Developing inverted simulated annealing algorithm and solving timetabling problem

Simulated Annealing (SA) is a statistical method for solving constraints satisfaction and optimization problem: Its main focus is on neighbor function and cooling scheduling. Neighbor function depends on the applied problem. Neighbor function chooses a random solution that is neighbor to the current solution in some meaning. The temperature in cooling scheduling determines the acceptance threshold even when a neighbor solution is not better than the current solution. This paper suggests all inverted SA algorithm which is an improved algorithm of SA. In SA, neighbor solution is chosen first and then tested whether it can be accepted as a new current solution based on the threshold. But in the inverted SA, the threshold is determined first and then used to choose a neighbor solution satisfying the threshold. The inverted SA method is effective in solving the strictly constraint problem such as the timetable problem in universities where random style neighbor function is not suitable. In this paper, we use a modified best-first search considering threshold as a neighbor function to solve the timetabling problem. The simulation result shows that the inverted SA finds a better solution than SA.