Out-of-scope k variable usage when updating source terms.

[willGraham01]

The variable k is used in the Ksource update loop, with a specific instance here, despite not being a loop variable.

This is only possible since the indexing-variables are declared right at the top of execute(). In the previous two loops (Isource, Jsource), k is a loop variable so this is fine. However in the Ksource update loop only i and j are loop variables. It is only because k is globally scoped that the code compiles: but then we also notice that in this loop the value k is not changing, but is fixed to k = inputs.K1.index, and is still being used to fetch constants from the input matrices!

[willGraham01]

Having delved into the code some more, I believe that what's meant to be happening in these loops is that:

• When updating across the K0 plane, we should be using inputs.K0.index (possibly offset by -1) in place of k,
• When updating across the K1 plane, we should be using inputs.K1.index (which we are currently doing "implicitly").

However, we should clarify this and make these associations explicit if we are! We may additionally need to regenerate the test data if we are meant to be using inputs.K0.index in the K0 loop (as presently we are still using k).

[prmunro]

This is a good spot and it is indeed an error in the code (which I know you already know!). This part of the code hasn't been used for some time and I need to go back into the theory to test this. In particular, this part of the code will only be encountered when dispersive materials are modelled. As mentioned, I haven't used this feature for >15 years. Can we create an issue for me to generate a test that will use the dispersive materials feature, and also to double check what this k-value should be, as I am 99% confident that Will has the correct solution?

[prmunro]

Having delved into the code some more, I believe that what's meant to be happening in these loops is that:

• When updating across the K0 plane, we should be using inputs.K0.index (possibly offset by -1) in place of k,
• When updating across the K1 plane, we should be using inputs.K1.index (which we are currently doing "implicitly").

However, we should clarify this and make these associations explicit if we are! We may additionally need to regenerate the test data if we are meant to be using inputs.K0.index in the K0 loop (as presently we are still using k).

I agree with this @willGraham01

[prmunro]

As discussed, @willGraham01 , please make change as described above, I will review later.