Return to 2016-05-29-b-in.txt CVS log

Up to [local] / OpenXM / src / hgm / mh / src / Testdata

Added a test data to study numerical instatiblity.

%!version2.0
%Mg=
1
%Beta=
0.03
%X0g=
0.3105682
%Hg=
0.001
%Dp=
1
%Xng=
200.0
## (m+1)/2=1.0, (n1+n2)/2=(5+10)/2=7.5
## (n1+m+1)/2=(5+1+1)/2=3.5
%p_pFq=2, 1.0, 7.5
%q_pFq=1, 3.5
%ef_type=2

## a<-hgm.p2wishart(m=1,beta=c(0.03),q=0.3,q0=0.1,err=c(1e-15,1e-10))
## [1]  0.3105682  0.9997290 54.0082200
## 
##
##  Output of this input: Iv[0] = 9293.58
##  Ef=9.88128e-5 ,  Iv[0]*Ef = 0.9183247  with approxdeg=86
## Error rate:  only one digit is valid.
## > 9293.58*9.88128e-5
## [1] 0.9183247
## > 9293.58*9.88128e-5/a[2]
## [1] 0.9185736
##  ## Iv[0] herpaps should be 10117.4
##
## Possible reason. x/(0.03+x) is close to 1 when x is large relative to 0.03.