Closed ClarkMcGrew closed 4 months ago
I've checked the LLH scans for xsec and ND splinified systematics comparing with the previous Lena's and MaCh3 and they all agree. I don't have the LLH scans for the flux and the detector covariance matrix to compare against at the moment but they rarely give issues. A visual scan of the flux LLH scans looks good. I propose to merge now.
I'd say "go for it". I think the others are on vacation, it's barely a one liner, and the only functional change is to remove the std::isinf check.
The LLH calculation is throwing a runtime_error if the likelihood returned infinity (a valid likelihood). This is changed so that it will only throw if the likelihood returns NAN. That's probably a sign that something didn't get initialized and is still trapped as an error.