Mikkel T. Jensen

Reducing the run-time complexity of multiobjective EAs: The NSGA-II and other algorithms

The last decade has seen a surge of research activity on multiobjective optimization using evolutionary computation and a number of well performing algorithms have been published. The majority of these algorithms use fitness assignment based on Pareto-domination: Nondominated sorting, dominance counting, or identification of the nondominated solutions. The success of these algorithms indicates that this type of fitness is suitable for multiobjective problems, but so far the use of Pareto-based fitness has lead to program run times in O(GMN/sup 2/), where G is the number of generations, M is the number of objectives, and N is the population size. The N/sup 2/ factor should be reduced if possible, since it leads to long processing times for large population sizes. This paper presents a new and efficient algorithm for nondominated sorting, which can speed up the processing time of some multiobjective evolutionary algorithms (MOEAs) substantially. The new algorithm is incorporated into the nondominated sorting genetic algorithm II (NSGA-II) and reduces the overall run-time complexity of this algorithm to O(GN log/sup M-1/N), much faster than the O(GMN/sup 2/) complexity published by Deb et al. (2002). Experiments demonstrate that the improved version of the algorithm is indeed much faster than the previous one. The paper also points out that multiobjective EAs using fitness based on dominance counting and identification of nondominated solutions can be improved significantly in terms of running time by using efficient algorithms known from computer science instead of inefficient O(MN/sup 2/) algorithms.

Designer titania-supported Au-Pd nanoparticles for efficient photocatalytic hydrogen production

Photocatalytic hydrogen evolution may provide one of the solutions to the shift to a sustainable energy society, but the quantum efficiency of the process still needs to be improved. Precise control of the composition and structure of the metal nanoparticle cocatalysts is essential, and we show that fine-tuning the Au-Pd nanoparticle structure modifies the electronic properties of the cocatalyst significantly. Specifically, Pd(shell)-Au(core) nanoparticles immobilized on TiO2 exhibit extremely high quantum efficiencies for H2 production using a wide range of alcohols, implying that chemical byproducts from the biorefinery industry can be used as feedstocks. In addition, the excellent recyclability of our photocatalyst material indicates a high potential in industrial applications. We demonstrate that this particular elemental segregation provides optimal positioning of the unoccupied d-orbital states, which results in an enhanced utilization of the photoexcited electrons in redox reactions. We consider that the enhanced activity observed on TiO2 is generic in nature and can be transferred to other narrow band gap semiconductor supports for visible light photocatalysis.

Generating robust and flexible job shop schedules using genetic algorithms

The problem of finding robust or flexible solutions for scheduling problems is of utmost importance for real-world applications as they operate in dynamic environments. In such environments, it is often necessary to reschedule an existing plan due to failures (e.g., machine breakdowns, sickness of employees, deliveries getting delayed, etc.). Thus, a robust or flexible solution may be more valuable than an optimal solution that does not allow easy modifications. This paper considers the issue of robust and flexible solutions for job shop scheduling problems. A robustness measure is defined and its properties are investigated. Through experiments, it is shown that using a genetic algorithm it is possible to find robust and flexible schedules with a low makespan. These schedules are demonstrated to perform significantly better in rescheduling after a breakdown than ordinary schedules. The rescheduling performance of the schedules generated by minimizing the robustness measure is compared with the performance of another robust scheduling method taken from literature, and found to outperform this method in many cases.

Improving robustness and flexibility of tardiness and total flow-time job shops using robustness measures

The traditional focus of scheduling research is on finding schedules with a low implementation cost. However, in many real world scheduling applications finding a robust or flexible schedule is just as important. A robust schedule is a quality schedule expected to still be acceptable if something unforeseen happens, while a flexible schedule is a quality schedule expected to be easy to change. In this paper, the robustness and flexibility of schedules produced by minimising different robustness measures are investigated. One kind of robustness measure is the neighbourhood-based robustness measure, in which the basic idea is to minimise the implementation costs of a set of schedules located around a centre schedule. For tardiness problems another way of improving robustness is to increase the slack of the schedule by minimising lateness instead of tardiness. The problems used in the experiments are maximum tardiness, summed tardiness and total flow-time job shop problems. The experiments showed that the neighbourhood-based robustness measures improves robustness for all the problem types. Flexibility is improved for maximum tardiness and loose summed tardiness problems, while it is not improved for tight summed tardiness problems and total flow-time problems. The lateness-based robustness measures are found to also improve robustness and in some cases flexibility for the same problems, but the improvement is not as substantial as with the neighbourhood-based measures. Based on these observations, it is conjectured that neighbourhood-based robustness can be expected to improve flexibility on problems with few critical points.

