[Tutor] location of points in a covariance matrix

[Beanan O Loughlin]

Hi all, I'm a meteorology postgrad working with python for the first time.

I have found the location minimum in a large covariance matrix. this value
corresponds to the covariance of two points on a latitude, longitude grid.

I need to find a method to locate these two points on the lat,lon grid.

this is the code i have used, where 'se' is a 3-D array of data


>>> nt,nlat,nlon = shape(se)
>>>nt,nlat,nlon                            # 3-D array of data taken 1464
times, over 41 latitudes and 58 longitudes
(1464, 41, 58)
>>>
>>>
>>>m=reshape(se,(nt,nlat*nlon))    # reshape to (time,latitude longitude
data point) where 2378 = 41*58
>>>
>>>shape(m)
(1464,2378)
>>>
>>>
>>>covmat=cov(m)                           # calculate covariance matrix
>>>
>>>shape(covmat)
(2378,2378)

>>>def min(R):
       U = triu(R)                              #just use one half of the
diagonal matrix
       n = U.shape[0]
       U.flat[::n+1] = 1000000000.0    #give the diagonal elements a large
value so they wont be selected
       k = argmin(U.flat)                    #find the min value of the
flattened array
       i, j = divmod(k,n)                     #calculate the index of the
minimum data
       return i, j, R[i,j]

>>>
>>> min(covmat)
(7, 1914, -2.3016361721151051)

so the minimum is found at (7,1914) in the covariance matrix and has a value
of -2.3

This min point corresponds to the covariance between two 'lat,lon' data
points in my (41,58) sample grid.

Is there a way i can move back from my (2378,2378) covariance matrix to see
where these two points are located on the (41, 58) grid?

Thank you very much in advance

B.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.python.org/pipermail/tutor/attachments/20070724/1884bb5c/attachment.htm