Different results by LDA and GGA

[yongjunwinwin]

I calculated the band structure of monolayer MoTe2 respectively with LDA and GGA, but got the complete different result! The band structure by LDA suggest metal property while GGA suggest semiconductor property with a direct band gap of 1.118 eV. What cause the difference and which result is right? (The results and the script  of the unite cell of monolayer MoTe2 is attached)

[zh]

It has been reported in literature that the single atomic sheet of MoS2 is a semiconductor. If the LDA and GGA band structures are calculated for a same lattice of MoTe (i.e., the lattice constants and atomic coordinates in two cases are same), I guess that the possible reason comes from the  LDA pseudoptentials of Mo or Te. 

Now you had better check the LDA and GGA band structure of bulk MoTe2, and compare them with the experimental results and the previous theoretical work:
Th. Böker, et al, Phys. Rev. B 64, 235305 (2001), Band structure of MoS2, MoSe2, and α-MoTe2: Angle-resolved photoelectron spectroscopy and ab initio calculations
http://prb.aps.org/abstract/PRB/v64/i23/e235305 

[Anders Blom]

The results for LDA look strange, and actually the LDA basis set for Te looks wrong too (you can inspect it in VNL 11.2 by choosing Analyzer>Basis Set Explorer). There are 2 valence states (5s,5p), but if we check the DoubleZeta basis set, for GGA it looks normal, but in LDA we have 3 functions with nodes at r=0. If you zoom in, however, one of them is actually finite at r=0, but much too small. Also, the s state seems completely linear for small r.

Since the GGA results seem reasonable, you should definitely use GGA for this system.