Presenters: Diyi Liu and Shuchen Zhu
The success of many quantum simulation algorithms depends on the construction of efficient quantum circuits that can be implemented on a quantum computer to realize the effect of a many-body Hamiltonian on a properly prepared quantum state. These circuit constructions often rely on oracles that encode the Hamiltonian as a sequence of unitary transformations in some way. Such oracles form what is called the input model, which can generally be separated into two parts: a prepare oracle and a select oracle. One particular strategy for implementing a physical Hamiltonian on a quantum computer is known as block encoding, where the Hamiltonian is embedded into a larger unitary matrix.
We present an efficient block encoding scheme for a second quantized Hamiltonian to address the following two challenges. (a) The number of terms $L$ in Hamiltonian is large; (b) Unnecessary resources used for $\eta$ particle state simulation.
The quantum circuit we construct for the block encoding scheme has two desireable features: (a) T gate count is sublinear in $L$; (b) sublinear subnormalization factor in $L$ for $\eta$ particle state block encoding.
Under the most general setting where the Hamiltonian’s coefficients $\alpha_i=\mathcal{O}(1)$ , the T gate counts can be reduced from $\mathcal{O}(L)$ to $\mathcal{O}(\sqrt{L})$ and the Clifford gate count equals $L\log(\frac{1}{\epsilon})+o(L)$ where $\epsilon$ is the precision for input model. In addition, we also developed $\eta$ particle block encoding with subnormalization reduced from $\mathcal{O}(L)$ to $\mathcal{O}(\sqrt{L})$. Further improvements are discussed if certain decay properties are assumed for the coefficients.
