Date of Award
Bachelor of Science (BS)
Department of Mathematics and Statistics
Edwin O'Shea, Ph.D
Integer programming can be used to find upper and lower bounds on the cells of a multi-dimensional contingency table using the information from the released margins. The linear relaxation of these programs also provides bounds and the discrepancy between these bounds, the integer programming gap, can be large. While the more notable examples of large gaps have been shown to be rare, here we provide some results on the rarity of large gaps on small tables.
Watson, Noah J., "A Computational Investigation of Large Gaps in Contingency Tables" (2016). Senior Honors Projects, 2010-current. 151.