Optional MPLE existence check?

[mbojan]

Some tests with bigger data show that MPLE existence checking in

ergm/R/ergm.mple.R

Line 75 in 1f4401e

if(! control$MPLE.type%in%c("penalized","logitreg")) mple.existence(pl)

is time-consuming and memory-hungry. It's an LP so I guess there are little chances of making it leaner (I'm not sure if there are LP implementations for sparse matrices etc...?). Perhaps we can make a control option to disable it on demand?

[krivit]

We should be able to add a control parameter for that, sure.

[krivit]

@mbojan , can you come up with a heuristic for determining whether a check would be too long? Right now, you can turn it off manually, but it would be nice to autodetect the conditions instead.