QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: Dipankar Saha on November 8, 2013, 15:08

Title: computed multigrid residual is greater than the required accuracy
Post by: Dipankar Saha on November 8, 2013, 15:08

WARNING                                                             
# The computed multigrid residual is greater than the required accuracy.     
# Computed residual :   1.20533e-11                                         
# Required accuracy :   1.00000e-12                                           

............ What does this warning message signify???                                 

Title: Re: computed multigrid residual is greater than the required accuracy
Post by: kstokbro on November 8, 2013, 15:27

This error is related to the accuracy of the multigrid solver. 1e-11 is still a very small number, thus, I think it is safe to continue. However, please send your inputfile to [email protected], with an explanation, such that we can check the problem.

Title: Re: computed multigrid residual is greater than the required accuracy
Post by: Dipankar Saha on November 8, 2013, 18:38

Okay... I'm continuing with it.... / Besides, I will mail the *.py file with due explanations....to [email protected]
Thanks  :)

Title: Re: computed multigrid residual imuch much greater than the required accuracy
Post by: tara on September 22, 2014, 15:29

 The computed multigrid residual is greater than the required accuracy.     
# Computed residual :   1.20533e33                                         
# Required accuracy :   1.00000e-12
                     
I have used multigrid with periodic boundary contions in A and B directions in my structure that is gated .

what could be the reason for such huge difference.??

Title: Re: computed multigrid residual is greater than the required accuracy
Post by: Anders Blom on September 22, 2014, 21:08

Please tell us how we are supposed to help you when we have zero idea of what calculation you are performing, what input parameters are used, etc?

Title: Re: computed multigrid residual is greater than the required accuracy
Post by: tara on September 23, 2014, 12:01

sorry ,i forgot to mention the calculation and input parameters.

I am using DFT calculation,with poisson solver setting as multigrid.I have given boundary conditions as periodic in A and B directions.My structure is p-i-n gated transistor. For doping i have replaced some  Si atoms with boron atoms for p-type doping and with phosphorous atoms for n -type.

i am getting this warning repeatedly

The computed multigrid residual is greater than the required accuracy.     
# Computed residual :   1.20533e33                                         
# Required accuracy :   1.00000e-12

why am i getting such huge difference??

Title: Re: computed multigrid residual is greater than the required accuracy
Post by: Anders Blom on September 23, 2014, 22:53

Attach the whole script, otherwise we can't reproduce it.

Title: Re: computed multigrid residual is greater than the required accuracy
Post by: tara on September 24, 2014, 14:28

i am atttaching my script here.

please help as to why i am getting huge difference in required and obtained values

Title: Re: computed multigrid residual is greater than the required accuracy
Post by: Anders Blom on September 24, 2014, 14:35

Your gate is outside the simulation cell and therefore not included in the calculation. Moreover it looks like you attempt to have periodic boundary conditions in the Y direction while at the same time having a gate there, this will not work. If you are using gates, you need Dirichlet on the gate (cf. http://quantumwise.com/forum/index.php?topic=1597.0).

Also, you will need to passivate all the Si surface atoms to eliminate dangling bonds (this is not related to the multigrid residual though).