Stochastic Series Expansion (SSE)
The Stochastic Series Expansion (SSE) method is a finite-temperature QMC technique that expands the partition function of the quantum system in a power series of the Hamiltonian. It was originally applied to the Heisenberg model 1, but can be easily extended to other quantum models, such as Bose-Hubbard model.
The partition function of a quantum model is given by:
where is the inverse temperature and is the Hamiltonian. The key idea of SSE is to express the exponential operator as a Taylor series:
By inserting a complete set of basis states , the partition function can be rewritten as:
Depending on the temperatures, the SSE expansion order in the simulation will never exceed a finite order . The SSE method then truncates this expansion series at order and samples the terms stochastically. The Hamiltonian is typically decomposed into a sum of elementary interaction terms , such as bond operators for the Heisenberg model:
Each term acts on a pair of sites and can be represented in a suitable basis. The SSE algorithm then samples configurations consisting of a sequence of these operators.
For the Heisenberg model, the bond operator can be expressed as:
where is the -component of the spin operator, and and are the spin raising and lowering operators, respectively. The first term, , represents the diagonal part of the interaction, while the second term, , represents the off-diagonal part.
Diagonal and Off-Diagonal Matrix Elements
Heisenberg Model
In the SSE framework, the Heisenberg Hamiltonian is expressed in terms of diagonal and off-diagonal operators. For a given basis state , the matrix elements of the bond operator are:
Diagonal Matrix Elements: These correspond to the term and are given by:
where and are the -components of the spins in the state .
Off-Diagonal Matrix Elements: These correspond to the spin-flip terms and . For a state , the off-diagonal matrix elements are:
and
where and is the state obtained by flipping the spins at sites and .
Bose-Hubbard Model
The Bose-Hubbard model describes bosons on a lattice with on-site interactions and nearest-neighbor hopping. The Hamiltonian is given by:
where:
- is the hopping amplitude,
- is the on-site interaction strength,
- is the chemical potential,
- and are the bosonic creation and annihilation operators at site ,
- is the number operator,
- denotes nearest-neighbor pairs.
The Bose-Hubbard Hamiltonian is decomposed into a set of bond operators (for hopping) and (for on-site interactions):
where labels the bonds or sites. For the Bose-Hubbard model:
- Hopping terms: ,
- On-site terms: .
Insertion of Basis States
In the SSE method, the partition function is expanded in terms of basis states and operator sequences. A typical configuration in the SSE expansion consists of:
- A basis state (the initial state).
- A sequence of operators acting on the state.
The partition function can then be written as:
where is the cutoff of the expansion order and represents a sequence of operators. The matrix elements of the operators are evaluated in the basis states, and the sequence of operators must satisfy the condition that the final state matches the initial state .
Steps in the SSE Algorithm
- Initialization: Start with an initial state and an empty operator sequence.
- Operator Insertion: Propose to insert or remove diagonal operators into the sequence, updating the state accordingly.
- Diagonal Updates: Ensure that the sequence of operators is consistent with the Hamiltonian and the basis states.
- Loop Updates: Perform non-local updates to improve sampling efficiency, often using cluster or loop algorithms tailored to the spin or other bosonic models 2 3 4.
- Measurement: Compute physical quantities, such as energy, magnetization, and correlation functions, by averaging over the sampled configurations.
The SSE method is particularly advantageous for the Heisenberg model because it avoids the sign problem for certain geometries (e.g., bipartite lattices) and provides efficient sampling of both low-temperature and high-temperature regimes. It has been successfully applied to study a wide range of phenomena, including quantum phase transitions, spin dynamics, and Bosonic systems.
Sandvik, A. W., “Stochastic Series Expansion Method with Operator-Loop Update”, Physical Review B, 59, R14157-R14160 (1999). ↩︎
Syljuåsen, O. F. and Sandvik, A. W., “Quantum Monte Carlo with Directed Loops”, Physical Review E, 66, 046701 (2002). ↩︎
Pollet, L., et al., “Optimal Monte Carlo Updating”, Physical Review E, 70, 056705 (2004). ↩︎
Alet, F., et al., “Generalized Directed Loop Method for Quantum Monte Carlo Simulations”, Physical Review E, 71, 036706 (2005). ↩︎