19990730: a third run fully comparable with 19990727 and 19990729

GOAL

Run a third calculation homogeneous with 19990727 and 19990729: same FIELD and CONTROL files, different CONFIG file, but no $(Ce_{Zr}^\prime{}Ce_{Zr}^\prime)^{\prime\prime}$, $(Ce_{Zr}^\prime{}V_O^{\cdot\cdot})^\cdot$ or $(V_O^{\cdot\cdot}V_O^{\cdot\cdot})^{\cdot\cdot\cdot\cdot}$ tight pairs.

To check oxygen MSD reproducibility out.

Runtime problems on t3e

Just a note. I got exactly the same error as 19990727. I re-run EXACTLY the same input (literally: not even a bit of the input was changed) and the run completed successfully. This suggests that the runtime errors could be due to deadlock situations which happen depending upon the particular state of the PE's. According to Sigismondo Boschi from CINECA these are things to try out:

• compile without optimization: could be a compiler bug
• use MPI_Bsend in place of MPI_Send, which should be safer from the deadlock problem point of view

Results

The MSD plot has a slope different from those in 19990727 (23) and 19990729 (25).

The following is the comparison of the starting configurations for 19990727 (23), 19990729 (25) and 19990730 (present calc).