Integrating PPML analytical solution in to the trunk.
authordhackett <dhackett@df83fd30-4cca-4c36-ab9d-e8b5583ccbd1>
Wed, 4 Jan 2012 23:28:53 +0000 (23:28 +0000)
committerdhackett <dhackett@df83fd30-4cca-4c36-ab9d-e8b5583ccbd1>
Wed, 4 Jan 2012 23:28:53 +0000 (23:28 +0000)
commit1979931a4522d4ab07d84cb23ea650380350323a
tree966b2327e658998b8f643004315dc3caa944dd78
parent47d7e0cb21fee76d51b23df1c6ac8d420dc26510
Integrating PPML analytical solution in to the trunk.

Currently applicable to models specified in both path and matrix form, with either covariance matrix or raw data. Support for data sets with missingness in development.

For RAM models, if PPML is applicable in such a way that an analytical solution can be computed, the model will be transformed and solved.  Optimizations to models not analytically solvable are possible and in development, but not fully implemented yet.

Adds a new model option, UsePPML.  Can be set to "Yes" or "No".  mxDisablePPML(model) and mxEnablePPML(model) set this option.

PPML applicability filter has been separated off in to another function, imxCheckPPMLApplicable.  This function returns "Solve", "PartialSolve", "Split", or NA depending on how PPML is applicable to the given model.  The return value from this function can be passed to the transform function.  By default, however, the transform function will call imxCheckPPMLApplicable.

Following the analytical solution, the backend will be called for one cycle to compute Hessians and Standard Errors.

The UsePPML option is also used to track what PPML did to the model.  After the run is complete, it will have value "Solved", "PartialSolved", or "Split".

git-svn-id: http://openmx.psyc.virginia.edu/svn/trunk@1892 df83fd30-4cca-4c36-ab9d-e8b5583ccbd1
25 files changed:
R/MxOptions.R
R/MxPPML.R
R/MxRAMObjective.R
demo/LatentGrowthCurveModel_MatrixRaw.R
demo/LatentGrowthCurveModel_PathRaw.R
demo/LatentGrowthCurveModel_PathRaw_ModelRec.R
demo/LatentGrowthCurveModel_PathRaw_ObjectAdd.R
models/passing/LGC_MatrixCov.R
models/passing/LGC_PathCov.R
models/passing/PPML_100days.R [new file with mode: 0644]
models/passing/PPML_testFactorModelCov.R
models/passing/PPML_testFactorModelFakeLatent.R [deleted file]
models/passing/PPML_testFactorModelMatrixCov.R [new file with mode: 0644]
models/passing/PPML_testFactorModelMatrixCovMean.R [new file with mode: 0644]
models/passing/PPML_testFactorModelMatrixRaw.R [new file with mode: 0644]
models/passing/PPML_testFactorModelRaw.R [new file with mode: 0644]
models/passing/PPML_testLGC_MatrixCov.R [deleted file]
models/passing/PPML_testLGC_PathCov.R [deleted file]
models/passing/PPML_testMultiLayer.R [deleted file]
models/passing/PPML_testNHEVCov1.R [deleted file]
models/passing/PPML_testNHEVCov2.R [deleted file]
models/passing/PPML_testNHEVFakeLatent.R [deleted file]
models/passing/PPML_testNHEVMatrix.R [deleted file]
models/passing/PPML_testNHEVRaw1.R [deleted file]
models/passing/PPML_testNHEVRaw2.R [deleted file]