[Robast-commits] r308 - pkg/ROptEst/inst/scripts

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

Author: stamats
Date: 2009-06-30 06:32:35 +0200 (Tue, 30 Jun 2009)
New Revision: 308

Modified:
   pkg/ROptEst/inst/scripts/GammaModel.R
Log:
added comment due to new E method in distrEx which increases precision of certain computations.

Modified: pkg/ROptEst/inst/scripts/GammaModel.R
===================================================================
--- pkg/ROptEst/inst/scripts/GammaModel.R	2009-06-17 02:43:56 UTC (rev 307)
+++ pkg/ROptEst/inst/scripts/GammaModel.R	2009-06-30 04:32:35 UTC (rev 308)
@@ -4,6 +4,7 @@
 require(ROptEst)
 options("newDevice"=TRUE)
 
+
 ## generates Gamma Family with 
 ## scale = 2 and shape = 0.1
 G <- GammaFamily(scale = 2, shape = 0.1)
@@ -12,11 +13,16 @@
 distrExOptions(ErelativeTolerance = 1e-8) # increase precision for E
 checkL2deriv(G)
 
+## 30.06.09: new method for "E" with 
+## signature(object = "Gammad", fun = "function", cond = "missing")
+## in package distrEx introduced which slightly reduces the problem
+## documented below.
+
 ## more precisely:
 ## numerical integration gives
 E(Gammad(scale = 2, shape = 0.1), function(x) (log(x/2)-digamma(0.1))^2)
 
-## vhereas
+## whereas
 trigamma(0.1)
 
 ## Problem is more or less caused by integration of log(x/2)^2