Self-adaptive Matrix Equation Solving in Analog Memory Array 1

Linear matrix equations have been one of the main topics in matrix theory and its applications. In this work, a self-adaptive matrix equation solving method in analog non-volatile memory (NVM) arrays has been demonstrated. The solving method utilizes the gradient descent algorithm of the machine learning and can be illustrated by the linear neural network. Relying on the multiplication and accumulation operation of the NVM array, the self-adaptive method provides an efficient and reliable solution of the matrix equation. The influence of the device conductance states and write noise on the solution accuracy is also discussed.