We propose an efficient quantum algorithm for simulating the dynamics of general Hamiltonian systems. Our technique is based on a power series expansion of the time-evolution operator in its off-diagonal terms. The expansion decouples the dynamics due to the diagonal component of the Hamiltonian from the dynamics generated by its off-diagonal part, which we encode using the linear combination of unitaries technique. Our method has an optimal dependence on the desired precision and, as we illustrate, generally requires considerably fewer resources than the current state-of-the-art. We provide an analysis of resource costs for several sample models.
Main idea
Using the permutation matrix representation (PMR) of a time-independent Hamiltonian \(H = D_0 +\sum_i D_i P_i\) for diagonals \(D_i\) and permutation matrices \(P_i\vert z \rangle = \vert z'\rangle\), an algorithm to simulate the unitary evolution \(e^{i H t}\) is given, by expanding as a Taylor series in the number of permutations, and using LCU.
Background
The PMR method was originally developed for QMC by expanding the partition function \(\mathrm{Tr}e^{-\beta H}\) (see e.g. arXiv:1908.03740)
Here the same general technique is employed by for the unitary operator \(U(t) = e^{-i H t}\), where the expansion is in \(q\), the number of permutations, e.g. \(\prod_{i=1}^q P_{i_k}\)
The Linear Combination of Unitaries (LCU) method (arXiv:1202.5822) is used to simulate an operator \(A = \sum_i \alpha_i U_i\) probabilistically (though amplitude amplification can often be employed to boost the success)
Results
The expansion decomposes exactly into a product of unitaries \(U = U_{od}e^{-i D_0 t}\), where \(U_{od}\) is the off-diagonal contribution
The unitary \(U_{od}\) can be written itself as a sum of unitaries and simulated by the LCU method, where the sum is eventually truncated to order \(q=Q\) with calculable error \(Q = O\left( \log (T/\epsilon) / \log \log T/\epsilon \right)\) for evolution time \(T\) to error \(\epsilon\) (n.b. the simulation is actually done as a product of short-time evolutions)
Circuit description is given for the LCU simulation and how to use oblivous amplitude amplification to increase the success probability to arbitrary accuracy
The overall claim is that this outperforms standard LCU methods where \(H = \sum_i \alpha_i U_i\) since the construction here “is generally considerably more compact”, resulting in fewer terms in the LCU circuit (examples of this are given in Table 3)
The remaining diagonal terms are simulated directly via \(e^{-i D_0 \Delta t}\) with fixed cost \(O(C_{D_0} )\) where \(C_{D_0}\) is the cost to evaluate a matrix element of \(D_0\) (see after Eq. (12))
Comparing to the standard LCU prescription for qubit systems, an example with an \(O(N^2)\) reduction in the number of unitaries is given