paminco.algo
Algorithms¶
The following main algorithms are implemented in paminco. While EFA computes an exact parametric minimum cost flow function for piecewise quadratic edge cost, MCA and MCFI both output a alpha-beta-approximate minimum cost flow function.
Abbrev |
Full name |
Short summary |
Source of approximation error |
---|---|---|---|
Electrical Flow Algorithm |
Compute an optimal potential function \(\lambda \rightarrow \mathbf{\pi}(\lambda)\) for linear piecewise marginal costs that induces a parametric minimum cost function \(\lambda \rightarrow \mathbf{x}(\lambda)\). |
/ |
|
Marginal Cost Approximation |
Interpolate marginal edge cost and run EFA with piecewise quadratic cost . |
Linear interpolation of marginal edge costs \(f_e\). |
|
Minimum Cost Flow Interpolation |
Repeat: 1) find next parameter \(\lambda\) and 2) calculate minimum cost flow \(\mathbf{x}_{\lambda}\). Then: interpolate between breakpoint solutions. |
Interpolative nature and optimizer tolerance \(\epsilon\). |