Source-linked AI summary
Adiabatic Quantum Simulation of Quantum Chemistry
Ryan Babbush, Peter J. Love, Alán Aspuru-Guzik
TL;DR
The paper addresses the next step beyond annealing the classical Ising model: applying adiabatic quantum computation to molecular electronic structure. It presents a scalable mapping to 2-local Hamiltonians with experimentally suitable interactions, while preserving polynomial scaling of experimental resources and the eigenvalue gap.
Problem
Applying adiabatic quantum computation to molecular electronic structure is presented as the natural next step beyond annealing the classical Ising model.
Method
The paper maps molecular electronic structure to a 2-local Hamiltonian using a scalable construction with ZZ, XX, and XZ terms while removing experimentally challenging YY terms.
Results
The resulting electronic-structure Hamiltonians are equivalent to the requirements of universal adiabatic quantum computation, with experimental resources scaling polynomially.
Takeaways & Limitations
The construction provides a physically implementable spin-Hamiltonian representation of molecular electronic structure suitable for adiabatic quantum computation.
Abstract
from arXiv · showhide
We show how to apply the quantum adiabatic algorithm directly to the quantum computation of molecular properties. We describe a procedure to map electronic structure Hamiltonians to 2-local qubit Hamiltonians with a small set of physically realizable couplings. By combining the Bravyi-Kitaev construction to map fermions to qubits with perturbative gadgets to reduce the Hamiltonian to 2-local, we obtain precision requirements on the coupling strengths and a number of ancilla qubits that scale polynomially in the problem size. Hence our mapping is efficient. The required set of controllable interactions includes only two types of interaction beyond the Ising interactions required to apply the quantum adiabatic algorithm to combinatorial optimization problems. Our mapping may also be of interest to chemists directly as it defines a dictionary from electronic structure to spin Hamiltonians with physical interactions.
I. INTRODUCTION
The introduction motivates extending adiabatic quantum computation from classical optimization to molecular electronic structure, where exact classical calculations become exponentially difficult. It presents a scalable mapping to programmable physical systems using 2-local Hamiltonians with experimentally constrained interactions.
- Adiabatic quantum computation: AQC offers a framework for preparing ground states by evolving from an easy initial Hamiltonian to one whose ground state encodes the target problem.The adiabatic runtime is governed by the minimum gap between the ground and first excited states along the path.
- Motivation: The introduction identifies molecular quantum simulation as a natural next step beyond adiabatic annealing of classical Ising models.Molecular systems broaden the target from classical optimization Hamiltonians to interacting-fermion electronic structure problems.
- Motivation: Molecular electronic structure is difficult because directly solving the exact Hamiltonian’s eigenvalues grows exponentially with problem size, leaving many exact calculations out of reach.Approximate classical algorithms have advanced substantially, but the passage states that exact calculations remain inaccessible for many systems of interest.
- Contribution: The paper describes a scalable method for applying the quantum adiabatic algorithm to molecular electronic Hamiltonians on a programmable physical system.The method begins from a molecular electronic Hamiltonian and targets a physical encoding suitable for adiabatic computation.
- Contribution: Bravyi–Kitaev encoding and perturbative gadgets yield a 2-local Hamiltonian with only ZZ, XX, and ZX couplings, while removing experimentally challenging YY terms.The stated interaction set extends the Ising interactions used for combinatorial optimization by two additional coupling types.
II. SECOND QUANTIZATION
The second-quantized formulation represents molecular electronic structure in an occupation-number basis of spin orbitals. Creation and annihilation operators obey fermionic anticommutation relations, and the Hamiltonian coefficients are precomputed electron-overlap integrals.
- Occupation-number basis: The electronic structure problem is formulated in an occupation-number basis over spin orbitals, with each orbital represented as empty or occupied.Basis states are tensor products over configurations of the molecular spin orbitals.
- Spin orbitals: Spin orbitals combine spatial molecular orbitals with spin-up or spin-down functions; molecular hydrogen therefore has four spin orbitals in the example.The example labels the orbitals as products of two spatial orbitals and two spin states.
- Fermionic operators: Electron interactions are expressed through combinations of creation and annihilation operators that obey fermionic anticommutation relations.These operators act on the occupation-number representation of the electronic states.
