Mapping fermions to qubits: Jordan–Wigner, Bravyi–Kitaev
Electrons are fermions: no two share a state, and swapping two flips the sign of the wavefunction. Qubits are not fermions: they are independent two-level systems whose operators on different qubits simply commute. To simulate electrons on qubits, the mapping must build the fermion sign rule into the operators. The oldest way, from Pascual Jordan and Eugene Wigner in 1928, attaches a string of Z gates to every operator; Sergey Bravyi and Alexei Kitaev found in 2002 a cleverer encoding with much shorter strings. The choice changes circuit cost.
On the plate: a fermion operator turned into Pauli strings under three mappings.
Jordan–Wigner
Store the occupation of spin orbital directly in qubit . Creating an electron in orbital must also apply a sign for every occupied orbital before it, so:
The Z string counts the parity of the occupied orbitals to the left. One can check directly that these operators satisfy and ; this book’s code checks it for 4 and 8 modes. The cost: an operator on orbital acts on qubits, so terms involving distant orbitals become long Pauli strings, up to the number of orbitals. Established
Compare mappings
Choose a mapping: , and orbital 6 of 8. The plate shows the two Pauli strings that make up . The parity mapping stores in qubit the parity of orbitals 0 to , so the sign is local but updating an occupation flips all qubits to the right. Bravyi–Kitaev stores partial parity sums arranged as a binary tree (a Fenwick tree), so both updating and reading parity touch only about qubits. For 8 orbitals, every Bravyi–Kitaev string has weight at most 4, against up to 8 for Jordan–Wigner; the gap widens with size. This book’s code checks the anticommutation relations for the Bravyi–Kitaev operators too. Established
Tapering with symmetries
Molecular Hamiltonians conserve electron number and spin, and often have point-group symmetries. Each such symmetry, expressed as a Pauli operator commuting with the Hamiltonian, lets one qubit be removed by fixing its eigenvalue. For H₂ in STO-3G, the parity mapping turns 4 spin orbitals into 4 qubits, two of which store the total and spin parities and are fixed: two qubits remain, and the Hamiltonian is the five-term one used in Chapter 122, . Its lowest eigenvalue reproduces the exact energy of Chapter 184. Tapering is free accuracy-preserving compression, and every practical calculation uses it. Established
What you now have
The Jordan–Wigner transformation and its Z strings, the parity and Bravyi–Kitaev mappings with their update and parity sets, Pauli weight scaling (linear versus logarithmic), and qubit tapering using symmetries, ending at the two-qubit H₂ Hamiltonian.
· · ·
Three questions to carry with you.
And one to work out. Under Jordan–Wigner, on how many qubits does act, for 100 spin orbitals numbered from 0?
Sources: P. Jordan and E. Wigner, Z. Phys. 47, 631 (1928). S. B. Bravyi and A. Yu. Kitaev, Ann. Phys. 298, 210 (2002). J. T. Seeley, M. J. Richard and P. J. Love, J. Chem. Phys. 137, 224109 (2012). S. Bravyi et al., arXiv:1701.08213 (2017) (tapering). Checked by code: anticommutation relations hold exactly for Jordan–Wigner (4 modes) and Bravyi–Kitaev (8 modes); BK string weights at most 4 for 8 modes.