numerical oscillations

[mulvrova]

Hello,

I am experiencing weird behavior of Elmer. When I run a simulation long enough there are numerical oscillations in the output, more precisely looking at mean temperature in the computational box. After the restart is done, oscillations are gone, sometimes appearing again lately. Any clue why that happens? Although oscillations are really small it disturbs me. There is a picture enclosed that illustrates that.

By the way why the restart doesn't continue exactly the same way as before the restart? Could it have something to do with single/double precision?

Regards, Martina.