OpenMx Developer Forums
pjohnson
Joined: 09/19/2009
You could adhere to R conventions more closely
Hi, I'm new here. I'm a political scientist/programmer. I was drawn in by conversation about SEM among my psych colleagues at the University of Kansas. Now that you have the gcc-4.3 version of lpsolve.la in your package, I can build OpenMx without any trouble on Ubuntu 9.04 with R 2.9.2
I've been reading through your OpenMxUserGuide and testing the examples and I'm having some reluctance about the style of your R interface/code. I'm pretty sure that R users who come to try OpenMx will have difficulty digesting the way you have designed your interface.
tbates
Joined: 07/31/2009
mxConstraint
a place to discuss mxConstraint
- Read more about mxConstraint
- 17 comments
- Log in or register to post comments
tbates
Joined: 07/31/2009
mxRun
A place to discuss mxRun
- Read more about mxRun
- 6 comments
- Log in or register to post comments
tbrick
Joined: 07/31/2009
Checkpointing Proposal
At the developer meeting today, we discussed adding checkpointing functionality to OpenMx.
This is basically a recovery system in case the computer crashes during an optimization. Essentially, the back-end will write a file to disk every x minutes or every z iterations. The file will contain some representation of the state of optimization at the time it saved. That way if the power fails or the computer crashes, the state of not-too-long-ago is saved somewhere, and the optimizer can pick up where it left off.
- Read more about Checkpointing Proposal
- 7 comments
- Log in or register to post comments
mspiegel
Joined: 07/31/2009
Shrinking the binary releases
I had an idea for shrinking the binary releases. At the moment, we ship the binary release for each platform with copies of NPSOL for all platforms, even though the binary release was built using only one of those copies. After release 0.1.4 has been made available, I'm going to play with the configure script such that only one copy of NPSOL is shipped.
- Read more about Shrinking the binary releases
- 3 comments
- Log in or register to post comments
mspiegel
Joined: 07/31/2009
Developers Meeting 9/18/09
It was noted that another release of the OpenMx closed beta library will be produced today.
There are a discussion of the next immediate implementation tasks. On the front-end, it was decided that square bracket notation should be implemented, and a tool from converting Mx 1.0 algebra expressions to OpenMx algebra expressions is needed. On the back-end, the most immediate goal is to implement some sort of checkpointing to disk.
- Read more about Developers Meeting 9/18/09
- Log in or register to post comments
stueller
Joined: 08/04/2009
"saving" the score/information functions
First of all I want to say that OpenMx has been quite straightforward to start using given the documentation on this website.
- Read more about "saving" the score/information functions
- 1 comment
- Log in or register to post comments
mspiegel
Joined: 07/31/2009
R substitute question
The following question seems unusual, but if someone figures out how to do it then we can do a single mxEval() call for all the algebras in a model to test the algebras for conformability during mxRun before the optimizer is called, instead of one mxEval() call per algebra. This could speed up our test suite (not by much, expect something like 5 - 10% increase).
https://stat.ethz.ch/pipermail/r-help/2009-September/211731.html
- Read more about R substitute question
- Log in or register to post comments
tbates
Joined: 07/31/2009
Save the state of the optimiser for input to a new run (granted)
For lots of jobs, the ability to input the optimiser state from a previous run would be a massive productivity gain.
The cleanest way to implement it for the user would be as a parameter to mxRun, where you pass in the last fit, and this has in it the state of the optimiser which mxRun pushes into the current optimiser run as a set of start values.
fit = mxRun(model)
mxRun(model, fit)
#--> returns almost instantly having been started at the optimal location
tbates
Joined: 07/31/2009
Confidence intervals (granted)
Confidence intervals are basic to most results sections now, so implementing Mx's 'Intervals' likelihood-based confidence intervals (Neale & Miller, 1997) (plus standard error-based CIs for speed) would be high on the wish-list
maybe something like
mxInterval(..., percent=95, type="ML") where ... consists of matrix addresses
mxInterval(z.A) would mean all cells of A
mxInterval(z.A[1:2,], type="SE") first two rows of A, standard-error based method
- Read more about Confidence intervals (granted)
- 31 comments
- Log in or register to post comments