[CP2K-user] [CP2K:20950] Energy from restart file not equal to energy at the end of geo opt

Bram Van der Linden bramvdlinden98 at gmail.com
Thu Dec 5 13:34:04 UTC 2024


I set EPS_DEFAULT to 1e-14 and EPS_SCF to 1e-7.
I included the geo opt result below.
If I do an energy calculation from the restart, I obtain
ENERGY| Total FORCE_EVAL ( QS ) energy [a.u.]:             
 -17.164621690030497
So, again this energy is equal to the energy at the last step, it is not 
equal to the energy computed in the final scf of the geo opt.


 SCF WAVEFUNCTION OPTIMIZATION

  Step     Update method      Time    Convergence         Total energy   
 Change
  
------------------------------------------------------------------------------
     1 Pulay/Diag. 0.50E+00    0.0     0.00028837       -17.1646216823 
-1.72E+01
     2 Pulay/Diag. 0.50E+00    0.1     0.00002417       -17.1647099510 
-8.83E-05
     3 Pulay/Diag. 0.50E+00    0.1     0.00001927       -17.1646218371 
 8.81E-05
     4 Pulay/Diag. 0.50E+00    0.1     0.00000930       -17.1646577256 
-3.59E-05
     5 Pulay/Diag. 0.50E+00    0.2     0.00000078       -17.1646200227 
 3.77E-05
     6 Pulay/Diag. 0.50E+00    0.2     0.00000021       -17.1646214825 
-1.46E-06
     7 Pulay/Diag. 0.50E+00    0.1     0.00000007       -17.1646216684 
-1.86E-07

  *** SCF run converged in     7 steps ***


  Electronic density on regular grids:         -8.0000018479       
-0.0000018479
  Core density on regular grids:                7.9999994863       
-0.0000005137
  Total charge density on r-space grids:       -0.0000023616
  Total charge density g-space grids:          -0.0000023616

  Overlap energy of the core charge distribution:               
0.00000004553664
  Self energy of the core charge distribution:               
 -43.83289054591484
  Core Hamiltonian energy:                                     
12.82173498735973
  Hartree energy:                                             
 17.97397572044687
  Exchange-correlation energy:                                 
-4.12744187580467

  Total energy:                                               
-17.16462166837628

 ENERGY| Total FORCE_EVAL ( QS ) energy [a.u.]:             
 -17.164621690030479

 *******************************************************************************
 ***                 BRENT   - NUMBER OF ENERGY EVALUATIONS :       4       
 ***
 *******************************************************************************

 --------  Informations at step =    10 ------------
  Optimization Method        =                   CG
  Total Energy               =       -17.1646216900
  Real energy change         =        -0.0000000430
  Decrease in energy         =                  YES
  Used time                  =               14.528

  Convergence check :
  Max. step size             =         0.0005485208
  Conv. limit for step size  =         0.0010000000
  Convergence in step size   =                  YES
  RMS step size              =         0.0002589937
  Conv. limit for RMS step   =         0.0010000000
  Convergence in RMS step    =                  YES
  Max. gradient              =         0.0002215760
  Conv. limit for gradients  =         0.0010000000
  Conv. in gradients         =                  YES
  RMS gradient               =         0.0001046210
  Conv. limit for RMS grad.  =         0.0010000000
  Conv. in RMS gradients     =                  YES
 ---------------------------------------------------

 *******************************************************************************
 ***                    GEOMETRY OPTIMIZATION COMPLETED                     
 ***
 *******************************************************************************

                    Reevaluating energy at the minimum

 Number of electrons:                                                       
   8
 Number of occupied orbitals:                                               
   4
 Number of molecular orbitals:                                             
    4

 Number of orbital functions:                                               
  23
 Number of independent orbital functions:                                   
  23

 Extrapolation method: ASPC


 SCF WAVEFUNCTION OPTIMIZATION

  Step     Update method      Time    Convergence         Total energy   
 Change
  
------------------------------------------------------------------------------
     1 Pulay/Diag. 0.50E+00    0.1     0.00010461       -17.1646216890 
-1.72E+01
     2 Pulay/Diag. 0.50E+00    0.1     0.00000907       -17.1646546064 
-3.29E-05
     3 Pulay/Diag. 0.50E+00    0.1     0.00000721       -17.1646212859 
 3.33E-05
     4 Pulay/Diag. 0.50E+00    0.1     0.00000339       -17.1646345270 
-1.32E-05
     5 Pulay/Diag. 0.50E+00    0.1     0.00000029       -17.1646210776 
 1.34E-05
     6 Pulay/Diag. 0.50E+00    0.1     0.00000007       -17.1646216153 
-5.38E-07

  *** SCF run converged in     6 steps ***


  Electronic density on regular grids:         -8.0000018479       
-0.0000018479
  Core density on regular grids:                7.9999994863       
-0.0000005137
  Total charge density on r-space grids:       -0.0000023616
  Total charge density g-space grids:          -0.0000023616

  Overlap energy of the core charge distribution:               
0.00000004553565
  Self energy of the core charge distribution:               
 -43.83289054591484
  Core Hamiltonian energy:                                     
12.82173337728078
  Hartree energy:                                             
 17.97397690071527
  Exchange-correlation energy:                                 
-4.12744139290861

  Total energy:                                               
-17.16462161529174

 ENERGY| Total FORCE_EVAL ( QS ) energy [a.u.]:             
 -17.164621615291743

Op donderdag 5 december 2024 om 14:18:46 UTC+1 schreef Bram Van der Linden:

