Dear PWscf users,<br><br>I have some questions during my phonon calculations using PWscf 3.2.2. <br><br>1) If I switch the iverbosity=.true. in the input file, the output of ph.x will list the eigenvectors for each mode before entering the scf iterations. If I understand correctly, the code can determine an initial displacement patterns even WITHOUT solving the linear system. I looked into the source code and found that this is done in the set_irr.f90 subroutine. The comment line says:
<br><br>! This subroutine computes a basis for all the irreducible<br>! representations of the small group of q, which are contained<br>! in the representation which has as basis the displacement vectors.<br>! This is achieved by building a random hermitean matrix,
<br>! symmetrizing it and diagonalizing the result. The eigenvectors<br>! give a basis for the irreducible representations of the<br>! small group of q.<br><br>I noticed that the calculation of the first-order perturbation dVscf
also needs the initial displacement pattern for each mode to calculate
the contribution of the local potential (in dvqpsi_us.f90). Can anybody here provide me more details about the algorithm? thanks<br><br>2) I try to output the dVscf for each mode from the solve_linter.f90 after the convergence has been reached. However, it seems that the output value will change when I increase the Ecut (the FFT grid and also the symmetry operations will be different comparing with previous case). In fact, in all cases the Ecut has been already converged. I believe there should be something missed in my output of dVscf. Can anybody here give me some hints about that? Thanks!
<br><br>Best,<br>Jia-An Yan<br><br>---------------------------------------------------------------------------------------------------------<br>School of Physics, <br>Georgia Institute of Technology<br>837 State Street,<br>
Atlanta, Georgia 30332<br>U.S.A.<br>---------------------------------------------------------------------------------------------------------<br><br><br><br><br>