Random Sampling-Based Automatic Parameter Tuning for Nonlinear Programming Solvers

Abstract

Nonlinear programming solvers play important roles in process systems engineering. The performance of a nonlinear programming solver is influenced significantly by values of the solver parameters. Hence, tuning these parameters can enhance the performance of the nonlinear programming solver, especially for hard problems and time-critical applications like real time optimization and nonlinear model predictive control. Random sampling (RS) algorithm is utilized to tune the nonlinear programming solver for solving hard problems. By introducing an iterated search technique, heuristic rules and advanced termination criteria, an enhanced random sampling algorithm is developed to determine parameter configuration that work significantly better than the default. These random sampling-based methods can handle all kinds of parameters (e.g., categorical, integer, and continuous) of the nonlinear programming solver. Numerical results with parameter configurations from the proposed random sampling-based methods show remarkable performance improvement compared with the defaults. Future works toward the development and improvement of the automatic parameter-tuning tool for the nonlinear programming solvers are also outlined.

Process Intensification of Polymerization Processes with Embedded Molecular Weight Distributions Models: An Advanced Optimization Approach

“…These studies − , focused on minimization of the target MWD to the calculated MWD, rather than directly improving the economic performance of the process. Moreover, the optimization can be quite time-consuming with increasing model complexity (e.g., polymer thermodynamics). , By improving the modeling and solution strategy with the EO model described in section and IPOPT , with a parameter-tuning method, the challenges associated with the MWD optimization can be efficiently dealt with.…”

“…Moreover, the optimization can be quite time-consuming with increasing model complexity (e.g., polymer thermodynamics). 58,59 By improving the modeling and solution strategy with the EO model described in section 2 27 and IPOPT 29,61 with a parameter-tuning method, 62 the challenges associated with the MWD optimization can be efficiently dealt with.…”

Internal-growth Strategies for Dynamic Process Optimization with Differential-algebraic Equations

“…IPOPT is not only able to receive accurate Hessian Matrix, but also has flexible solution options. According to the specific situation of the exact problems, the users are able to set the solver flexibly and pertinently to accelerate the convergence speed and increase the success rate of solving (Chen et al, 2011).…”

Equation-Oriented Optimization on an Industrial High-Density Polyethylene Slurry Process with Target Molecular Weight Distribution

“…Instead, bound constraints are replaced with a logarithmic barrier term added into the objective function, and a sequence of equality-constrained optimization problems is solved with a monotonically decreasing or adaptive updated barrier parameter. This algorithm exhibits excellent convergence properties and superior performance, especially on the large-scale NLPs. − In this project, AMPL, an EO modeling system, is used to construct the complete optimization formulation, and IPOPT 3.10.1 is used to solve the optimization problem.…”

“…Hence, the method of PAT developed by the authors is exploited to help IPOPT solve the HDPE optimization problems. The enhanced random sampling-based (ERS) strategy containing an iterated search technique, heuristic rules, and advanced termination criteria is applied to find the parameter settings that work significantly better than their defaults . Figure presents the whole problem solving process.…”