In this thesis, we study online algorithms for different packing problems from the field of combinatorial optimization. As considering fully adversarial inputs leads to strong impossibility results for most of these problems, we consider the random-order model. This model is an alternative to pure worst-case analysis and became increasingly popular over the past years. For the k-secretary problem and the knapsack problem, we propose new algorithms and evaluate them in terms of competitive ratio. For bin packing, we revisit a well-known algorithm and present several improved bounds.
«
In this thesis, we study online algorithms for different packing problems from the field of combinatorial optimization. As considering fully adversarial inputs leads to strong impossibility results for most of these problems, we consider the random-order model. This model is an alternative to pure worst-case analysis and became increasingly popular over the past years. For the k-secretary problem and the knapsack problem, we propose new algorithms and evaluate them in terms of competitive ratio....
»