基于dijkstra和广度搜索的加权有向图有必经点的点对点的最短路径算法，路径必须经过要求的必经点，且不成环。

该算法采用了一种自适应调整的方法，经过多次迭代，使解收敛。 但只是寻找了一个略优的可行解（每次迭代都基于贪婪算法寻找），不能保证最优解。而且如果图过于稀疏，因为收敛速度过快，可能导致问题无解（收敛于一个局部最优解，没有经过所有点）。 解决相对稠密的图（每个点的平均出入度4以上），表现良好。

备注：里面有一个QT的工程，可以直接打开（源码里没用QT的库）-Weighted Dijkstra and breadth first search based on a shortest path algorithm must point to point to the map, the path must pass through points, and not a ring.

This algorithm adopts an adaptive method to adjust, after several iterations, the convergence of solution. But only to find a slightly better feasible solution (each iteration is based on the greedy algorithm to find), can not guarantee the optimal solution. And if the graph is too sparse, because the convergence speed is too fast, it may lead to the problem of no solution (convergence to a local optimal solution, not after all points). To solve the relatively dense graphs (each entry point average above 4), good performance.