Inspired by scenarios where the strategic network design and defense or immunisation are of the central importance, Goyal et al. [3] defined a new Network Formation Game with Attack and Immunisation. The authors showed that despite the presence of attacks, the game has high social welfare properties and even though the equilibrium networks can contain cycles, the number of edges is strongly bounded. Subsequently, Friedrich et al. [10] provided a polynomial time algorithm for computing a best response strategy for the maximum carnage adversary which tries to kill as many nodes as possible, and for the random attack adversary, but they left open the problem for the case of maximum disruption adversary. This adversary attacks the vulnerable region that minimises the post-attack social welfare. In this paper we address our efforts to this question. We can show that computing a best response strategy given a player u and the strategies of all players but u, is polynomial time solvable when the initial network resulting from the given strategies is connected. Our algorithm is based on a dynamic programming and has some reminiscence to the knapsack-problem, although is considerably more complex and involved.
翻译:在战略网络设计和防御或免疫具有核心重要性的情景的启发下,Goyal等人[3] 定义了一个新的攻击和免疫网络形成游戏。作者们表明,尽管有攻击,游戏具有很高的社会福利特性,尽管平衡网络可以包含循环,但边缘数是紧密结合的。随后,Friedrich 等人[10] 提供了一种多元时间算法,用于计算最大屠杀对手的最佳反应战略,该对手试图杀死尽可能多的节点,以及随机攻击对手,但他们为最大破坏对手打开了问题。这个对手攻击了脆弱区域,最大限度地减少攻击后的社会福利。在本文中,我们讨论这个问题的努力。我们可以表明,计算出一个玩家的最佳反应战略,以及所有玩家的战略,但除了u,当最初的网络与特定战略相联系时,是多数值时间可以溶解的。我们的算法基于一个动态的编程,对 knapsack-Problem 有一些共鸣,尽管相当复杂和涉及。