gate counts for Trotter steps: often lower for BK on dense Hamiltonians.
- Conceptual clarity: JW is simpler to reason about; BK is more efficient at scale.
Why locality matters
Shorter Pauli strings mean fewer two-qubit gates per Hamiltonian term when Trotterizing, which reduces circuit depth and accumulated noise. For molecular Hamiltonians with many terms, the logarithmic locality of BK can substantially lower the resource estimate compared with JW.
Variants and context
Related encodings, including the parity encoding and the Bravyi-Kitaev superfast method, occupy different points on the occupation-versus-parity spectrum. Superfast encodings can reach constant locality for Hamiltonians with bounded connectivity, at the cost of extra qubits. The right encoding depends on the Hamiltonian's connectivity graph and the hardware's native gates, so encoding choice is part of the compilation problem for chemistry simulation, not a fixed decision.