Set1: contains GRLP instances with unitary node costs with between 50 and 150 nodes.
Set2: contains weighted GRLP instances with between 50 and 150 nodes.
Set3: contains larger GRLP instances with unitary node costs (with between 175 and 500 nodes)

File names: for example "steiner_dataN50S13M253.dat" denotes a GRLP instance with 50 nodes, 13 s-nodes and 253 NDC pairs.
The corresponding weighted GRLP instance is called "weight_steiner_dataN50S13M253.dat".

Input format:
Input graphs are directed. Each s-node i is split into i' and i''.
Weights of all arcs are zero, except along the split nodes (where the cost of an arc (i',i'') represents the cost for installing a regenerator i).
Each t-node is split into t_s and t_t (representing the node t as being a source or a target of an NDC pair, respectively).
So, for example, for the instance N50S13M253, our input graph contains 100 nodes. First 26 nodes are split s-nodes, the remaining nodes are split t-nodes.

The format is:

<number of nodes>    <number of ndc pairs>
link
<link_index>    <source_node>    <target_node>    <weight>
...
<link_index>    <source_node>    <target_node>    <weight>
NDC
<source>    <target>
...
<source>    <target>