This is an example on how an readme file can look, see this file for info about how to contribute.
The paper consider 10 instances for the classical bi-objective linear assignment problem.
Instances are named Template_AP_n<n>.<raw/xml>
where n
is the size of the problem. The paper considers
instances of size 5-50; however, the instance set also contains 5 instances of size 60-100. Costs
are generated random in [0,30].
All instance files are given in both xml and raw format. The xml format is self explainable (see e.g. ex1).
We use the following parameter names:
-
$n$ = dimension/size -
$c^{k}_{r,c}$ =$k$ 'th cost of assigning row$r$ to column$c$ .
The instances have the following format:
n
c^{0}_{0,0}... c^{0}_{0,n-1}
c^{0}_{1,0}... c^{0}_{1,n-1}
...
c^{0}_{n-1,0}... c^{0}_{n-1,n-1}
c^{1}_{0,0}... c^{1}_{0,n-1}
c^{1}_{1,0}... c^{1}_{1,n-1}
...
c^{1}_{n-1,0}... c^{1}_{n-1,n-1}
That is, first the dimension, then the costs for the first criterion and next the cost for the second criterion.
The instances are contained in the sub folder MMAP
. Instances are named
Template_MMAP_d<n>_e<I1>_c<I2>_m<M>_s<S>_<Y>.xml
where Y
is the instance number of a BiMMAP of
size I1
and cost range I2
using method M
and shape S
.
A total of 8000 instances are provided. That is, 100 instances of each of the following 80 possible configurations were generated:
- n: 4, 6, 8, 10.
- I1: 2-8 (not for method 2) and 10-30.
- I2: 0-500 and 0-10000.
- (M,S): (1, −60), (1,0), (1, 60), (2, 3), (2, 4), (3, 0).
Restults are given in the results
folder using the json
format (see Step 3).