New Heuristics for global optimization of complex bioprocesses

A reconsideration on the resolution of phase stability analysis using stochastic global optimization methods: Proposal of a reliable set of benchmark problems

Due to their universal applicability, global stochastic optimization methods are popular for designing improvements of biochemical networks. The drawbacks of global stochastic optimization methods are: (i) no guarantee of finding global optima, (ii) no clear optimization run termination criteria and (iii) no criteria to detect stagnation of an optimization run. The impact of these drawbacks can be partly compensated by manual work that becomes inefficient when the solution space is large due to combinatorial explosion of adjustable parameters or for other reasons. SpaceScanner uses parallel optimization runs for automatic termination of optimization tasks in case of consensus and consecutively applies a pre-defined set of global stochastic optimization methods in case of stagnation in the currently used method. Automatic scan of adjustable parameter combination subsets for best objective function values is possible with a summary file of ranked solutions. https://github.com/atiselsts/spacescanner . egils.stalidzans@lu.lv. Supplementary data are available at Bioinformatics online.

Solution of implicitly formulated inverse heat transfer problems with hybrid methods

In this paper, the design of non-uniform concentric circular antenna arrays (CCAAs) of isotropic radiators with optimum side lobe level reduction is presented. The newly-proposed global evolutionary optimization methods; namely: the biogeography based optimization (BBO) is used to determine a set of weights that provides a radiation pattern with optimum side lobe level reduction with the constraint of a fixed major lobe beamwidth. The results obtained from BBO are compared with those obtained using the Matlab function Fmincon which uses a sequential quadratic programming (SQP) method. The comparison shows that the design of non-uniformly excited CCAAs using SQP method provides a side lobe level reduction that is better than that obtained using global stochastic optimization methods, indicating that global optimization techniques are not really needed in this problem.

Engineering design problems often involve global optimization of functions that are supplied as ’black box‘ functions. These functions may be nonconvex, nondifferentiable and even discontinuous. In addition, the decision variables may be a combination of discrete and continuous variables. The functions are usually computationally expensive, and may involve finite element methods. An engineering example of this type of problem is to minimize the weight of a structure, while limiting strain to be below a certain threshold. This type of global optimization problem is very difficult to solve, yet design engineers must find some solution to their problem – even if it is a suboptimal one. Sometimes the most difficult part of the problem is finding any feasible solution. Stochastic methods, including sequential random search and simulated annealing, are finding many applications to this type of practical global optimization problem. Improving Hit-and-Run (IHR) is a sequential random search method that has been successfully used in several engineering design applications, such as the optimal design of composite structures. A motivation to IHR is discussed as well as several enhancements. The enhancements include allowing both continuous and discrete variables in the problem formulation. This has many practical advantages, because design variables often involve a mixture of continuous and discrete values. IHR and several variations have been applied to the composites design problem. Some of this practical experience is discussed.

In this study, the performance of Scatter Search (SS) meta-heuristic for the post-enrolment course timetabling problems reported in recent literature was reviewed. The aim is to address the strengths and limitations of the SS structure and mechanisms; to empower more studies; to investigate the capabilities of SS; and to enhance it for solving timetabling problems as a whole. The SS is almost similar to memetic algorithms. However, it has a memory of elite solutions (which considers both high quality and diverse solutions) and combines two or more solutions explicitly based on elitism. SS contains a mechanism to strike a balance between diversification and intensification of the search. SS has five major steps: initializing a diverse collection of solutions; general solutions improvement; memorizing elite solutions, solutions combination; and improving the selected solution. Based on the outcomes of SS (tested on post-enrolment course timetabling problems), which was reported in the literature, this study concluded that the impact of SS's strategies are significant for a successful SS search performance. These strategies are: memory update, solutions selection, diversification (i.e. similarity measurement), and solutions combination. Indeed, updating the memory and solutions combination, have the greatest impact on the performance of SS. Therefore, future studies regarding the design of these two strategies for solving timetabling problems is recommended to be more carefully designed.

Scatter search is a population-based method that has recently been shown to yield promising outcomes for solving combinatorial and nonlinear optimization problems. Based on formulations originally proposed in 1960s for combining decision rules and problem constraints such as the surrogate constraint method, scatter search uses strategies for combining solution vectors that have proved effective in a variety of problem settings. In this paper, we develop a general purpose heuristic for a class of nonlinear optimization problems. The procedure is based on the scatter search methodology and treats the objective function evaluation as a black box, making the search algorithm context-independent. Most optimization problems in the chemical and bio-chemical industries are highly nonlinear in either the objective function or the constraints. Moreover, they usually present differential-algebraic systems of constraints. In this type of problem, the evaluation of a solution or even the feasibility test of a set of values for the decision variables is a time-consuming operation. In this context, the solution method is limited to a reduced number of solution examinations. We have implemented a scatter search procedure in Matlab (Mathworks, 2004) for this special class of difficult optimization problems. Our development goes beyond a simple exercise of applying scatter search to this class of problems, but presents innovative mechanisms to obtain a good balance between intensification and diversification in a short-term search horizon. Computational comparisons with other recent methods over a set of benchmark problems favor the proposed procedure.

