In very rare cases, setting &ELECTRONS startingwfc = 'random' can lead to a 
different converged value. Have you checked that the inputs are identical?
Also, did you set conv_thr tight enough? The default value is not good enough. 
And in my case, I only trust the total energy down to the sixth decimal place. 
The unit is eV.

Best regards,
GiMyung Park
Yonsei University
-----Original Message-----
From: "Simon Michnowicz via users"<[email protected]>
To: <[email protected]>;
Cc:
Sent: 2026-08-28 (금) 10:52:09 (GMT+09:00)
Subject: [QE-users] Result precision across different builds

Dear User List
I am helping a researcher install Quantum Espresso. I find different 
combinations of compilers/libraries produce slightly different results upon a 
sample input given to me. The researcher said the results should be consistent 
up to 5 decimal points.  Is the  variation that I see to be expected, or could 
there some other problem here?

Tool                                       Final Energy
Intel Compiler,MKL               -314.1147158171
gcc/15.2.0-gcc-11.5.0/MKL  -314.1256316418
gcc/11.4/openblas                -314.1256341822

much thanks

---
Simon Michnowicz                
Senior Application Specialist,  High-Performance Computing

Monash eResearch Centre
Monash University
15 Innovation Walk, Building 75, Clayton Campus
Wellington Road, VIC 3800
Australia

T:  +61 3 9902 0794                   
M: +61 3 0418 302 046
E: [email protected]
monash.edu
_______________________________________________________________________________
The Quantum ESPRESSO Foundation stands in solidarity with all civilians 
worldwide who are victims of terrorism, military aggression, and indiscriminate 
warfare.
--------------------------------------------------------------------------------
Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
users mailing list [email protected]
https://lists.quantum-espresso.org/mailman/listinfo/users

Reply via email to