APPLIED MATHEMATICS & INFORMATION SCIENCES
Because the existence of dependent relation between tasks，the resources assignment will have an impact on the other tasks in the grid dependent task scheduling process. Therefore, in order to improve the performance of grid application, grid dependent tasks scheduling generally adopts static scheduling strategy, this strategy is generate a global scheduling plan before runtime, and rescheduling at the time of resources changing.
In order to achieve the scheduling goal of optimal application performance, it needs global optimization method, the grid dependent tasks rescheduling will take all unfinished tasks as scheduling object, so the grid dependent tasks rescheduling problem is a NP complete problem. The solution to this problem will influenced by quantity of the gird application resources and tasks. If the quantity of the gird application resources and tasks is too large, the solution to this problem will requires a long time. At present, the study in this filed is using heuristic algorithm or AI optimization algorithm to improve the efficiency of the algorithm. However, rescheduling happens at application operating time, the executing efficiency of rescheduling algorithm is not only influence the cost of rescheduling，but also affect the performance of application, therefore, compared with the initial scheduling, rescheduling efficiency demanded more strict. In rescheduling triggered frequently cases，only improve the efficiency of rescheduling optimization algorithm is hard to ensure efficiency of the grid dependent tasks rescheduling, apart from this, also need to consider to reduce the size of the resources and tasks involved in rescheduling. However，in general sense，to reduce the size of the resources and tasks will influence the application performance. Therefore, how to determine the involved scope of rescheduling tasks is a key problem in the gird dependent tasks rescheduling.
In view of this problem, this paper analyses the degree of data transmission dependent between tasks, the degree of resource sharing conflicts between tasks and the degree of task support to performance optimization in dynamic environment, put forward rescheduling tasks spread domain computation method which initial scope is the set of rescheduling tasks as minimum required, oriented as the sharing conflicts and data transmission dependent between tasks, bordered by the degree of task support to performance optimization, defined with gradually expand the scope of the rescheduling task. To calculate by this method, suitable rescheduling tasks scope can be obtained, strongly influence each other within the scope, weakly influence each other without the scope, thus，the rescheduling of the tasks scope will improve the efficiency, without affecting the gird application optimization as much as possible .