A Large Software Company develops its own social network. Analysts have found that during the holidays, major sporting events and other significant events users begin to enter the network more frequently, resulting in great load increase on the infrastructure. As part of this task, we assume that the social network is 4 n processes running on the n servers. All servers are absolutely identical machines, each of which has a volume of RAM of 1 GB = 1024 MB (1) . Each process takes 100 MB of RAM on the server. At the same time, the needs of maintaining the viability of the server takes about 100 more megabytes of RAM. Thus, each server may have up to 9 different processes of social network. Now each of the n servers is running exactly 4 processes. However, at the moment of peak load it is sometimes necessary to replicate the existing 4 n processes by creating 8 n new processes instead of the old ones. More formally, there is a set of replication rules, the i -th ( 1 ≤ i ≤ 4 n ) of which has the form of a i → ( b i , c i ) , where a i , b i and c i ( 1 ≤ a i , b i , c i ≤ n ) are the numbers of servers. This means that instead of an old process running on server a i , there should appear two new copies of the process running on servers b i and c i . The two new replicated processes can be on the same server (i.e., b i may be equal to c i ) or even on the same server where the original process was (i.e. a i may be equal to b i or c i ). During the implementation of the rule a i → ( b i , c i ) first the process from the server a i is destroyed, then appears a process on the server b i , then appears a process on the server c i . There is a set of 4 n rules, destroying all the original 4 n processes from n servers, and creating after their application 8 n replicated processes, besides, on each of the n servers will be exactly 8 processes. However, the rules can only be applied consecutively, and therefore the amount of RAM of the servers imposes limitations on t