Project Title: Benchmark Optimization API for Office Space Allocation 
Student: Zheyu Zhang
Course: MSc in Computer Science

Abstract:
The problem of office space allocation is a reasonable and effective way to use space 
resources and reduce the waste of space and excessive use. This problem often involves 
massive data, multiple conflicting goals and constraints, and is a combinatorial optimization 
problem for NP-hard. Traditional optimization methods, such as method of exhaustion, 
blind search methods, etc., are difficult to achieve with good results. In this paper, the 
multi-objective genetic algorithm in the meta-heuristic technology is used to allocate the 
office space allocation problem to minimize the number of space abuse and violations of 
conditional constraints. At the same time, the algorithm is enhanced stood on the principle 
of multi-objective genetic algorithm and the analysis of internal mechanism. First, using the 
modified fitness function scheme, the conditions on the problem are merged into the penalty 
function in a dynamic manner to guide the genetic search. In the evolutionary stages of each 
generation, the high-adaptive parent is retained in the progeny to find the global best result. 
The highest fitness individuals after the parent is crossed and mutated are sent to the offspring. 
Finally, the Boltzmann update mechanism is introduced in the survival strategy to converge to 
the global best result. The University of Nottingham test case is used to the office space 
allocation problem. The experimental result shows that the improved genetic algorithm to deal 
with the office space allocation problem can produce high quality and diversified solutions. 
Due to maintaining the stability of the improved genetic algorithm, the generated optimal 
solution has better convergence and can obtain satisfactory results.