Commit messages

April 12, 2014

jpritikin committed r3334 in OpenMx at 14:55:
Place BA81 expectation in observed mode by default
in /trunk:

April 11, 2014

jpritikin committed r3328 in OpenMx at 18:23:
Rename observed.fit to observedFit and require a name instead of MxCompute
in /trunk:
jpritikin committed r3327 in OpenMx at 18:23:
Rename free.set to freeSet
in /trunk:
jpritikin committed r3326 in OpenMx at 18:23:
ComputeEM: remove post.mstep argument
in /trunk:

April 6, 2014

jpritikin committed r3294 in OpenMx at 12:53:
Newton-Raphson rewrite
in /trunk:

April 3, 2014

tbrick committed r3267 in OpenMx at 14:58:
Replaced @ with $ in demos and tests.
in /trunk:

March 19, 2014

jpritikin committed r3160 in OpenMx at 09:10:
Reparameterize lower & upper drm bounds in logits (copied from mirt)
in /trunk:

March 13, 2014

jpritikin committed r3141 in OpenMx at 16:32:
Re-architect fitfunction derivatives API using Eigen

Why do we need another matrix algebra library? We already have two, that
is, omxMatrix (original) and Matrix (from CSOLNP). For IFA models with
many items, it is essential to perform a sparse matrix-vector product
(Hessian %*% gradient) in Newton-Raphson. I initially wrote bespoke code
for sparse matrix-vector product. This was working well. However, it
came to my attention that inverting the Hessian can also benefit from
sparse matrix algebra. Rather than re-invent the wheel, Eigen looks like
a promising implementation.

Additional changes:

+ Internal derivatives are no longer reported back to R. You need to use
mxComputeReportDeriv to request them. Report derivs by default seemed
like a bad idea if they are so big that we are using a sparse
representation.

+ MxComputeNumericDeriv got a verbose parameter to enable debugging at
runtime.

+ ifa-drm-wide.R is moved to the failing directory temporarily. This
commit does not optimize manipulation of the Hessian but uses a simple
dense representation. dsytrf/dsytri are used to invert the Hessian. This
doesn't scale, but the improved accuracy results in many fewer
Newton-Raphson iterations highlighting the poor accuracy of the replaced
code.

+ Eigen has a great debug mode that NaN initializes memory and does
bounds checking. With these tools to assist debugging, I decided to
never store the lower triangle of a Hessian.
in /trunk:

March 7, 2014

jpritikin committed r3129 in OpenMx at 15:03:
Rearrange ComputeEM for SEM integration
in /trunk:

March 6, 2014

jpritikin committed r3123 in OpenMx at 19:49:
Add tests for convergence speed
Syndicate content