dwave-examples/distributed-computing

Minimize messaging between computers in a distributed computing system by using the LeapHybridCQMSampler. This problem is also known as the graph k-partitioning problem.

distributed-computing's Issues

Not all invalid inputs are detected

The code in build_graph tries to detect and correct various cases, including cases where number of nodes is not divisible by number of partitions. However, it does this on a graph-specific basis (i.e., the logic is different for each value of args.graph), and some cases are missed. For example:

python demo.py -n 1001 -g internet

or

python demo.py -n 5 -g rand-reg

These inputs produce impossible problems because the partition size constraint is not an integer.

I would suggest trying to refactor some of the checks in build_graph, in particular doing a single check for divisibility that applies to all graph types. A lot of the checking could also be simplified by not trying to "fix up" invalid inputs -- that way, the actual values are known up front, and can be checked more easily. Otherwise, modifying the values on the fly makes the validation more difficult.

Add visualization

The README implies that some visualization is produced (aside from command-line output), but none is created. Could we add one?