Example 67

Dear Dr. julian
The results of example67 show that the thermal conductivity’s tensor is not diagonal and the off diagonal arrays are not zero although the thermal conductivity is expected to be diagonal for the cubic structures. I tested it for another cubic structure and again obtain the off diagonal one. Please help me to know why the results are like this.
The same calculation with another code such as ShengBTE lead to the diagonal tensor as expected!
Sincerely,
Samira

Dear Samira,
Hard for me to say for certain, since this depends on Alamode as much as GULP, but I can give you an opinion. Whenever finite differences are involved there will be noise in the calculation that depends on the order of the finite differences, step size, etc. Therefore it’s not uncommon to get answers that might deviate from the exact answer because of this. Some codes might enforce symmetry to reduce noise (i.e. if it uses the fact it’s cubic then it might just zero the values or ignore them). You’d have to look into ShengBTE to see if it does anything like this, or whether it’s a difference in the finite differencing strategy.
Regards
Julian

Dear Dr. Julian
Thank you for your prompt reply.
The Bests,
Samira

Dear Dr. Julian
In the manual it is said that ‘space’ option is optional. But when I run example 67 with ‘space’ option and without it obtaining very different results! Please help me to know why the results are like this.
Sincerely,
Samira

What would be the point of an option that didn’t make a difference?!
I’ll leave it to your supervisor to explain what a space group is and why it including it or not would make a difference when you only input the asymmetric unit since that’s fundamental and not related to GULP.
PS The answer as to why things are different should be obvious by comparing the output from the 2 runs.