[理学]math_modeling_chenwei.ppt

Example: Shift Scheduling A plant production manager is attempting to devise a workforce shift schedule. Each day of every working week is divided into three eight-hour shift periods: Night shift (12:01 AM–8:00 AM) Day shift (8:01 AM–4:00 PM) Late shift (4:01 PM–12:00 midnight) This table specifies the number of workers required for each shift: This means, for example, that on Monday night, there need to be at least five workers while on Saturday late there need to be at least two. The workers’ union agreement specifies: A worker may be assigned to only one shift. Once assigned, a worker must remain on the same shift. Work assignments are for four consecutive days during any seven day period. This means that somebody who starts at Tuesday night will also work on Wednesday, Thursday and Friday night shifts; and somebody who starts at Sunday day will also work on Monday, Tuesday and Wednesday day shifts. The production manager must: Meet the scheduling objectives in the table. While minimizing the total weekly workforce. Subject to the constraints of the union agreement. Mon Tue Wed Thu Fri Sat Sun Night 5 3 2 4 3 2 2 Day 7 8 9 5 7 2 5 Late 9 10 10 7 11 2 2 The MathWorks * The MathWorks * All the functions listed in previous slide are supported by state of the art algorithms, including Line search Quasi-Newton Trust region Interior point Sequential quadratic programming Gauss Newton LM methods Nelder Mead simplex And more All functions have default behavior, but there are a variety of optional arguments to tune behavior, including tolerances, the type of display, and choice of methods We also provide a framework to perform basic diagnostics to assess your Optimization Finally, many Optimization Toolbox functions incorporate optional ‘large-scale’ algorithms. These methods are especially well suited to problems with many dimensions and inherent structure, or those that can be desc