Topic Title: Solving Ultra Large Scale Mixed Integer Programing (MIP)
Technical Area: Operations Research; Optimization
Mixed integer programming continues to be an important and challenging problem. It has been widely used in many applications such as logistics, production planning, network planning, energy, cloud computing, and etc. As a fundamental tool, it has received a significant amount of attentions from both academia and industry in the past decades. It is however computationally intractable to solve most of MIP instances due to its NP-hardness. Alibaba has a very diverse business ecosystem including logistics, e-commerce, finance, cloud computing, entertainment, map, and etc. Many problems can be formulated as mixed integer programs, thus there is a strong need of efficient MIP solver. However, the ultra large size and complicated constraints of real problems prohibit the direct application of commercial solvers such as Gurobi and CPlex. For instance, there are billions of variables in the planning problem in cloud computing. We are looking forward to new and more generic approaches to solve ultra large scale MIPs in Alibaba.
Most of common used approaches such as meta heuristics are problem specified, which highly rely on domain knowledge and are hard to apply to other problems.
We invite researchers who are experts in this field to work on new but more generic approaches to solve ultra large scale MIPs in reasonable time.
Related Research Topics:
Some related research topics are listed as follows:
- Heuristics for solving MIPs: efficient heuristics to get a good solution with tighter bound.
- Choosing good branching variables in branch-and-bound algorithms: heuristics or machine learning based node/branch selection.
- Parallel computing for solving MIPs: decomposition techniques, parallel computing in multiple searching tree, local search, etc.