Category: Uncategorized

Attempted to fix bug in score calc for complicated factor analytic modelsAdd multi-trait incidence counts of traits summary to .res fileFixed bug when path of working folder was more than 70 characters;limit is now 120

Fixes recent bug in processing VPREDICT units

The Mac ARM 64 build is now built with the OPENBLAS library, comparable with the INTEL ONEAPI MKL library used for the other platforms. Added !DF f and !YSS s qualifiers (as in ASReml) which allow adjustment of the residual degrees of freedom (by f) […]

For weighted analyses, whether (DH)GL(M)M or not, the base assumption isthat the weights are the inverse of the true variance, and therefore do notrequire a scaling (heterogenity/dispersion) factor . In line with ASReml 4.3,a scaling factor can be requested now via the RESIDUAL statement.(It was […]

Bug parsing at(Trait 1).(Year lin(Cal) ) fixedBug processing Selection !A 454 !L filteredped.csv !LSKIP 1 Record !A 1891 !L weatherped.csv !LSKIP 1 resolved

Fix bug recognising content of .grr file (marker data)

Fix bug parsing at(Trait,1).(Year, sin(Month,12), cos(Month,12)) !R +Fix bugs fitting DHGLM models: milk dev(milk) ~ Trait.DIM !r diag(Trait).spl(DIM)Fix bug fitting diag(T).id(5)Fix bugs in GAMMA dispersion modellingImprove DHGLM model fittingAdd !NOGRAPHICS (-n) qualifier (option) to facilitate Batch processing under Unix

Fixed some bugs parsing complex models

Fixed bug affecting large dense models; miss allocation of workspace used when forming sparse C inverse resulted in miscalculation of Scores and hence lack of proper convergence.

The initial build of Echidna (1.94) for Apple Silicon Mac computers ran 5-10 times slower than on machines with INTEL chips (compiled using the Intel compiler using the Intel MKL library). I have written my own version of the BLAS/LAPACK DGEMM subroutine based on the […]