MAVRIC - Poor man's parallel and convergence test

[Raymond Wang]

Hi,

I am using MAVRIC/SCALE 6.2.4 to do some calculations. I planned to use poor man's 

parallel (i.e. same input file with different random number seeds) for my calculations. I

 think it might be possible to post-process and perform the statistical checks for regional tallies using the information available in rt*.txt files from different runs. 

I think the statistical checks for

1. mean value,

2. standard deviation,

3. relative uncertainty, and

6. FOM

are relatively easy to implement. However, I don't know how to combine the relative VOV
from different runs. I checked the equation and the output, and it seems there are no sufficient information to do that. Do you have any good idea about this?

By the way, I don't think there is a way to post-process existing data to have statistical checks on the combined mesh tallies from MAVRIC...The histories at each batch are not available in the output files. Please correct me if my understanding is wrong.

Thank you.

Regards,

Ray

[Cihangir]

Ray,

All the equations for combining region tallies and mesh tallies are meant to get you better statistics in the mean values and associated uncertainties, so statistical checks on other metrics do not apply.

Thanks,

Cihangir

[Raymond Wang]

Hi Cihangir,

Thank you very much for your quick reply. There are standard deviations in each batch in file rt*.txt.

Although apparently the standard deviation is averaged with the ones from previous runs, I think it 

is possible to estimate what is the standard deviation at specific batch I using the information from
batch I and batch I-1.  We can do the same thing on the mean value. Then what we might be able to 

do is to combine the runs with different random number seeds together.

For example, run A and run B has 300 batches. Through some post processing, we can append run B to run

A (batch 301 ~ 600) and then perform the statistic check accordingly. It's a little bit complicated but I think 

it's doable.

Please correct me if I am wrong. Thank you.

Ray

2023年11月2日木曜日 15:05:05 UTC-4 Cihangir:

[Cihangir]

Yes, you can combine batches by appending statistics from batch 301 to 600 on top of the previous one; effectively yielding a single continuous run with a batch of 600.