[Rd] lapack test failed (PR#7294)

From: <root_at_u-003-scfe01.uni-tuebingen.de>
Date: Tue 19 Oct 2004 - 20:12:47 EST

Hallo,
I've just installed R on our Opteron System (AMD Opteron 246). Running make check-all results in a failure during the lapack test:

> ## ------- tests of random real and complex matrices ------
>
> ## 100 may cause failures here.
> eigenok <- function(A, E, Eps=1000*.Machine$double.eps)

+ {
+     V <- E$vect; lam <- E$values
+     stopifnot(abs(A %*% V - V %*% diag(lam)) < Eps,
+               abs(lam[length(lam)]/lam[1]) < Eps || # this one not for singular A :
+               abs(A - V %*% diag(lam) %*% t(V)) < Eps)
+ }

>
> Ceigenok <- function(A, E, Eps=1000*.Machine$double.eps)
+ {
+     V <- E$vect; lam <- E$values
+     stopifnot(Mod(A %*% V - V %*% diag(lam)) < Eps,
+               Mod(A - V %*% diag(lam) %*% Conj(t(V))) < Eps)
+ }

>
> ## failed for some 64bit-Lapack-gcc combinations:
> sm <- cbind(1, 3:1, 1:3)
> eigenok(sm, eigen(sm))

Error: abs(A %*% V - V %*% diag(lam)) < Eps is not TRUE Execution halted

So this seems to be one of the combination of 64bit and gcc which leads to failures. Using the configure option:
--with-lapack="-lacml"
results in the same error. I am not sure wether the build uses the acml (Version 1.5) or not. Using pgf77 instead of g77 results in problems building R. (Missing function definitions or invalid compiler options (fpic))
If there is a way to overcome this problem please let me know. Thanks in advance
w.dilling


W.Dilling                               Tel.: (49) 7071/29-70206
Universitaet Tuebingen                  Fax.: (49) 7071/29-5912
Zentrum fuer Datenverarbeitung          mail: dilling@zdv.uni-tuebingen.de
Waechterstrasse 76
72074 Tuebingen

--please do not edit the information below--

Version:
 platform = x86_64-unknown-linux-gnu
 arch = x86_64
 os = linux-gnu
 system = x86_64, linux-gnu
 status = Patched
 major = 2
 minor = 0.0
 year = 2004
 month = 10
 day = 18
 language = R

Search Path:
 .GlobalEnv, package:methods, package:stats, package:graphics, package:grDevices, package:utils, package:datasets, Autoloads, package:base



R-devel@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel Received on Tue Oct 19 20:44:02 2004

This archive was generated by hypermail 2.1.8 : Fri 18 Mar 2005 - 09:00:37 EST