OpenMx 0.2.8 (standard errors news)
- Read more about OpenMx 0.2.8 (standard errors news)
- Log in or register to post comments
This version is primarily a bug fix release. Here is the full change log:
people are referring to GenEpiHelperFunctions.R without a path. This means we can't know what version they have, nor can anyone not au-fee with the source of that file use the posted code.
To keep the file up to date, and accessible, should we move it into the svn under a directory like "libraries" or "user", or even "demo"?
That would mean all users would get it, and it would be kept up to date.
Otherwise perhaps encourage something like
source("http://www.vipbg.vcu.edu/~vipbg/Tc24/GenEpiHelperFunctions.R")
Please see the comment in thread http://openmx.psyc.virginia.edu/thread/394#comment-1355 . There is definitely at least a cosmetic bug and quite possibly something more serious. It's critical as this is an example for the workshop next week.
Lots of changes in this new release. A high level summary -- (1) the back-end can now explicitly calculate the Hessian matrix and standard error estimates. This functionality has been disabled by default for the current binary release. It will be enabled by default in the next binary release. See ?mxOption for explicitly enabling these options. (2) omxMnor and omxAllInt have been implemented. See ?omxMnor and ?omxAllInt. (3) Matrix subranges and the range operator have been implemented.
Since we have implemented mixture distributions and categorical data, how about we change the list of things people can vote for? Let's delete those two and add joint analysis of ordinal and continuous data analysis.
Running a model reported
frontend elapsed time: 0.7429769 secs
backend elapsed time: 2.985495 mins
looks ok but in fact I system.time'd the run and it reported:
> system.time(multivTwinSatFit<-mxRun(multivTwinSatModel))
Running multivTwinSat
user system elapsed
177.301 0.608 179.884
so there are some 175 seconds missing from the tally.
Also, the results from a run with @independent=T under snowfall are different and seem to be basically reporting zero seconds for backend elapsed time.
I did an svn up and sudo make install to r1073, with no errors. Issue #1 is
i) The version number in summary() is stuck at 1050:
> openmx version number: 0.2.5-1050
I noticed that the tar filename is also stuck at 1050 during the build:
> cd build; R CMD INSTALL OpenMx_0.2.5-1050.tar.gz
perhaps this propagates the version #?
> summary(mxRun(multivTwinSatModel))
Running multivTwinSat
The Major iteration limit was reached (Mx status BLUE).
name matrix row col Estimate Std.Error
1
2
180
Observed statistics: 960
Estimated parameters: 180
Degrees of freedom: 780
-2 log likelihood: 7259.518
Saturated -2 log likelihood: NA
numObs: 101
Chi-Square: NA
p: NA
AIC (Mx): 5699.518
OpenMx includes support for parallel processing by using the R package snowfall. Snowfall is an optional package for the library. If snowfall is loaded at runtime then independent models will be executed in parallel. Many thanks go to Greg Carey for writing the sequential version of the following simulation on the forums. The parallel version of the script can be found at https://openmx.ssri.psu.edu/repoview/1/trunk/demo/BootstrapParallel.R.