MUNA, I. Modely a metody pro svozové problému v logistice [online]. Brno: Vysoké učení technické v Brně. Fakulta strojního inženýrství. 2019.

Posudek vedoucího

Popela, Pavel

The author fulfilled the requirements of his diploma thesis in an original way and he has achieved the defined goals at a very good level. I must emphasize that the amount of work that had been done by the author on the subject of thesis during the winter and spring periods was considerably greater than he has presented at the end. However, the author has decided not to show all steps from his initial results in his thesis and, especially, not to present his successful implementations in GAMS despite his supervisor's suggestions. So, by my subjective opinion, it is a pity that the reader will not get the full information of the author's efforts on the subject. But I must clarify that this author's decision has contributed positively to the more compact version of the final thesis. The adequacy of thesis results can be judged as very good with respect to the solved challenging issues. As an expert in the area, I would like to mention that it could be also interesting to formulate a two-stage scenario-based model in more detail way and to solve its small instance, e.g., by GAMS/CPLEX. Such a use of traditional computational technique could identify its bottlenecks, and so, could give more explicit motivation of next author's ideas. So, the author originally applied the EEV approach according to Birge in Section 3.2 and explained it by a well chosen example. He focused on the evaluation of the computed suboptimal solution of the reduced program obtained for the average demands by using scenarios. Furthermore, he has chosen another original method of the problem reduction, when he adopted a combination of a fuzzy clustering method with a traveling salesman problem approach for the capacitated vehicle routing problem with stochastic demands in Section 3.3. Then, he has illustrated it again by his example. Especially, the MAPLE implementations of Chapter 3 heuristics modified by the author, and then, utilized in Chapter 4 are quite original. However, it is again a pity that he did not pay more attention to detailed description of his original MAPLE implementation. So, the impact of some original thesis results is not so impressive on a common reader as it could be. Then, the author analysed various results of computational tests in detail and derived some conclusions from them in Chapter 4 and more general ones in Chapter 5. It is obvious that the future usability of thesis results is not in theory but it is mainly in a methodological area. Specifically, it will be useful for the existing application based cooperation between Institute of Mathematics and Institute of Process Engineering. It is a consequence of the fact that the thesis shows possibilities of using two approximative techniques for real-world CVRPSD tasks, e.g., applied to waste collection routing problems. Regarding the formal aspects of thesis, we can say that the theme of thesis is presented in a compact and logical way. The layout of thesis text into chapters is well balanced. Maybe the author could pay more attention to the size of gaps between numbered sections and to spacing within some formulas. Although overview paragraphs in Chapters 1 and 2 are less formally conceived, as concepts taken from the literature are only briefly described in the text (without the definitions and theorems). they are well illustrated by examples. Some parts of thesis like Table 2.1 and Figure 2.6 would deserve a more extensive description. We can see that the author has decided to address the information about stochastic optimization ideas by references and by short remarks in specific paragraphs. However, the inclusion of a brief overview of selected used stochastic programming concepts could be more fruitful for an uninformed reader. We can welcome the enclosed RAR file with all programs in MAPLE, as it is quite impressive because they illustrate the author’s quite serious approach to the studied topic. Unfortunately, the importance of the attached programs is not emphasized enough by more detail description within the text of thesis. The graphical layout of thesis is very nice. In particular, I highly evaluate many figures illustrating the results. I mainly like the fact that the author originally programmed their generation in MAPLE. Most typos and grammatical errors have been corrected during the final reading of thesis but it has to be stated that the English text and its style could be further improved. I also appreciate the author’s work with literature. In addition to his selection of recommended general sources, he has independently searched for extra classical and recent sources. Then, he further used them in the thesis in a suitable way. In particular, I must emphasize the original use of the existing CVRP library source by the author. Literature citations are correctly inserted in the right paragraphs. I greatly appreciate how the student has worked hard, systematically, and independently on his master thesis. He has attended the consultations regularly, well prepared and with specific questions. He was flexible, patient and polite all the time even when he was often waiting for his busy supervisor. The recommendations given to him by a supervisor or by e-mails from supervisor specialist Dusan Hrabec from UTB Zlin were carefully and promptly incorporated. After all my positive and critical comments, I conclude that the author has satisfied the goals of his thesis at very good level, and so, I recommend his thesis for the thesis defence and classify it by B mark.

Posudek oponenta

Roupec, Jan

This thesis deals with a very topical problem of optimization in logistics. The thesis is divided into five chapters. The first chapter briefly describes motivation and slightly introduces the reader to the solved problems. The second chapter presents a literature search and brings a mathematical background of graph theory and classic related problems. In my opinion, this chapter is too brief given the complexity of the issue. Especially the stochastic demand model deserved a more detailed description. The solved problem, its mathematical model and used methods are described in the third chapter. I appreciate the use of non-traditional approaches in this area. Especially combination of fuzzy clustering and TSP is very interesting. Results are presented in the fourth chapter. Finally, the fifth chapter contains conclusions related to results achieved. The author used adequate methods to solve the problem, the obtained results seem to be interesting. The formal level of this thesis satisfies all defined requirements. In my opinion, the assignment is completely fulfilled. I recommend this diploma thesis for defense and classify it with A grade.

eVSKP id 117593