Computing norm of solution obtained by LUSolver

[conpierce8]

@atalasik

However, I observed that in Fenics 2019.1.0, the functionality of assign is the other way around.

You are correct. My mistake.

@atalasik

I would also like to ask that if the calculateNorm function, receives a vector having various fields (for example, temperature, pressure, entropy and velocity), then why does the split function needs to contain a boolean argument True? For example:

v,p,T,H = u.split(True)

I am asking this because, without True as an argument the norm doesn’t get calculated properly and yields fluctuating values. I had not used the True argument in the case without the LUSolver, and norm was observed to steadily decrease with each iteration.

Can you provide a code example that demonstrates this behavior? The code you have provided in this thread does not use split, and the code you have provided in your previous post does not contain an implementation of calculateNorm. Could you also provide your implementation of the “case without the LUSolver”? It’s difficult for me to determine why you are not getting the expected results without having a code that can produce the expected results.