Author: Vaibhav Gosain
Editorialist: Vaibhav Gosain
maxflow, min cut: Topcoder Tutorial
This problem can be solved via the concept of minimum cut.
We model the following graph, containing M+2 layers:
0th and (M+1)th layer for source and sink, M other layers for each role in firm, each containing N nodes.
- Among all pairs of nodes i,j within the same layer, an edge of capacity D[i][j] from j to i.
- Edge from source to all nodes of layer 1, with capacity INF.
- Edge from all nodes in layer M to sink , with capacity MAX-P[i][m].
- Edge from all nodes in layer (j-1) to corresponding nodes in layer j with capacity P[i][j-1] for 2<=j<=M
- Edge from all nodes in layer j to corresponding nodes in layer (j-1) with capacity INF for 2<=j<=M
where MAX = maximum possible value of P[i][j]
The required maximum productivity of the firm = N*MAX - mincut
Why does this work?
Let S be the source and T be the sink.
The infinite edge from layer j to layer j-1 ensures that if ith node in layer j is on the S side in the cut, corresponding node in layer j-1 will also be on the S side in the cut.
Now, say the last layer whose ith node lies in S side of the cut is R[i]. We claim N*MAX - (cut of above graph) is the value of total productivity of the firm if person i is given role R[i] for all i.
- Productivity contributed by P[i][R[i]] is due to the edge between layer R[i] and layer R[i]+1.
- For any 2 people i and j such that R[j] > R[i], there will be R[j]-R[i] layers for which node j is on S side of cut and node i is on the T side. Hence the value D[i][j] will be added R[j]-R[i] times to the mincut.
Author’s solution can be found here.