Parallel parameter optimization algorithm in dynamic general equilibrium models

We present a parallel parameter optimization algorithm for reproducing future projections of certain model outputs in dynamic general equilibrium models. The optimization problem is reduced to a nonlinear system of equations. The Jacobian matrix for a Newton type solver in the problem is generated in parallel. The parameter optimization algorithm is implemented for parallel systems with distributed memory by using MPI. To achieve better performance of the parallel algorithm we use the parallel Fair-Taylor algorithm for computing an equilibrium path. Calculation of prices, input-output ratios and international trade for different time steps is carried out in parallel at each iteration of the method. The solution method is implemented for parallel systems with shared memory by using OpenMP. The effectiveness of the hybrid MPI+OpenMP parallel code for parameter optimization is demonstrated in the example of a global multi-sector energy economics model with scenarios that are used for studying climate change impacts on land use. (C) 2018, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved.