0.9-52 - fixed input check for maxiter (vector logicals) - reactivated globaltest in vignette 0.9-51 - Fixed bug with non-compatible matrices - update to incorporate new length.Surv - made globaltest use conditional 0.9-50 - Fixed bug with baseline hazard returning NA - increased minsteps default in profL1 0.9-49 - Parts of the package rewritten in C++ by Matthew Lueder 0.9-48 - Fixed error in profL2 in case unpenalized =~0 (Thanks to Shixiao Cui for the bug report) - Fixed error in predict when using a matrix in unpenalized (thanks to Pilipp Deutsch for the bug report) 0.9-47 - fixed error in predict arising from earlier fix (Thanks to Bernd Bischl for the bug report) - made addition of response column in data not nessary for predict (Thanks to Bernd Bischl for the bug report) 0.9-46 - fixed error in predict() if unpenalized of the form ~0+X (Thanks to mark van de Wiel for the bug report) - updated namespace to new standards 0.9-45 - minor changes to DESCRIPTION 0.9-44 - fixed error in predict penfit.R in the presence of an offset term - .ridge takes a half Newton-Raphson step if likelihood decreases otherwise 0.9-43 - changed default minsteps in profL1. - fixed error when setting col when plotting breslow obect - fixed error in predict in cox.R in the presence of an offset term 0.9-42 - Fixed error with NA loglikelihood in the Cox model introduced in version 41. - Fixed error "Error in strata[status == 1] : object of type 'closure' is not subsettable" (thanks to Ilari Scheinin for the bug report) 0.9-41 - Fixed wrong calculation of cvl at infinite penalty in optL2 for Cox. - fixed error message in case of ?Inf loglikelihood in the cox model. 0.9-40 - added functionality for fused lasso 0.9-39 - fixed Inf * 0 problem in cox.R (thanks to Levi Waldron for the bug report) - fixed problem in case unpenalized = ~strata() without further covariates (thanks to Holger Schwender for the bug report) - added strata example to vignette 0.9-38 - fixed a problem in .coxmerge arising when censored observations were present at time 0 (thanks to Thierry Picot for the bug report) - fixed cross-validated partial likelihood for k