ergm

ergm copied to clipboard

Benchmark different termination criteria (confidence, Hummel, etc.) and MCMC parameters (fixed vs ESS)

1

As of this writing, the default is to use "confidence" criterion, but it might be worth going back to "Hummel", and others might be worth trying as well. This choice...

krivit

Proposals in MHproposal_bipartite.c rely on the defunct discordant network API

1

The affected functions are `MH_BipartiteHammingConstantEdges()` and `MH_BipartiteHammingTNT()`. Both to be updated to use the storage or auxiliary storage API.

krivit

Currently, the test only computes summary for some randomly selected networks and computes the edge statistic.

krivit

Adaptive MCMC, especially with complex models, spends a lot of time calculating the ACF.

3

Here's an example of `Rprof` results for a recent run of a 60-parameter multilayer model: It was run on a cluster with `parallel=8`. The runtime was: ``` > proc.time() user...

krivit

Implement term labeler

1

Mentioned by @krivit in https://github.com/statnet/ergm/issues/217#issuecomment-765741463, we can extend `param_names()`. We can also have a dedicated `term_labels()` that creates unique labels for the terms. This is needed in any reporting context...

mbojan

Outsource MCMC standard error calculations to mcmcse::mcse.multi()?

5

Essentially, instead of `specturm0.mvar()`, we call that function; it seems to do everything we need except handle parallel chains, and it may well do so faster and more robustly. (At...

krivit

At this time, we represent the information needed to obtain the MPLE in three objects: * Covariates: u*p-matrix of doubles (change statistic vector for the dyad) * Response: u-vector of...

krivit

Implement an alias() method for ergm objects.

2

This could work in tandem with the existing nonidentifiability detection; in particular, the existing `alias.formula()` and `alias.lm()` methods provide more detail: not only which statistics are aliased by those before...

krivit

Back in 2016, we discovered that the `bd()` constraint that was enforced as follows: 1. Propose a toggle. 2. Check if the proposal violates the constraint. 3. If violates, repeat...

krivit

Improve documentation for new "levels" argument

4

There are a couple of problems with the current documentation -- which needs to be able to serve a wide range of users, from first-timers needing to understand the syntax...

martinamorris