Community detection algorithm is very important to parallel computing and Internet structure analysis.This paper designs a new algorithm for Community Discovery based on building the various forces in network making the network movie randomly by Hooke s Law and Coulombs Law

then reach a balance by a variety of forces

this balance is used to map network to two-dimensional vector.And then the nodes of original network become the data points in 2d plane.This way can compress(n+m)network structure information down to 2n(Here n is the number of nodes in the network

m is the number of edges in the network)

and then make cluster algorithm to carry out the division of these 2d data points.In this paper

the algorithm is compared with the traditional GN algorithm

the experiment indicates that new algorithm can match the efficient GN algorithm

and can get more efficient and accurate by selecting some sample of data points in the plane.The algorithm provides a new way to partition network and takes a new way to study the balance of network.