Partition of candidate splitting nodes

[Carex-Chunhui]

Hello!
I want to know whether the candidate splitting nodes are evenly divided by the global maximum and minimum in federated learning?
In addition, in this federated learning, whether the eigenvalues of each node need to be uploaded and how the histograms are aggregated in the aggregator？
I look forward to your reply very much. Thank you！