19990823: who is who: road map of the calculations between 19990614 and 19990819

GOAL

Road map of the last calculations, for not getting lost in the forest.

Results

19990614 19990617 19990702 19990707 19990712
$Zr_{54}Ce^{IV}_{38}Ce^{III}_{16}O_{208}(V_O^{\cdot\cdot})_8$system.

Here I was trying to get reproducible  MSD plots for oxygen bystandardizing the starting random configuration. Wrote thedlsub code for this purpose. At present, it seems that thereproducibility problems were due to the small number of vacancies,which determined  an insufficient statistics.

19990712 is a summary for 19990614 19990617 19990702 19990707

19990727 19990729 19990730 19990731 19990802 19990805
		
$Zr_{54}Ce^{IV}_{38}Ce^{III}_{16}O_{208}(V_O^{\cdot\cdot})_8$system.


Here I was thinking that the apparently irreproducibleoxygen diffusion coefficient values could be due to an equilibrationtime too short: so I repeated the previous calculations keeping thestarting configuration of each unchanged, but setting an equilibrationtime ten times longer. The following is the correspondence map:







19990730 was fully comparable with the other long equilibrated runs, buthad a yet different starting configuration, with no counterpart in theshort equilibrated calculations.


19990805 summarizes results for the whole series: 19990614 19990617 1999070219990707 19990727 19990729 19990730 19990731 19990802

19990602 19990709 19990728
		
These calculations compare the same input on two different hardwares:utsj90 and t3e:

 


utsj90
t3e

19990602
19981127

19990707
19990709

19990728
19990727




The comparison 19990602/19981127 seemed to indicate a big difference, butthen I could find the reason, and the following two pairs ofcalculations show good agreement between the two machines.


19990809		
Here I realized that the unwanted dependence of the oxygen diffusioncoefficient upon the starting configuration could be due to the smallnumber of vacancies, which in turn determined an insufficientstatistics.


So this was an attempt to run a calculation with 100 oxygen vacanciesand the same composition as before:$Zr_{686}Ce^{4+}_{482}Ce^{3+}_{204}O_{2642}(V_O^{\cdot\cdot})_{102}$. Itturned out that the computational demand was too high.


19990810 19990812		
Here I simulated the system$Zr_{250}Ce^{4+}_{50}Ce^{3+}_{200}O_{900}(V^{\cdot \cdot }_O)_{100}$,which is reasonably tractable (but now composition is different, witha very high reduction degree). I got good results, but then I realizedthat the cell parameter was inappropriate for the new composition.

19990814 19990815 19990817
		
These are$Zr_{250}Ce^{4+}_{50}Ce^{3+}_{200}O_{900}(V^{\cdot \cdot }_O)_{100}$calculations with the correct cell parameter. The startingconfigurations are the two previously run plus a third one:

 


cell parameter

wrong
right

19990810
19990814

19990812
19990815

 
19990817




19990817 summarizes the results, which look nice.

(19990815) 19990818 19990819
		
Here I took the starting configuration of the best of the threeprevious calculations (19990815) and, keeping that fixed, I changed thetemperature to get an activation energy.

 


calc
$T\;K$

19990818
1273

19990819
1473

19990815
1759




Results look nice. Summary for these three calcs is in 19990819.