I am having some issues with running the same code in my laptop which has Kernel Rev 15.0.7 and in my PC which has 15.1
somehow I am getting different results from the same code and same data
any idea on this?
If your data is very badly scaled you can lose lots of precision calculating the hessian. If it is bad enough, you can lose all 15.5 places of precision and then your calculations will be based on the round-off error of the computer, which can certainly be different on a different machine. Usually this round-off difference is irrelevant, but if you have catastrophic cancellation, then there is nothing left except the round-off error. Or model misspecification problems can sometimes also cause similar sorts of issues.