I am the author of the nleqslv package for solving systems of non linear 
equations.
The main code is written in Fortran.
At the beginning of March, I released a new version with a few minor 
corrections in the R code.
Nothing in the C or Fortran code was changed.

The package passed all the CRAN checks/tests on Linux, Windows and Solaris 
systems and Mac OS X Intel platforms.
However, on the Mac OS X PowerPc platform one of the examples gave a bus error. 
With previous versions of the package this did not happen. 
Moreover, the relevant Fortran code had not changed. 
So I simply do not understand what is going on.
I run Mac OS X Snow Leopard 10.6.7 which is Intel only, so I have no means of 
trying to find out what may be the cause.

I would like to ask anyone with a Mac running Mac OS Leopard PowerPc to run the 
example below to see what happens.
And maybe the cause of the bus error can be found.

I have attached the R script and the output of running 
Rscript -e ppcissue_mailexample.R > ppcissue_mailexample.rout.txt

best regards

Berend Hasselman

Attachment: ppcissue_mailexample.R
Description: Binary data

R version 2.12.2 (2011-02-25)
Platform: i386-apple-darwin9.8.0/i386 (32-bit)

locale:
[1] en_GB/en_GB/C/C/en_GB/en_GB

attached base packages:
[1] stats     graphics  grDevices utils     datasets  base     
[1] 2.0 0.5
[1] 2.2500000 0.8432818
Chkjac  possible error in jacobian(   1,   1)
        0.8000000000000D+01 Estimated =  0.4000000029802D+01
Chkjac  possible error in jacobian(   2,   2)
        0.1250000000000D+01 Estimated =  0.7500000298023D+00
$x
[1] 2.0 0.5

$fvec
[1] 2.2500000 0.8432818

$termcd
[1] -10

$message
[1] "Analytical Jacobian most likely incorrect"

$scalex
[1] 1 1

$nfcnt
[1] 0

$njcnt
[1] 0

_______________________________________________
R-SIG-Mac mailing list
[email protected]
https://stat.ethz.ch/mailman/listinfo/r-sig-mac

Reply via email to