Paul Hansen wrote:
The problem I was thinking about was if the core exceeds the yield strength while the surface is still elastic. I think in that case, the sword would still take a bend.


Yes you are absolutely right. If any part of the blade exceeds the local yield strength it will take a set. The yielded areas will prevent the unyielded areas from returning to their original shape and there will therefore also be some residual stresses.

Paul Hansen wrote:

Also I think calculating sword strength is a bit complicated if we consider pattern welding and welded-on steel edges, as well as hardening thickness.


I agree. The complexity is just too high to get exact numbers and there's not really much point in it anyway. But, if you wanted, you could at least get a good idea of what hardening thickness you would need so that the soft core will not affect performance. I'm guessing you would not need a very tick hardened layer to do all the heavy lifting.