The ability to quickly fabricate sensorcraft, or other small unmanned aircraft, via additive manufacturing techniques opens a range of new possibilities for the design and optimization of these vehicles. In this paper we propose a design loop that makes use of surrogate modeling and additive manufacturing to reduce the design and optimization time of scientific sensorcraft. Additive manufacturing reduces the time and effort required to fabricate a complete aircraft, allowing design iterations to be quickly manufactured and flight tested. Co-Kriging surrogate models allow data collected from test flights to correct Kriging models trained with numerically simulated data. The resulting model provides physically accurate and computationally cheap aircraft performance predictions. A global optimizer is used to search this model to find an optimal design for a bespoke aircraft. This paper presents the design loop and the progress made in implementing this design loop. Results are shown from Kriging models trained using numerically simulated data. Progress towards extracting aerodynamic data from flight testing small unmanned aircraft is also documented.

Stochastic Methods for Global Optimization and Problem Solving

We discuss the use of surrogate models in the field of genetic programming. We describe a set of features extracted from each tree and use it to train a model of the fitness function. The results indicate that such a model can be used to predict the fitness of new individuals without the need to evaluate them. In a series of experiments, we show how surrogate modeling is able to reduce the number of fitness evaluations needed in genetic programming, and we discuss how the use of surrogate models affects the exploration and convergence of genetic programming algorithms.

The use of surrogate models in computational mechanics is an area of high interest due to the potential for significant savings in computational cost. However, assessment and presentation of evidence for surrogate model credibility has yet to reach a standard form. The present study utilizes a deep neural network as a surrogate for a computational fluid dynamics simulation in order to predict the coefficients of lift and drag on a NACA 0012 airfoil for various Reynolds numbers and angles of attack. Using best practices, the credibility of the underlying simulation predictions and of the surrogate model predictions are analyzed. Conclusions are drawn which should better inform future uses of surrogate models in the context of their credibility.

Efficient global sensitivity analysis for flow-induced vibration of a nuclear reactor assembly using Kriging surrogates

Model-based methods for continuous and discrete global optimization

During the early stages of the design process, designers rarely have accurate models of system behavior, yet the success of their designs depends on understanding the effect of changes in the design parameters on the system response. When models are available, they are often expensive to evaluate and difficult to run, in large part due to the imprecise knowledge available at the beginning of the design process. To circumvent these problems, a common approach proposed in the literature is the use of surrogate models. In this paper, we propose a framework for building surrogate models in multiple stages for time-dependent systems. Because they are built in stages, the surrogate models can respond to changes in constraints and can be fine-turned as design decisions are made. When designers have mathematical models available, they can perform repeated sensitivity analyses, explore trade-offs and perform optimization studies. In this framework, the observed responses are viewed as a set of time-correlated spatial processes. The framework uses optimal sampling techniques to improve the accuracy of the resulting surrogate model while keeping the number of samples to a minimum. A new non-stationary covariance structure is proposed and tested with an example design application. The resulting models are compared with the surrogates obtained with the stationary covariance structure proposed by Romero et al. [1]. The results show increased accuracy using the non-stationary covariance structure due to its superior interpolation capabilities.Copyright © 2004 by ASME

This study outlines an approach for speeding up the simulation of the dynamic response of vehicle models that include hysteretic nonlinear tire components. The method proposed replaces the hysteretic nonlinear tire model with a surrogate model that emulates the dynamic response of the actual tire. The approach is demonstrated via a dynamic simulation of a quarter vehicle model. In the proposed methodology, training information generated with a reduced number of harmonic excitations is used to construct the tire hysteretic force emulator using a Neural Network (NN) element. The proposed approach has two stages: a learning stage, followed by an embedding of the learned model into the quarter car model. The learning related main challenge stems from the attempt to capture with the NN element the behavior of a hysteretic element whose response depends on its loading history. The methodology is demonstrated in conjunction with a simple nonlinear quarter vehicle system as well as an ADAMS based model that uses a complex tire element. The results obtained with the surrogate model prove to be accurate and are obtained at a fraction of the CPU time required to handle the original models. The approach proposed is anticipated to be useful for reducing the duration of vehicle simulations, or when a tire model is not available but experimental data can be used to generate a surrogate model.