Fig. 1: Schematic workflow of our neural network for learning entanglement \({\mathcal{E}}\equiv \{{{\mathcal{S}}}^{(n)},{{\mathcal{P}}}_{n},{\mathcal{C}}\}\) from local Pauli measurements.
From: Direct entanglement detection of quantum systems using machine learning
a For the ground states \(\vert {\psi }_{g}\rangle\), local Pauli operators are measured and they are directly used to learn \({\mathcal{E}}\) via FCNN. b For the dynamical states \(\vert {\psi }_{t}\rangle\), we only measure and input the expectation values of single-qubit Pauli operators in the range [0, Ttra]. It can predict not only the dynamics of \({\mathcal{E}}\) during the training window, but also the long-time dynamics of \({\mathcal{E}}\) at the unseen time [Ttra, Ttot].
