xyxxy.bsky.social
@xyxxy.bsky.social
10 followers 14 following 19 posts
Refugee from musky ol' Twitter.
Posts Media Videos Starter Packs
...Existing scripts have recently been updated to reflect current recommended best practices in doing mxGREML analyses, which are described in the README file. (n+2)/n
Here is a git repository of R scripts that demonstrate the use of OpenMx's GREML feature: github.com/RMKirkpatric... . It already has some scripts that demonstrate the new GREML capabilities in this OpenMx release, and more such scripts will be coming soon... (n+1)/n
GitHub - RMKirkpatrick/mxGREMLdemos
Contribute to RMKirkpatrick/mxGREMLdemos development by creating an account on GitHub.
github.com
Again, read the full announcement of new features, bug-fixes, known issues, GREML-related details, and NPSOL availability here: openmx.ssri.psu.edu/node/4942 n/n
OpenMx 2.22 released! | OpenMx
openmx.ssri.psu.edu
The big non-GREML news is that OpenMx can now automatically compute analytic first & second derivatives of the ML fitfunction for some RAM models. We recommend enabling these derivatives for models that have a large number of manifest variables and free parameters. 6/n
Other GREML news includes partial support for use of `mxCheckIdentification()` and `mxAutoStart()` with a GREML expectation; see under "Known Issues" for details. 5/n
Thus, the following are now possible in an mxGREML model: regressions ("one-headed paths") or nonrecursive paths ("feedback loops") between endogenous phenotypes; accommodating heteroskedasticity; regressing out covariates via a nonlinear or non-identity-link function! 4/n
...you can model the mean "explicitly", by naming an arbitrary vector-valued MxMatrix or MxAlgebra to be the phenotypic mean vector (similar to the MxExpectationNormal case)! 3/n
First, you don't have to use REML to fit a "GREML" model anymore--you can fit it by ordinary ML instead, with argument `REML=FALSE`! Second, if you're using ordinary ML, instead of modeling the phenotypic mean "implicitly" by naming exogenous covariates... 2/n
The OpenMx Team is pleased to announce the release of OpenMx version 2.22!: openmx.ssri.psu.edu/node/4942

The big news in this release relates to GREML (Genomic-relatedness-matrix REstricted Maximum-Likelihood): 2 bug-fixes, a patched memory leak, and 2 new features! 1/n
OpenMx 2.22 released! | OpenMx
openmx.ssri.psu.edu
I'm co-author on a paper accepted at _Psychometrika_ about how OpenMx can verify whether or not your model is identified, even if your model has nonlinear constraints and/or "definition variables"!: www.cambridge.org/core/journal...
Show Me Some ID: A Universal Identification Program for Structural Equation Models | Psychometrika | Cambridge Core
Show Me Some ID: A Universal Identification Program for Structural Equation Models
www.cambridge.org
Reposted
🚨BREAKING. From a program officer at the National Science Foundation, a list of keywords that can cause a grant to be pulled. I will be sharing screenshots of these keywords along with a decision tree. Please share widely. This is a crisis for academic freedom & science.
I think the most exciting new feature is that `mxCheckIdentification()` is now compatible with definition variables! 4/5
Of course, non-NPSOL builds of the new version, including for macOS, are available the usual way from CRAN. 3/5
As usual, we provide our own NPSOL-enabled builds of the new version for x86_64 Windows and x86_64 Linux/GNU. However, our own builds for macOS are not yet available, and v2.21.8 was the last version to support 32-bit (x86) Windows. 2/5
Today is the 40th anniversary of the disappearance of Will Byers.
#strangerthings
you can never have too many external hard drives
_Tales of Phantasia_ is an underrated 16-bit JRPG.

www.youtube.com/watch?v=W3SA...