Abstract:We present parametric breadth-first search (PBFS), a new algorithm for solving the parametric minimum cut problem in a network with source-sink-monotone capacities. The objective is to find the set of breakpoints, i.e., the points at which the minimum cut changes. It is well known that this problem can be solved in the same asymptotic runtime as the static minimum cut problem. However, existing algorithms that achieve this runtime bound involve fairly complicated steps that are inefficient in practice. PBFS uses a simpler approach that discovers the breakpoints in ascending order, which allows it to achieve the desired runtime bound while still performing well in practice. We evaluate our algorithm on benchmark instances from polygon aggregation and computer vision. Polygon aggregation was recently proposed as an application for parametric minimum cut, but the monotonicity property has not been exploited fully. PBFS outperforms the state of the art on most benchmark instances, usually by a factor of 2-3. It is particularly strong on instances with many breakpoints, which is the case for polygon aggregation. Compared to the existing min-cut-based approach for polygon aggregation, PBFS scales much better with the instance size. On large instances with millions of vertices, it is able to compute all breakpoints in a matter of seconds.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: in a network with source - sink monotonic capacity, how to effectively solve the parameterized minimum cut problem and find all the breakpoints where the minimum cut changes.
### Specific problem description
1. **Parameterized minimum cut problem**:
- In some applications, the capacity of an edge is not static but a function of a certain parameter \(\lambda\). As \(\lambda\) changes, the minimum cut also changes.
- The goal of the parameterized minimum cut problem is to calculate the minimum cut corresponding to each possible value of \(\lambda\) within a given parameter range.
2. **Source - sink monotonic parameterized minimum cut (MPMC) problem**:
- In the MPMC problem, the capacity of the edges connected to the source is non - decreasing with \(\lambda\), the capacity of the edges connected to the sink is non - increasing with \(\lambda\), and the capacities of other edges remain unchanged.
- In this case, the solutions of the minimum cut are nested: for any \(\lambda_1 < \lambda_2\), the source component of the minimum cut corresponding to \(\lambda_1\) is completely contained in the source component of the minimum cut corresponding to \(\lambda_2\).
- A breakpoint refers to the value of \(\lambda\) when the minimum cut changes.
3. **Deficiencies of existing algorithms**:
- Although existing algorithms can theoretically achieve the same asymptotic running time as the static minimum cut, they are often too complex and inefficient in practical applications.
- For example, the GGT algorithm finds breakpoints by bisection method, but it needs to maintain multiple flows and contracted graphs, resulting in a long actual running time.
### Main contributions of the paper
The paper proposes a new algorithm - Parameterized Breadth - First Search (PBFS) to solve the MPMC problem. The main features of PBFS include:
- **Discover breakpoints in ascending order**: PBFS gradually discovers breakpoints in the order of \(\lambda\) from small to large, which enables more efficient flow updates.
- **Simple and efficient implementation**: Compared with the GGT algorithm, the implementation of PBFS is simpler and shows better performance in actual tests.
- **Good scalability**: Especially when dealing with large - scale instances, PBFS performs excellently. For example, when dealing with polygon aggregation instances containing millions of vertices, PBFS can calculate all breakpoints within a few seconds.
### Application scenarios
The paper also shows the superior performance of PBFS in application scenarios such as polygon aggregation and computer vision. Especially for the polygon aggregation problem, PBFS has a significant advantage over existing methods when dealing with large - scale instances.
In summary, this paper aims to provide a simpler and more efficient algorithm to solve the parameterized minimum cut problem, and pays special attention to the source - sink monotonic parameterized minimum cut problem, thereby providing better support for applications in related fields.