> Dear Jurg,
>
> Thank you for your answer. Does this mean that the final energy 
> calculation in a geo opt uses slightly different coordinates than the ones 
> reported in the restart file?
> Otherwise I do not understand why an energy calculation from the restart 
> file yields the same energy as the energy at step 11, and not the final 
> energy from the geo opt.
>
> With kind regards,
>
> Bram
>
> Op donderdag 5 december 2024 om 14:11:19 UTC+1 schreef Jürg Hutter:
>
>> Hi 
>>
>> this is due to the very loose accuracy/convergence settings in this 
>> example. 
>> In the final calculation an additional SCF step is perfomed (the first 
>> energy 
>> is the same as at the end of the geometry optimization). 
>> This single SCF step decreases the energy by another 10^-5 au. 
>> Tighter settings (e.g. EPS_SCF 10^-7) will decrease this difference. 
>>
>> regards 
>> JH 
>>
>> ________________________________________ 
>> From: cp... at googlegroups.com <cp... at googlegroups.com> on behalf of Bram 
>> Van der Linden <bramvdl... at gmail.com> 
>> Sent: Thursday, December 5, 2024 12:43 AM 
>> To: cp2k 
>> Subject: [CP2K:20945] Energy from restart file not equal to energy at the 
>> end of geo opt 
>>
>> Dear all, 
>>
>> I was following this tutorial: Geometry Optimisation — CP2K documentation<
>> https://manual.cp2k.org/trunk/methods/optimization/geometry.html> 
>> If I perform the geo opt, I get the snippet below at the end of the 
>> output file. 
>> If I do an energy calculation from the restart file, I obtain that 
>> ENERGY| Total FORCE_EVAL ( QS ) energy [a.u.]: -17.164620477328342a 
>> So, the energy of the energy calculation is equal to the energy at the 
>> last step (11), it is not equal to the energy that is recomputed after the 
>> last geometry optimization step. 
>> Could someone explain why this is the case? 
>> Many thanks in advance! 
>> Bram 
>>
>> -------- Informations at step = 11 ------------ 
>> Optimization Method = SD 
>> Total Energy = -17.1646204773 
>> Real energy change = -0.0000000529 
>> Decrease in energy = YES 
>> Used time = 2.886 
>>
>> Convergence check : 
>> Max. step size = 0.0003393777 
>> Conv. limit for step size = 0.0010000000 
>> Convergence in step size = YES 
>> RMS step size = 0.0001493511 
>> Conv. limit for RMS step = 0.0010000000 
>> Convergence in RMS step = YES 
>> Max. gradient = 0.0001787811 
>> Conv. limit for gradients = 0.0010000000 
>> Conv. in gradients = YES 
>> RMS gradient = 0.0000786768 
>> Conv. limit for RMS grad. = 0.0010000000 
>> Conv. in RMS gradients = YES 
>> --------------------------------------------------- 
>>
>> ******************************************************************************* 
>>
>> *** GEOMETRY OPTIMIZATION COMPLETED *** 
>> ******************************************************************************* 
>>
>>
>> Reevaluating energy at the minimum 
>>
>> Number of electrons: 8 
>> Number of occupied orbitals: 4 
>> Number of molecular orbitals: 4 
>>
>> Number of orbital functions: 23 
>> Number of independent orbital functions: 23 
>>
>> Extrapolation method: ASPC 
>>
>>
>> SCF WAVEFUNCTION OPTIMIZATION 
>>
>> Step Update method Time Convergence Total energy Change 
>> ------------------------------------------------------------------------------ 
>>
>> 1 Pulay/Diag. 0.50E+00 0.0 0.00005610 -17.1646204769 -1.72E+01 
>> 2 Pulay/Diag. 0.50E+00 0.1 0.00000563 -17.1646347273 -1.43E-05 
>>
>> *** SCF run converged in 2 steps *** 
>>
>>
>> Electronic density on regular grids: -8.0000016293 -0.0000016293 
>> Core density on regular grids: 7.9999992519 -0.0000007481 
>> Total charge density on r-space grids: -0.0000023775 
>> Total charge density g-space grids: -0.0000023775 
>>
>> Overlap energy of the core charge distribution: 0.00000004555422 
>> Self energy of the core charge distribution: -43.83289054591484 
>> Core Hamiltonian energy: 12.82175609863938 
>> Hartree energy: 17.97395116248504 
>> Exchange-correlation energy: -4.12745148803461 
>>
>> Total energy: -17.16463472727081 
>>
>> ENERGY| Total FORCE_EVAL ( QS ) energy [a.u.]: -17.164634727270808 
>>
>> -- 
>> You received this message because you are subscribed to the Google Groups 
>> "cp2k" group. 
>> To unsubscribe from this group and stop receiving emails from it, send an 
>> email to cp2k+uns... at googlegroups.com<mailto:cp2k+uns... at googlegroups.com>. 
>>
>> To view this discussion visit 
>> https://groups.google.com/d/msgid/cp2k/7bd75aef-e648-41ca-b4c0-cab444f39165n%40googlegroups.com
>> <
>> https://groups.google.com/d/msgid/cp2k/7bd75aef-e648-41ca-b4c0-cab444f39165n%40googlegroups.com?utm_medium=email&utm_source=footer>. 
>>
>>
>

-- 
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+unsubscribe at googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/96a1e419-e78a-44bd-a62b-a7a8318f5a26n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20241205/10e09954/attachment-0001.htm>


More information about the CP2K-user mailing list