Author Topic: Computed multigrid residual greater than required accuracy  (Read 3461 times)

0 Members and 1 Guest are viewing this topic.

Offline Adila

  • Heavy QuantumATK user
  • ***
  • Posts: 31
  • Country: my
  • Reputation: 0
    • View Profile
Dear all,

I am running phosphorene FET. However, while running its IV characteristics, a warning as stated in the title, "THE COMPUTED MULTIGRID RESIDUAL IS GREATER THAN THE REQUIRED ACCURACY" came upon in the Log Window. I would like to know, what is this about? It seems like my simulation has stopped running since the Log Window did not move. However, from the Job Manager, the simulation is still running. Should I terminate this simulation or not? Attached is my ".py file", the screenshot of the Log Window to show that the simulation has stopped running and the screenshot of the Job Manager to show that the simulation is still running.

Thank you in advance for your quick response.

Offline Jess Wellendorff

  • QuantumATK Staff
  • Supreme QuantumATK Wizard
  • *****
  • Posts: 933
  • Country: dk
  • Reputation: 29
    • View Profile
Re: Computed multigrid residual greater than required accuracy
« Reply #1 on: October 27, 2015, 16:02 »
You have used a MultiGrid algorithm for solving the Poisson Equation for the electrostatic potential. The warning says that a solution this equation was found, but that it was not quite as accurate as it was supposed to be. This is a Warning, so the calculation should keep on running. What version of VNL-ATK are you using? If it is not 2015.0, the Job Manager may not display the log file correctly.

Offline Anders Blom

  • QuantumATK Staff
  • Supreme QuantumATK Wizard
  • *****
  • Posts: 5576
  • Country: dk
  • Reputation: 96
    • View Profile
    • QuantumATK at Synopsys
Re: Computed multigrid residual greater than required accuracy
« Reply #2 on: October 27, 2015, 16:59 »
Your cell is very small in the Y direction, you probably need more vacuum. Also the model should in general be longer in Z, in the central region, for screening to be correctly described + you don't want the gate to appear too close to the edge of the electrodes.

Also always use Dirichlet for the edge where you have the gate, not Neumann (the other side can be Neumann).