Skip to content
Draft
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
The table of contents is too big for display.
Diff view
Diff view
  •  
  •  
  •  
29 changes: 29 additions & 0 deletions PDPTW/README.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,29 @@
# PDPTW

This folder contains pickup and delivery problem with time windows (PDPTW) instances from two benchmark sets: Li & Lim (2003) and Sartori & Buriol (2020).
The original instances are modified to fit PyVRP's VRPLIB format.
The solution cost represents the total distance traveled.
All files were retrieved on 8 July 2026.

The instances are organised into two top-level folders, `li/` and `sartori/`, each with a subfolder per instance size.

## Li & Lim

The `li/` folder contains the Li & Lim (2003) instances with 100, 200, 400, 600, 800 and 1000 customers.
Downloaded from [SINTEF](https://www.sintef.no/projectweb/top/pdptw/li-lim-benchmark/) and converted from the Solomon-style format.
Distances are Euclidean and follow PyVRP's `exact` convention: each edge distance is scaled by 1000 and rounded to the nearest integer.
The best-known solutions (BKS) are taken from the [SINTEF](https://www.sintef.no/projectweb/top/pdptw/) solution pages.
Two solutions did not have an explicit solution file: `lc1_4_4` and `lc2_4_4`.
We used multiple solvers to find the best-known solution for these two instances.

## Sartori & Buriol

The `sartori/` folder contains the Sartori & Buriol (2020) instances with 100 up to 5000 customers.
The instances are downloaded from the [Mendeley dataset](https://data.mendeley.com/datasets/wr2ct4r22f/2) and the solutions from the [companion GitHub repository](https://github.com/cssartori/pdptw-instances).
These instances provide an explicit integer distance matrix, so PyVRP's `none` rounding convention is used and distances are taken directly from the instance data.

### References

Li, H., & Lim, A. (2003). A metaheuristic for the pickup and delivery problem with time windows. International Journal on Artificial Intelligence Tools, 12(2), 173–186. https://doi.org/10.1142/S0218213003001186

Sartori, C. S., & Buriol, L. S. (2020). A study on the pickup and delivery problem with time windows: matheuristics and new instances. Computers & Operations Research, 124, 105065. https://doi.org/10.1016/j.cor.2020.105065
11 changes: 11 additions & 0 deletions PDPTW/li/100/lc101.sol
Original file line number Diff line number Diff line change
@@ -0,0 +1,11 @@
Route #1: 81 78 104 76 71 70 73 77 79 80
Route #2: 57 55 54 53 56 58 60 59
Route #3: 98 96 95 94 92 93 97 106 100 99
Route #4: 13 17 18 19 15 16 14 12
Route #5: 32 33 31 35 37 38 39 36 105 34
Route #6: 90 87 86 83 82 84 85 88 89 91
Route #7: 43 42 41 40 44 46 45 48 51 101 50 52 49 47
Route #8: 67 65 63 62 74 72 61 64 102 68 66 69
Route #9: 5 3 7 8 10 11 9 6 4 2 1 75
Route #10: 20 24 25 27 29 30 28 26 23 103 22 21
Cost 828937
Loading