Menu

#133 Faithful failure and probable bad path join

rpy
open
nobody
Win32 (20)
5
2012-07-18
2010-09-06
docduke
No

I'm a very new user of Rpy, so this may be my problem, but I don't think so. I have installed Python 2.7, numpy-1.5.0-win32-superpack-python2.7, R-2.9.1-win32 and rpy-1.0.3-R-2.9.1-win32-py2.7. When I run the faithful.py demo, I get the nonfatal error
Warning message:
In function (x, y, ..., alternative = c("two.sided", "less", "greater"), :
cannot compute correct p-values with ties
I found that after attempting to debug a very simple bivariate function, "r('fm2<-lm(ibm~dj+ti)')". That function appears to run properly, but I am unable to extract the results. Specifically, the statement "r('print(summary(fm2))')" produces:
Warning message:
In chol2inv(Qr$qr[p1, p1, drop = FALSE]) :
unable to load shared library 'C:\c\R\R-2.9.1/modules//lapack.dll':
LoadLibrary failure: The specified module could not be found.
Traceback (most recent call last):
File "C:/a/rpy/ibm1.py", line 14, in <module>
r('print(summary(fm2))')
File "C:\c\Python27\lib\site-packages\rpy.py", line 322, in call
return self.eval(self.parse(text=s))
RPy_RException: Error in chol2inv(Qr$qr[p1, p1, drop = FALSE]) :
lapack routines cannot be loaded
Note that the path in the warning message: .../modules//lapack.dll contains a double slash. The lapack dll is present at .../modules/lapack.dll
I haven't a clue how to debug this. Neither grep nor the pythonwin stack viewer sees any symbol like "chol2inv" or "Qr$qr". I can run the identical program (not imbedded in Python) in R on the same computer and it runs without error.

My objective is simple. I want to be able to provide 3 variables, and get the regression and the standard error estimates. I can transfer the results of the "lm" call to Python. The coefficients are easy to find, but it does not appear to contain the standard error data. I can calculate it from the residuals, but it seems to me that this defeats the purpose of using Rpy!

I also considered moving to Rpy2, but the download page has this prominent warning that R must be compiled with "--enable-R-shlib", and there is no indication whether that is the case for the windows binary distributions of R.

Help, please!

Discussion

  • docduke

    docduke - 2010-09-06

    Rpy program that produces the fatal lapack error.

     
  • lgautier

    lgautier - 2010-09-23

    notes:

    1/ Compiling with "--enable-R-shlib" is also required for RPy to work

    2/ Rpy2 support on Windows is currently interrupted (last working version is 2.0.x)

     

Anonymous
Anonymous

Add attachments
Cancel