RMKirkpatrick
RMKirkpatrick
See d400ab6 . I disabled the warning about excess equalities that's raised during the ComputeNumericDeriv step. Eigen's 'FullPivHouseholderQR' class gives false alarms about rank-deficiency, at least at its default threshold...
623057c is relevant, but does not completely resolve this issue.
Well, actually, 623057c *does* resolve what I thought was the *only* issue. But, it still doesn't get `imxRowGradients()` working with OP's model in [the referenced forums thread](https://openmx.ssri.psu.edu/comment/8493). The real problem...
I changed the name of this issue to better reflect what is still unresolved.
Not to my knowledge.
> Can we close this after R drops support for win32? Probably. We'll see.
> ```r > Moving type forward should help users find it. Hiding parameters seems seldom a good idea, as they aren't document in the Rd where people want to find...
> It seems that we should check for active bounds if a code 5 or 6 is found, and only issue this note if a bound is seen to be...
It looks like we do have safeguards against this issue causing a spurious status 6, but as my script in the OP demonstrates, we do not have safeguards against spurious...