[Vinecopula-commits] r40 - in pkg: R man

[noreply at r-forge.r-project.org]

Author: ulf
Date: 2013-11-14 15:57:36 +0100 (Thu, 14 Nov 2013)
New Revision: 40

Modified:
   pkg/R/RVineMLE.R
   pkg/man/VineCopula-package.Rd
Log:
Typisch: Grad abgeschickt und schon wieder einen Fehler gefunden. Diesmal in der MLE. Wenn man z.B. nur Gauss copulas hat, dann ist er bei der parscale trotzdem in den Fall reingelaufen, dass er zwei-parametrische copulas hat.

Modified: pkg/R/RVineMLE.R
===================================================================
--- pkg/R/RVineMLE.R	2013-11-11 11:40:21 UTC (rev 39)
+++ pkg/R/RVineMLE.R	2013-11-14 14:57:36 UTC (rev 40)
@@ -302,11 +302,14 @@
 		pscale[i] = ifelse(Copula.Types[i] %in% c(1,2,43,44), 0.01, 1)
 	}
   pscale2=numeric()
-  for(i in 1:nParams2)
+  if(nParams2>0)
   {
-    pscale2[i] = ifelse(Copula.Types[i] %in% c(104,114,124,134,204,214,224,234), 0.05, 1)
-  }
-	pscale=c(pscale,pscale2)
+	  for(i in 1:nParams2)
+	  {
+		pscale2[i] = ifelse(Copula.Types[i] %in% c(104,114,124,134,204,214,224,234), 0.05, 1)
+	  }
+		pscale=c(pscale,pscale2)
+	}
 
 	if(!exists("factr"))		# Toleranz etwas hoch setzen (groeber)
 		factr=1e8

Modified: pkg/man/VineCopula-package.Rd
===================================================================
--- pkg/man/VineCopula-package.Rd	2013-11-11 11:40:21 UTC (rev 39)
+++ pkg/man/VineCopula-package.Rd	2013-11-14 14:57:36 UTC (rev 40)
@@ -81,7 +81,7 @@
 Package: \tab VineCopula\cr
 Type: \tab Package\cr
 Version: \tab 1.2\cr
-Date: \tab 2013-10-11\cr
+Date: \tab 2013-11-11\cr
 License: \tab GPL (>=2)\cr
 Depends: \tab R (\eqn{\geq 2.11.0}{>= 2.11.0}), MASS, mvtnorm, igraph \cr
 Suggests: \tab CDVine, TSP, ADGofTest \cr