Hello Ruslan Zhachuk,
Your answer helps a lot! Thank you very much! Best wishes! Leila 发件人: Руслан Жачук [mailto:[email protected]] 发送时间: 2009年12月17日 13:41 收件人: [email protected] 主题: Re: [SIESTA-L] 2 questions about Meshcutoff 2) I learned from Eur. Phys. J. B 40, 371-377(2004) that: Hamiltonian matrix elements are partly computed on a real space grid, whose fineness Δx is controlled by a grid cutoff, Ec=(p/Δx)2/2. Does it mean that the charge density is represented on a regular real space grid ofΔx? If not, where can I find the regular real space grid that represents the charge density? And what are the units ofΔx and Ec in that formula? I will be grateful if anyone can help. Thanks in advance. Best wishes! Leila Hello Leila, Yes, the density is represented on a regular real space grid ofΔx, but the equation Ec=(p/Δx)2/2 is valid only when you have orthogonal unit cell. It is in atomic units (Δx - Bohr, Ec - Rydbergs). The stiffness of this grid is controlled by MeshCutoff parameter. If you look at your Siesta output you''l find something like: outcell: Cell vector modules (Ang) : 27.174106 27.174107 25.000000 outcell: Cell angles (23,13,12) (deg): 90.0000 90.0000 60.0000 outcell: Cell volume (Ang**3) : 15987.5228 InitMesh: MESH = 216 x 216 x 216 = 10077696 InitMesh: Mesh cutoff (required, used) = 200.000 206.314 Ry Here 216 is the number of grid points along each spanning vector of the unit cell. If you want to have access to charge density values in real space grid, then yo'll need to find the file m_iorho.F in siesta/Src, change line 37 from character(len=*), parameter :: fform = "unformatted" to character(len=*), parameter :: fform = "formatted" and then recompile Siesta. Then add SaveRho .true. in your fdf file and you 'll get readable *.RHO file. Kind regards Ruslan Zhachuk
