Discrete Optimization ( IF 1.1 ) Pub Date : 2022-10-12 , DOI: 10.1016/j.disopt.2022.100743 Hua Chen , Lin Chen , Guochuan Zhang
We consider 4-block -fold integer programming, which can be written as , where the constraint matrix is composed of small matrices such that the first row of is , the first column of is , the main diagonal of is , and all the other entries are 0. There are copies of , , and . The special case where is known as -fold integer programming.
Prior algorithmic results for 4-block -fold integer programming and its special cases usually take , the largest absolute value among entries of , as part of the parameters. In this paper, we explore the possibility of solving the problems polynomially when the number of rows and columns of the small matrices are constant. We show that, assuming , this is not possible even if and . However, this becomes possible if or , or more generally if , and the rank of matrix satisfies that .
中文翻译:
块结构整数规划:我们可以在没有最大系数的情况下进行参数化吗?
我们考虑 4-block-fold 整数规划,可以写成, 其中约束矩阵由小矩阵组成这样第一行是,第一列是, 的主对角线是,所有其他条目为0。有的副本,, 和. 特殊情况被称为-折叠整数规划。
4块的先验算法结果-fold 整数规划及其特殊情况通常需要, 的条目中最大的绝对值,作为参数的一部分。在本文中,我们探讨了当小矩阵的行数和列数恒定时,多项式求解问题的可能性。我们证明,假设, 这也是不可能的,即使和. 然而,这成为可能,如果或者,或更一般地,如果,和矩阵的秩满足.