[dune-pdelab] Accumulated view wieghts on numerical jacobian

[Ospina De Los Rios, Santiago]

Hi everybody,


I was taking a look at local matrix/vector code to see if I could do some refactor and include an matrix sparse version. Doing so, I came across the accumulated views on the numerical jacobians. Do you know why some of these operators apply the weights on the matrix while others do so on the vector? I don't think that the order of weight application makes any difference there, but I might be wrong on that. It seems to me that is about helping the compiler to "optimize" the application of the weight over the loop, but I am unsure.


Best,

Santiago
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.dune-project.org/pipermail/dune-pdelab/attachments/20200520/ad76a8b7/attachment.htm>