LAUSR

In this paper, we introduce two novel discrete differential evolution (DDE) methods for the optimization of simulated tree pruning within a software support tool for demonstration of tree training techniques. Therein, the pruning is posed as a combinatorial optimization problem of performing the cuts on a virtual tree model, whereby the objective function is defined by an empirical model of light interception. The proposed path-based and set-based DDE methods are closed to a discrete search domain under the implemented mutation operators. We compare both methods to several popular discrete optimization algorithms and a selection of efficient metaheuristics from continuous optimization, including existing DDE variants that map a discrete problem into continuous search space using real-valued solution encodings. We demonstrate that the path-based DDE achieves the best overall performance in the experiments on problem instances of different complexity classes.