Take abs(ev) when inverting Hessian (suggested by Timo von Oertzen)
[openmx:openmx.git] / src / Compute.cpp
2014-05-03 jpritikinTake abs(ev) when inverting Hessian (suggested by Timo...
2014-05-02 jpritikinSparse invert, handle the non-overlapping case
2014-04-28 jpritikinIgnore non-convergence in ComputeEM M-step
2014-04-28 jpritikinImprove diagnostics
2014-04-28 jpritikinPlace likelihood-based CI code into a separate MxComput...
2014-04-17 jpritikinRemove unused arguments from omxRaiseError, omxRaiseErrorf
2014-04-12 jpritikinTidy mxComputeEM user interface, add doc
2014-04-11 jpritikinRename observed.fit to observedFit and require a name...
2014-04-11 jpritikinRename free.set to freeSet
2014-04-11 jpritikinComputeEM: remove post.mstep argument
2014-04-11 jpritikinCheckpoint rewrite
2014-04-06 jpritikinNewton-Raphson rewrite
2014-04-06 jpritikinRework reporting of iterations and optimizer status...
2014-04-04 jpritikinmxDataDynamic
2014-04-03 jpritikinHook up recordIterationError for FIML
2014-04-03 jpritikinRemove lock from FitContext::resetIterationError
2014-04-03 jpritikinAdd FitContext::recordIterationError
2014-03-27 jpritikinmxCI should not interfere with SEs (or fit)
2014-03-26 jpritikinMake S-EM asymmetric Hessian available in debug slot
2014-03-24 jpritikinEliminate a few more memory leaks
2014-03-24 jpritikinPlug HessianBlock memory leaks
2014-03-24 jpritikinAddress memory corruption in ComputeEM debug output
2014-03-22 jpritikinChange paramInfo.type from integer to string
2014-03-22 jpritikinFix loop indexing for IFA derivs
2014-03-22 jpritikinDisentangle Ramsay1975 caution settings (maybe improves...
2014-03-21 jpritikinHelp gcc issue more accurate warnings
2014-03-21 jpritikinFix uninitialized memory read in multivariate normal...
2014-03-20 mhunterFinish killing %lu related debugging compiler warnings.
2014-03-16 jpritikinAdd protect to MxRList and simplify API
2014-03-15 jpritikinFactor out lots of calls to Rf_mkChar
2014-03-13 jpritikinRe-architect fitfunction derivatives API using Eigen
2014-03-12 jpritikinEnable R_NO_REMAP for a cleaner namespace
2014-03-09 jpritikinRemove non-reproducable pointer addresses from logs
2014-03-09 jpritikinRemove deadcode (dead after free.set fixes)
2014-03-07 jpritikinRearrange ComputeEM for SEM integration
2014-03-07 jpritikinAdd a 3rd string argument to omxExpectation computeFun...
2014-03-07 jpritikinFix various bugs in free.set variable groups
2014-03-06 jpritikinFix uninitialized memory read in Ramsay1975
2014-03-03 jpritikinAllow simultaneous computation of gradient and gradient...
2014-03-02 jpritikinFactor out code to force a matrix to be positive semi...
2014-02-28 jpritikinImprove mxComputeOnce, mxComputeEM APIs
2014-02-27 jpritikinRename EM fit functions to match DLR (1977) terminology
2014-02-27 jpritikinMove positive definite check from ComputeSE to ComputeEM
2014-02-27 jpritikinLots of work on SEM
2014-02-20 jpritikinAlways report @output$fit but don't report @output...
2014-02-16 jpritikinRename ComputeEstimatedHessian to ComputeNumericDeriv
2014-01-27 jpritikinRevive latent distribution gradients
2014-01-23 jpritikinPermit rescaling of log-likelihood
2014-01-17 jpritikinAdd more matrix operations
2014-01-17 jpritikinFactor out code to invert the Hessian
2014-01-17 jpritikinRewrite second order test; rename to ComputeHessianQuality
2014-01-07 jpritikinRevert "Fix standard error calculation"
2014-01-05 jpritikinReport Hessian consistently
2014-01-05 jpritikinFix standard error calculation
2014-01-02 jpritikinConsolidate reporting of the gradient
2014-01-02 jpritikinConsolidate reporting of the fit
2014-01-02 jpritikinReport estimate from 1 place
2014-01-02 jpritikinAllow MxCompute C objects to set values in their R...
2013-12-31 jpritikinSEM reorg
2013-12-31 jpritikinSEM debugging
2013-12-31 jpritikinCompute condition number of information matrix
2013-12-31 jpritikinSEM create pseudo-history if history is not working
2013-12-31 jpritikinRe-think semantics of COMPUTE_INFO
2013-12-31 jpritikinPartially fix sandwich
2013-12-31 jpritikinRework SE reporting
2013-12-30 jpritikinRevert "Move some result reporting to glue"
2013-12-30 jpritikinMove some result reporting to glue
2013-12-17 jpritikinRemove latent distribution gradients
2013-12-17 jpritikinMaybe sandwich working
2013-12-17 jpritikinGive up immediately if iteration limit is reached
2013-12-16 jpritikinImprove log message
2013-12-07 jpritikinSplit SE calculation into separate Compute object
2013-11-13 jpritikinGeneric ComputeEM with Ramsay (1975) acceleration
2013-09-27 jpritikinReport initial fit value when iterating
2013-09-12 jpritikinImprove error message
2013-08-22 jpritikinRemember N-R max caution across E-M cycles
2013-08-17 jpritikinSparse matrix vector product
2013-08-17 jpritikinCombine calculations into the E-step (helps two-tier...
2013-08-17 jpritikinAllow ComputeIterate to test maximum absolute change
2013-08-05 jpritikinDon't copy derivs between FitContext instances (by...
2013-08-04 jpritikinAuto-tune Newton-Raphson
2013-08-04 jpritikinInvert Hessian piecewise
2013-08-01 jpritikinProper handling for an empty free.set
2013-07-29 jpritikinAvoid final copyParamToModel when not needed
2013-07-29 jpritikinFix Hessian to work with equality constraints
2013-07-28 jpritikinImprove diagnostics
2013-07-27 jpritikinMerge branch Csolnp
2013-07-26 jpritikinRewrite Newton-Raphson with better math
2013-07-26 jpritikinRework initialization of starting values
2013-07-25 jpritikinFix/remove improper printf style formats
2013-07-22 jpritikinLogging levels, not just TRUE/FALSE
2013-07-22 jpritikinRename start to adjustStart
2013-07-18 jpritikinAll compute objects need a varGroup
2013-07-18 jpritikinRewrite dependency tracking
2013-07-16 jpritikinGradients for the latent distribution
2013-07-11 jpritikinAdjust BA81 expectation context argument
2013-07-10 jpritikinRework specification of free param groups per Tim's...
2013-07-08 jpritikinUpdate diagnostics
2013-07-07 jpritikinPartially fix reporting for custom Compute plans
2013-07-07 jpritikinRemove most instances of setFinalReturns
next