Using broadcast variables in PySpark for GMM implementation

Question

So I\'m working on an implementation of a Gaussian Mixture Model on Spark and my question here is if it\'s a good idea or not to use broadcast variables.

The thing is