Changelog

nethist 1.0.0

update: 2026/08/04

New features

• multinethist()
  • New implementation of Song and Olhede (2026+) for multi-layer network histogram estimation.
  • common_f option assumes a common graphon across layers.
  • Accepts a wider range of inputs: 3D array, list of igraph/network objects, and ergm.multi::combined_networks object.
• nethist()
  • New method = "LSE" option, implementing Gao et al. (2015) for single-layer network histograms.
  • Now implemented as a single-layer wrapper around multinethist().
• hnethist(): new hybrid network histogram method for shape-based model selection.
• fitted(): new S3 method for nethist and multinethist objects.
• New dataset: IndianVil, a multilayer network.

Renamed functions

• summary() -> covariate_plot()
• violin_netsummary() -> netsummary_plot()

Bug fixes

• netsummary_plot() (formerly violin_netsummary()): return 0 instead of NaN when a subsampled network is empty.

Other changes

• netsummary_plot(): x-axis values are now shown as igraph plots of v-shapes and cycles instead of labels.
• Renamed output field LSE to MSE in nethist/multinethist objects.
• Documentation terminology unified from “node” to “vertex”.
• multinethist(): reject plain matrix/Matrix elements in list input, since vertex ordering across layers cannot be verified.

nethist 0.2.4

update: 2023/04/04

-bug fix in plot()

nethist 0.2.3

update: 2023/03/15

• Documentation updates
• Internal functions (not visible to users)

nethist 0.2.2

update: 2022/12/08

• plot.nethist()
  • You can display the estimated probabilities in p_mat of nethist objects by setting prob=TRUE. Use prob.col and prob.cex to set the color and size of text, respectively.
• nethist()
  • Check the convergence criterion every 5 iterations from the fifth iteration (5,10,15,…) instead of every 5 iterations from the first iteration (1,6,11,…).
  • minor speed improvement.

nethist 0.2.1

update: 2022/11/26

• Initialization step of nethist() is faster than earlier by computing hamming distance from the (simple) adjacency matrix instead of Manhattan distance.

nethist 0.2.0

update: 2022/11/19

• Computation speed of nethist() and violin_netsummary() are improved.

Bug fixes

• nethist(): computation of the diagonal part o in greedy search Rcpp routine is fixed.
• nethist(): computation of log-likelihood when there is a smaller size of group is fixed, in greedy search Rcpp routine.

nethist 0.1.1

update: 2022/11/08

• Documentations update: provide extra about default values of arguments, rewrite examples.

Bug fixes

• Resolving default value issue of R in violin_netsummary().

nethist 0.1.0

update: 2022/11/04

• Initial release of nethist R package (development)