Visualization of runtime behavior of branch-and-bound algorithms

Miyamura, Hiroko  ; Shinano, Yuji*; Miyashiro, Ryuhei*; Saito, Takafumi*

In branch-and-bound algorithms for integer programming, runtime behavior of the algorithms depends much on branching strategy. However, from a huge computation log of a large program, it is difficult to explore a key factor for effective branching. To analyze which factor of branching strategy is essential, we develop a system for visualization of growing process of a large branch-and-bound tree. The proposed system provides intuitive understanding how branching strategy affects branch-and-bound process.