Analysis and modeling of control tasks in dynamic systems

Most applications of evolutionary algorithms deal with static optimization problems. However, in recent years, there has been a growing interest in time-varying (dynamic) problems, which are typically found in real-world scenarios. One major challenge in this field is the design of realistic test-case generators (TCGs), which requires a systematic analysis of dynamic optimization tasks. So far, only a few TCGs have been suggested. Our investigation leads to the conclusion that these TCGs are not capable of generating realistic dynamic benchmark tests. The result of our research is the design of a new TCG capable of producing realistic nonstationary landscapes.

Guiding single-objective optimization using multi-objective methods

This paper investigates the possibility of using multi-objective methods to guide the search when solving single-objective optimization problems with genetic algorithms. Using the job shop scheduling problem as an example, experiments demonstrate that by using helper-objectives (additional objectives guiding the search), the average performance of a standard GA can be significantly improved. The helper-objectives guide the search towards solutions containing good building blocks and helps the algorithm avoid local optima. The experiments reveal that the approach only works if the number of helper-objectives used simultaneously is low. However, a high number of helper-objectives can be used in the same run by changing the helper-objectives dynamically.

An optimal polynomial time algorithm for the common cycle economic lot and delivery scheduling problem

Abstract The previous algorithm for solving the economic lot and delivery scheduling problem does not guarantee a globally optimal solution. In this paper, we develop a polynomial time algorithm which guarantees a globally optimal solution to the problem. We provide some insights into problem parameters for which the previous algorithm provides sub-optimal solutions.

A new look at solving minimax problems with coevolutionary genetic algorithms

A two-dimensional semiconductor image sensor has row lines which are selectable over a vertical shift register for driving first selection transistors of the sensor elements. Column lines are connected to a read-out line by way of first switches which are sequentially driven by the outputs of a horizontal shift register. Blooming is reduced. This is achieved by providing second selection transistors for the sensor elements which are connected in series with the first selection transistors and which are sequentially driven by the outputs of the horizontal shift register, and is further achieved by further switches which connect the column lines to a reference potential and which are inversely selectable relative to the first switches.

Scalable carbon dioxide electroreduction coupled to carbonylation chemistry

Significant efforts have been devoted over the last few years to develop efficient molecular electrocatalysts for the electrochemical reduction of carbon dioxide to carbon monoxide, the latter being an industrially important feedstock for the synthesis of bulk and fine chemicals. Whereas these efforts primarily focus on this formal oxygen abstraction step, there are no reports on the exploitation of the chemistry for scalable applications in carbonylation reactions. Here we describe the design and application of an inexpensive and user-friendly electrochemical set-up combined with the two-chamber technology for performing Pd-catalysed carbonylation reactions including amino- and alkoxycarbonylations, as well as carbonylative Sonogashira and Suzuki couplings with near stoichiometric carbon monoxide. The combined two-reaction process allows for milligram to gram synthesis of pharmaceutically relevant compounds. Moreover, this technology can be adapted to the use of atmospheric carbon dioxide.Electroreduction of CO2 to CO is a potential valorisation pathway of carbon dioxide for fine chemicals production. Here, the authors show a user-friendly device that couples CO2 electroreduction with carbonylation chemistry for up to gram scale synthesis of pharmaceuticals even under atmospheric CO2.

Palladium-Catalyzed Carbonylative α-Arylation of tert-Butyl Cyanoacetate with (Hetero)aryl Bromides.

A three-component coupling protocol has been developed for the generation of 3-oxo-3-(hetero)arylpropanenitriles via a carbonylative palladium-catalyzed α-arylation of tert-butyl 2-cyanoacetates with (hetero)aryl bromides followed by an acid-mediated decarboxylation step. Through the combination of only a stoichiometric loading of carbon monoxide and mild basic reaction conditions such as MgCl2 and dicyclohexylmethylamine for the deprotonation step, an excellent functional group tolerance was ensured for the methodology. Through the use of (13)C-labeled carbon monoxide generated from (13)COgen, the corresponding (13)C-isotopically labeled β-ketonitriles were obtained, and these products could subsequently be converted into cyanoalkynes and 3-cyanobenzofurans with site specific (13)C-isotope labeling.