James Boyle wrote:
In the quadmesh demo the shapes of the data(Z) and the coordinates(Qx,Qz) are exactly the same (n x n).
Now my understanding is that what is being plotted are values in a mesh which needs (n+1,n+1) vertices to describe it.
If I have a mesh of quadrilaterals defined by the vertices - how can pcolormesh plot this?
In collections.py the comments of QuadMesh appear to be addressing the description of the vertices, how is this used by pcolormesh?
It is very likely something very simple I am missing but I am at a loss at present.
I think that all the pcolor-type functions and methods simply throw away the last row and column of Z. This behavior is inherited from Matlab. Perhaps it could be clarified (made explicit) in the documentation. I would also be inclined to make these functions accept a Z with dimensions (m-1, n-1) for a grid with dimensions (m,n). It is silly to have to tack on an extra dummy row and column just to be able to use all the values in a Z array.
Other people are working on the pcolor-type functions now, so I don't want to barge in and make these changes, but I don't see any reason they can't be made once the dust settles.