Tommaso Mannelli Mazzoli

AI Architect at PORR AG, Vienna, Austria

← Back to collection

Validate a BDSP Solution

Check the cost and feasibility of a solution right here in your browser — your file is not uploaded anywhere. If it beats the best known solution (BKS), the results tell you how to submit it.

A file named after the instance (e.g. realistic_10_1.csv) selects it automatically.
Runs automatically once both are chosen.
Solution file format

A header-less CSV (.csv or .sol) binary matrix: row i is employee i, and column j is 1 if leg j is assigned to that employee, 0 otherwise. Legs are ordered by start time (matching the instance JSON), so every row has one column per leg. Every leg must be covered exactly once; all-zero rows are skipped.

1,0,1,0,0
0,1,0,0,1
0,0,0,1,0
How the cost is computed

Each employee (shift) costs

\[ O_e = 2W' + T + \text{ride} + 30\cdot\text{changes} + 180\cdot\text{splits}, \quad W' = \max(\text{work\_time},\; 390) \]

and the total is the sum over employees. Employees that break a hard constraint (driving, working or total time, rest breaks) add \(1000\times\) penalties. Full definitions are on the problem reference page.

Use your own instance file

Validate against an instance JSON that is not in the collection (no BKS comparison):

Submitting a new best known solution

When your solution is feasible and strictly better than the BKS, the results show a Submit box: download the correctly named file, open the pre-filled GitHub issue, attach the file and submit. A bot re-checks it and publishes it within minutes, credited to your GitHub account. You need a (free) GitHub account. See the submission guide for details, including submitting from the command line.