| import OAI.Combinatorics.ContingencyTables.FactorialLift | |
| import OAI.Combinatorics.ContingencyTables.UnconditionalMain | |
| /-! | |
| # Sampling and counting contingency tables | |
| Almost-uniform sampling has polynomial cost on every execution, while | |
| exact uniform sampling has polynomial expected bit cost. Cell-bounded | |
| counting admits a fully polynomial randomized approximation scheme, | |
| including structural zeros and infeasible inputs. The finite-box | |
| Prékopa–Leindler inequality, finite Cheeger inequality and shortest | |
| augmenting path bound supply the analytic and feasibility estimates. | |
| -/ | |