This page comprises supplementary information for the Algorithm Selection Problem
(ASP) for large 2-Dimensional Strip Packing Problem (2SP).
More on the problem, test instance generation, algorithm portfolio construction
and evaluation, can be found in:
K.Piechowiak, M.Drozdowski, É.Sanlaville,
Framework of Algorithm Portfolios for Strip Packing Problem,
Available online 4 August 2022, 108538,
https://doi.org/10.1016/j.cie.2022.108538
Extended results - Research report
Example algorithm portfolios, and their performance scores, for the considered
problem are shown in this report:
K.Piechowiak, M.Drozdowski, É.Sanlaville,
Extended Performance Results of Algorithm Portfolios for 2D Strip Packing,
Institute of Computing Science,
Poznan University of Technology,
Research Report RA-4/2021, 2021.
Collections of the instances:
• dataset1 - 1st set of random instances (training).
• dataset2 - 2nd set of random instances (validation).
• real instances - collection of test instances published
in the erlier literature.
The instances are collected in the following files
| # | instances | comments |
|---|---|---|
| 1. | dataset1.zip | dataset1 ~6.8MB |
| 2. | dataset2.zip | dataset1 ~6.8MB |
| 3. | real-instances.zip | real instances ~88.3MB |
File Name Convention
• 0.in - 99.in -- dataset1 instances with n=10 items,
• 100.in - 199.in -- dataset1 instances with n=20 items,
• 200.in - 299.in -- dataset1 instances with n=50 items,
• 300.in - 399.in -- dataset1 instances with n=100 items,
• 400.in - 499.in -- dataset1 instances with n=200 items,
• 500.in - 599.in -- dataset1 instances with n=500 items,
• 600.in - 699.in -- dataset1 instances with n=1000 items,
• 700.in - 799.in -- dataset1 instances with n=2000 items,
• 800.in - 899.in -- dataset1 instances with n=5000 items,
• 900.in - 999.in -- dataset1 instances with n=10000 items,
• 1000.in - 1099.in -- dataset2 instances with n=10 items,
• 1100.in - 1199.in -- dataset2 instances with n=20 items,
• 1200.in - 1299.in -- dataset2 instances with n=50 items,
• 1300.in - 1399.in -- dataset2 instances with n=100 items,
• 1400.in - 1499.in -- dataset2 instances with n=200 items,
• 1500.in - 1599.in -- dataset2 instances with n=500 items,
• 1600.in - 1699.in -- dataset2 instances with n=1000 items,
• 1700.in - 1799.in -- dataset2 instances with n=2000 items,
• 1800.in - 1899.in -- dataset2 instances with n=5000 items,
• 1900.in - 1999.in -- dataset2 instances with n=10000 items.
• 2000.in - 3305.in -- real instances.
Mapping of literature names
mapping.json -- mapping between file number and
literature instance names
File Format
Instance files are text files with data in the following order:
- 1st line: W LB
- 2nd line: n
- lines 3 to n+2 - item data (in the order): width height
where:
n - number of items,
LB - height Lower Bound,
W - width of the strip.