Cross section generation is triggered for XS IDs that are not even in the core

[keckler]

I have a case that runs many cycles and has a lot of assembly definitions. Many of these assembly definitions are not present in the initial core. Some of those have their own unique XS ID.

I've noticed that when cross sections are generated for the initial core, they are generated for these XS IDs that are not represented in the core, even though it is known that they will not be used. This is a waste of computational resources.

For cases when LatticePhysicsFrequency is BOC or higher, we do not need to be generating all of these cross sections so frequently. Instead, we should look to see what is in the core and only work on those.

Note, before #1288 , this would not have been possible because the shuffling took place after the XSGM was run. But at this point, I'm pretty sure the information is there for us to be smarter about this.

[john-science]

Okay, Chris and I talked about this and I think this is a fair point.

Somewhere in the XSGM:

armi/armi/physics/neutronics/crossSectionGroupManager.py

Line 906 in 04c0670

class CrossSectionGroupManager(interfaces.Interface):

we create representative blocks for all the assemblies that exist, then calculate cross sections for them. So far it appears we only NEED to calculate XS for assemblies that are in the Core.

On first glance, this seems like an important enough performance boost to make it worth doing.