A follow up to #126, which introduces PartitionedGraph wrapping an ITensorNetwork as a data structure to help with implementing the belief propagation code.
An unfortunate consequence of that design is that functions that work on regular ITensorNetwork may not work on one wrapped in a PartitionedGraph, such as:
[ ] scalartype
It would be good to assess how to make those functions work in an elegant way. scalartype could maybe work if we made ITensorNetwork iterable over the vertex data, in which case scalartype would iterate over the ITensors of the network.
A follow up to #126, which introduces
PartitionedGraphwrapping anITensorNetworkas a data structure to help with implementing the belief propagation code.An unfortunate consequence of that design is that functions that work on regular
ITensorNetworkmay not work on one wrapped in aPartitionedGraph, such as:scalartypeIt would be good to assess how to make those functions work in an elegant way.
scalartypecould maybe work if we madeITensorNetworkiterable over the vertex data, in which casescalartypewould iterate over the ITensors of the network.