[SciPy-User] Speed-up simple function
Charles R Harris
charlesr.harris at gmail.com
Mon Jan 10 10:21:00 EST 2011
On Mon, Jan 10, 2011 at 7:45 AM, <g.plantageneto at runbox.com> wrote:
>
> Thanks for the quick answers. To be more specific, the slow part is
> something like this (see comments in the code):
>
> for tstep in xrange(0,end): # I need this cycle otherwise I run out of
> memory
> VAR1 = file.variables['VAR1'][tstep,:,:,:] #Read from a NetCDF
> file
> VAR2 = file.variables['VAR2'][tstep,:,:,:] #Read another
> variable from the same NetCDF file
> # Use the data read above in two functions. I do also some reshaping
> to have the correct
> # input for the function. The polynomials computed in the functions
> are of order 10 at most.
> COMP1 =
> FUNC1(VAR1,VAR2,np.tile(5000,VAR1.shape),np.reshape(np.repeat(SOMETHING,nrep),VAR1.shape))
> COMP2 = FUNC2(VAR1,COMP1,np.tile(5000,VAR1.shape))
> #Compute an average on the output of previous computations
> RESULT[tstep] = np.average(COMP2,weights=W)
>
>
This looks terribly over complicated. What are you trying to do and where is
the polynomial?
<snip>
Chuck
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20110110/b904210c/attachment.html>
More information about the SciPy-User
mailing list