Mx RAM speedup comparison test

1 reply [Last post]
neale's picture
Joined: 07/31/2009


I tried running a model with vs. without a feedback loop. There is some penalty if there's feedback in the model, but not as much as I would have expected. This is in a system with 48 latent+observed variables, so I would have thought that inverting 2 48x48 matrices every iteration would have slowed things down a lot more. But perhaps the inversion routine is pretty smart about cases like this. Script attached.

> system.time(twinACEFit <- mxRun(twinACEModel))
Running twinACE
user system elapsed
239.818 0.429 240.693
> #summary(twinACEFit)
> ##Now add paths for feedback loop
> twinACEModel$MZ@matrices$A@values[1,2]<-.1
> twinACEModel$DZ@matrices$A@values[1,2]<-.1
> twinACEModel$MZ@matrices$A@values[7,8]<-.1
> twinACEModel$DZ@matrices$A@values[7,8]<-.1
> system.time(twinACEFitFeedback <- mxRun(twinACEModel))
Running twinACE
user system elapsed
253.815 0.474 253.921

Steve's picture
Joined: 07/30/2009
In the version you were

In the version you were running, I don't believe the solve(I-A) automatic speedup had yet been implemented. In fact, I believe it is still waiting to commit. Maybe Mike Spiegel and Tim Brick can give us some timeline on that tomorrow.