Application of Deep Neural Networks to the Operator Space of Nonlinear PDE for Physics-based Proxy Modeling