How can EIGRP topologies be designed to converge as fast as possible in the event of a point-to-point link failure?

How can EIGRP topologies be designed to converge as fast as possible in the event of a point-to-point link failure?

A. Limit the query domain by use of summarization. B. Limit the query domain by use of default routes. C. Build neighbor adjacencies in a squared fashion. D. Limit the query domain by use of distribute lists. E. Build neighbor adjacencies in a triangulated fashion.