[Pw_forum] Question about tests

[Andreas Kissavos]

Hi,

after compiling, I ran most of the test included. example01 produced  
identical results, while in example02, the Si Gamma point  
calculation, everything is identical up to the inversion of the  
dynamical matrix. After the inversion, my first 3 frequencies deviate  
a bit from the reference ones:

My are 0.101977 [THz] while the reference is 0.093301 [THz]. Note  
that everything is identical (to 6 decimals at least) before this.

The same thing occurs for all other examples where the dynamical  
matrix is inverted, while all other examples produce perfect agreement.

I noticed that the examples were calculated with version 3.2, while  
my calculations were performed with the latest one, 3.2.3. I also  
note that there has been some problems with phonons earlier. My  
question now is:

Has anything been changed with regards to this inversion between 3.2  
and 3.2.3 that can account for the error I am getting?

I certainly hope so, since I am not too fond of the idea of having  
defect  LAPACK-routines.

Best regards,

Andreas Kissavos