- Hamiltonian: The second-quantized molecular electronic Hamiltonian uses one- and two-electron overlap integrals, which are precomputed classically.The passage distinguishes the single-electron coefficients h_ij from the double-electron coefficients h_ijkl.
III. QUBIT REPRESENTATION
The section compares fermion-to-qubit mappings for electronic structure and motivates Bravyi–Kitaev as a scalable route to 2-local Hamiltonians. Jordan–Wigner introduces locality that grows linearly with system size, whereas Bravyi–Kitaev reduces this overhead logarithmically.
- Mapping limitations: The Jordan–Wigner transformation is not scalable for experimentally realizable adiabatic Hamiltonians.Its nonlocal parity strings make the resulting interactions unsuitable for the targeted hardware setting.
- Mapping limitations: Jordan–Wigner produces k-local interaction terms with k growing linearly in system size.This locality growth creates an exponential control-precision requirement when perturbative gadgets reduce interaction order.
- Alternative mappings: The parity-basis construction offers no advantage over Jordan–Wigner because updating occupancy still requires operations on a number of qubits that grows with system size.It localizes parity information but makes occupancy updates nonlocal.
- Bravyi–Kitaev mapping: Bravyi–Kitaev stores parity and occupancy information nonlocally, allowing both to be accessed using O(log n) qubits.Neither quantity is determined from a single qubit, but both have logarithmic access complexity.
- Bravyi–Kitaev mapping: The Bravyi–Kitaev transformation yields an n-qubit Hamiltonian that is (log n)-local.This logarithmic locality is the basis for an efficient subsequent reduction to 2-local form.
IV. HAMILTONIAN GADGETS
The paper uses perturbative Hamiltonian gadgets to embed logarithmically local electronic-structure Hamiltonians into 2-local systems with experimentally relevant interactions. The construction approximates low-energy spectra, while its main practical trade-offs involve control precision, ancilla count, and higher-order contamination.
- Construction: The proposed mapping embeds electronic structure in an experimentally realizable Hamiltonian and reduces the Bravyi–Kitaev Hamiltonian to 2-local form.The gadget Hamiltonian uses 2-local physical interactions to approximate the target Hamiltonian in a restricted low-energy subspace.
- Construction: Perturbative gadgets embed the eigenspectra of an n-qubit target Hamiltonian into a more constrained N-qubit gadget Hamiltonian.The effective Hamiltonian acts in a typically low-energy subspace of the gadget system.
- Hardware trade-offs: Bit-flip gadgets require less control precision than other constructions but generally use more ancillae.The paper focuses on this gadget family because of that precision–ancilla trade-off.
- Limitations: The gadget construction requires high control precision and should be avoided when possible.For entirely diagonal Hamiltonians, exact gadgets can require substantially less precision and often fewer ancillae, although gap scaling need not be preserved.
- Spectral approximation: Under the gadget theorem’s spectral-gap and norm assumptions, each low-energy gadget eigenvalue is ε-close to the corresponding effective-Hamiltonian eigenvalue.The theorem assumes a spectral gap around the cutoff, ∥V∥ ≤ ∆/2, and an ε-close self-energy approximation.
- Higher-order effects: Cross-gadget contamination produces unwanted higher-order processes, but for Pauli operators that square to identity it contributes only a compensable constant energy shift.This property permits compensation through the energy parameter Λ.
V. EXAMPLE PROBLEM: MOLECULAR HYDROGEN
The molecular-hydrogen example factors the Hamiltonian into commuting operator terms and reduces its local structure using perturbative gadgets. The construction introduces ancilla systems and compensating interactions to obtain a 2-local Hamiltonian with controlled interaction types.
- Hamiltonian decomposition: The Hamiltonian is separated into 2-local and 4-local parts, then targeted for reduction to a 2-local ZZ/XX/XZ Hamiltonian.The authors choose this division because splitting into 2-, 3-, and 4-local terms is less efficient for their procedure.
- Hamiltonian decomposition: The molecular-hydrogen Hamiltonian is factored into commuting three-operator terms, enabling the bit-flip gadgets to operate correctly.Each term is written as A_iB_iC_i with pairwise commuting operators.
