0.12.22

Prelude

Binary and constrained quadratic model generators that used to reside in
dwave-networkx (now dwave-graphs) have been moved to
dimod.generators.

New Features

• Add generators for binary quadratic models with ground states
  corresponding to a vertex coloring
  (dimod.generators.vertex_coloring) and a minimum vertex coloring
  (dimod.generators.min_vertex_coloring).

• Add a binary quadratic model generator for a Markov network
  (dimod.generators.markov_network).

• Add generators binary quadratic models with ground states
  corresponding to a graph matching (dimod.generators.matching),
  graph's maximal matching (dimod.generators.maximal_matching) and
  graph's minimum maximal matching
  (dimod.generators.min_maximal_matching).

• Add a generator for a constrained quadratic model with ground states
  corresponding to a balanced k-partition of graph
  (dimod.generators.graph_partition).

• Add a binary quadratic model generator to calculate the structural
  imbalance of a signed social network
  (dimod.generators.structural_imbalance).

• Add a generator for a binary quadratic model with ground states
  corresponding to a minimum TSP route
  (dimod.generators.traveling_salesperson).

• Add support for NetworkX graph output during function argument
  coercion via @graph_argument(..., as_networkx=True).

• Add estimator for effective sample size based on
  Revisiting the Gelman-Rubin Diagnostic https://arxiv.org/abs/1812.09384.

Upgrade Notes

• Drop support for Python 3.9.

Bug Fixes

• To avoid @graph_argument(..., as_networkx=True) failing during
  import/compile time, defer networkx checks to the decorated function
  run time.