Make A 2d Matrix Of Strings To Match Mesh Coordinates

I have a structured mesh. The mesh has coordinates (X,Y), but each mesh cell also has an alphanumerical code assigned. The mesh X and Y coordinates are generated as 2D arrays with

Solution 1:

You can use numpy.array_split or even numpy.split to split the arrays into multiple sub-arrays. But the former does not raise an exception if an equal division cannot be made.

In [2]: np.array(np.array_split(df['Code'].values, 4))
Out[2]: 
array([['aa1', 'aa2', 'aa3'],
       ['aa4', 'bb1', 'bb2'],
       ['bb3', 'bb4', 'ab1'],
       ['ab2', 'ab3', 'ab4']], dtype=object)

EDIT :

You mean like this?

In [5]: np.array(np.array_split(df.as_matrix(columns=['Code']), 4))
Out[5]: 
array([[['aa1'],
        ['aa2'],
        ['aa3']],

       [['aa4'],
        ['bb1'],
        ['bb2']],

       [['bb3'],
        ['bb4'],
        ['ab1']],

       [['ab2'],
        ['ab3'],
        ['ab4']]], dtype=object)

Solution 2:

Well, if you know the length of each row you can use just a list. You can make the 2D array in "your head" and convert the coordinates into their position in an array.

Example:

You have got a row_length x column_length grid, therefore you have a list with a row_length*column_length entries. To access a specific coordinate you access the following entry of the list:

Pos(x|y) = mygrid[xcoord*rowlength+columnlength]

Explanation/Clarification:

Instead of using an actual 2D array, you can just use a 1D list. If you know the size of each row, that isn't a problem but instead, increases the speed your program is working with. Let's assume we have got a 3x3 grid containing a letter at each position, our coordinates will look like that: (0|0, 0|1, 0|2, 1|0, 1|1, ...)

We could represent this grid with:

      0          1         2

0    'a''q''x'

1    'm''f''b'  

2    'l''s''r'

Or, instead of creating an actual 2D grid, we just create a 1D array.

data = ['a', 'q', 'x', 'm', 'f', 'b', 'l', 's', 'r']

To get the index of a specific coordinate we can now multiply the row_number with the length of each row and add the column_number to it.

For example, to access the coordinate (2|1) of the grid above, we can just access:

data[2*3+1]

If you check both values, you will see that both are delivering the letter 's' as it should.