- Ancilla construction: Each logical operator receives an associated ancilla qubit, while each three-operator term receives a fully connected ancilla system.The unperturbed Hamiltonian is constructed from fully connected ancilla systems with a spectral gap labeled for the first perturbative stage.
- Perturbative construction: A 2-local compensation Hamiltonian is chosen to cancel unwanted perturbative contributions, leaving the intended effective terms in the large-gap limit.The compensation cancels the second-order contribution exactly and removes the unwanted third-order contribution asymptotically.
- Perturbative construction: Second-order bit flips generate self-interaction terms, while third-order flips transition between degenerate ancilla ground states and generate the desired A_iB_iC_i terms.The third-order transition requires flipping all three ancilla bits through operators coupled to A, B, and C.
A. Second Order
At second order, only single-bit flips followed by flips of the same bit return the ancilla system to its ground space. These processes therefore produce effective self-interactions for the associated logical operators.
- A. Second Order: Only processes that flip one ancilla bit and then flip the same bit back return from the ground state to the ground state at second order.The resulting processes are the three bit-flip processes shown for each term.
- A. Second Order: Second-order perturbation creates effective interactions between each logical operator and itself.The contribution contains squared operators associated with the A, B, and C operators in each ancilla system.
B. Third Order
At third order, sequences of three ancilla bit flips generate the target three-operator terms, while competing processes create unwanted contributions. A second perturbative gadget then embeds the construction into a 2-local Hamiltonian.
- B. Third Order: An unwanted third-order process combines one interaction with H2-local or Λ1 in the high-energy ancilla subspace.These competing processes are illustrated alongside the desired process in the third-order diagrams.
- B. Third Order: Choosing Λ1 appropriately cancels the unwanted third-order contribution while also canceling the second-order term in the large-∆1 limit.The compensation is selected from the perturbative operators and the first spectral gap.
VI. CONCLUSION
The paper presents a scalable mapping from molecular electronic-structure Hamiltonians to 2-local spin Hamiltonians with experimentally motivated interactions. The construction preserves the target spectrum while requiring polynomially scaling experimental resources, although further reductions tighten precision demands and the resulting models are non-stoquastic.
- Mapping and interactions: The mapping reduces molecular electronic-structure Hamiltonians to 2-local Hamiltonians containing only ZZ, XX, and XZ interactions.Further reductions can limit the interaction set to ZZ and XX or ZZ and XZ terms.
- Scalability and implementation: All experimental resources—including qubits, control precision, and graph degree—scale polynomially with the number of orbitals.The paper describes the method as scalable, while a detailed resource study remains underway.
- Mapping and interactions: Perturbative gadgets embed the entire target Hamiltonian, preserving its eigenvalue gap during the reduction.Bit-flip gadgets remove experimentally challenging YY terms.
- Scalability and implementation: The resulting interactions are equivalent to those required for universal adiabatic quantum computation, while repeated reductions impose more stringent precision requirements.The chosen target set balances control precision against a reasonable number of controllable interaction types.
- Measurement and implications: The construction supports direct eigenspectrum measurement through tunneling spectroscopy of a probe qubit coupled to the simulated system.Transitions occur when the probe bias approaches an eigenvalue, revealing the original system’s eigenspectrum.
- Measurement and implications: The mapped spin Hamiltonians are non-stoquastic, so classical simulation techniques suffer from the fermionic sign problem.This motivates quantum hardware for simulating the mapped electronic-structure problem.
APPENDIX
The appendix develops the perturbative-gadget calculation by defining low- and high-energy projectors and evaluating perturbation components term by term. It then combines these projected terms to obtain the relevant perturbative-series contributions.
- Projector construction: The gadget analysis begins by defining projectors onto the ancilla ground space and its orthogonal high-energy subspace.The high-energy projector is the complement of the ground-space projector.
- Projected perturbation terms: The perturbation is decomposed into projected components acting within and between the low- and high-energy subspaces.The appendix explicitly calculates V− and V+ using projector orthogonality.
- Term-by-term expansion: The first perturbative-series contribution is computed as V−+V+V+− using projector expressions rather than expanded operator forms.The calculation relies on orthogonality between the high- and low-energy subspaces.