eigen
Peter Dalgaard BSA
p.dalgaard@biostat.ku.dk
10 Apr 1999 12:31:30 +0200
Douglas Bates <bates@stat.wisc.edu> writes:
> >>>>> "DMB" == Douglas Bates <bates@stat.wisc.edu> writes:
>
> DMB> I think the compiler that I used for the 0.63.4 builds was
> DMB> egcs-1.0.3 but I'm not sure. I'll go back and re-build 0.64.0
> DMB> with egcs-1.0.3 instead of egcs-1.1.1 and see what happens.
>
> Changing to egcs-1.0.3 for a re-build of 0.64.0 didn't clear up the
> problem.
I also get the same result with different compilers (gcc 2.7.2.3 and
egcs-2.91.57) Note BTW that it *fails* the
consistency test from the example file
> m %*% V - V %*% diag(lam)
[,1] [,2] [,3] [,4]
[1,] 2.775558e-17 1.387779e-17 2.775558e-17 -3.382711e-17
[2,] 4.117019e-01 2.872937e-01 -2.687366e-01 7.624831e-02
[3,] 4.615645e-02 3.220889e-02 -3.012842e-02 8.548301e-03
[4,] -2.688891e-01 -1.876361e-01 1.755162e-01 -4.979899e-02
- whereas the older version did
> m %*% V - V %*% diag(lam)
[,1] [,2] [,3] [,4]
[1,] -8.326673e-17 0.000000e+00 0.000000e+00 -2.622414e-18
[2,] -5.551115e-17 1.110223e-16 0.000000e+00 -1.368263e-16
[3,] 3.469447e-18 -2.775558e-17 -5.551115e-17 3.957338e-17
[4,] -5.551115e-17 -1.110223e-16 5.028530e-16 1.502704e-16
[0.63.4 in progress (March 9, 1999) on my office machine]
The test on symmetric matrices does not fail, though. (Make check
would have caught this, save for the m<-m+t(m) line...)
Most likely, we're being bitten either by a code 'cleanup' or some
sort of change in the linked libraries.
--
O__ ---- Peter Dalgaard Blegdamsvej 3
c/ /'_ --- Dept. of Biostatistics 2200 Cph. N
(*) \(*) -- University of Copenhagen Denmark Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard@biostat.ku.dk) FAX: (+45) 35327907
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !) To: r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._