About convergence

[lesheng]

Dear all,

I am a new user of Yambo, and after I followed the tutorial on Si to perform the GW calculation, I have a general question about the convergence test. For each system I am dealing with, I have to carry out a set of calculations with different variables (such as EXXRLvcs, BndsRnX, GbndRnge etc) in order to determine the "best" value for these variables, so I can obtain an accurate result (with convergence outcomes). And such a strategy will be absolutely "time expensive". Is that correct?

Thanks,

[Daniele Varsano]

Dear Lesheng,
the heaviness of the convergence studies depends totally on your system size (number of electrons, k-points etc..).
Most of them are independent so you can push a variable keeping the others low (but reasonable) in order to assess the each converged
value and finally run the production calculations.

Cheers,

Daniele

[lesheng]

Dear Lesheng,
the heaviness of the convergence studies depends totally on your system size (number of electrons, k-points etc..).
Most of them are independent so you can push a variable keeping the others low (but reasonable) in order to assess the each converged
value and finally run the production calculations.

Cheers,

Daniele

Thank you Daniele, and others in this forum, I find very helpful information during reading these posts.

Thanks,

Lesheng Li

[claudio]

Dear Lesheng,

moreover I advise you to use mHa as unit for the different parameters instead of RL (reciprocal lattice vectors = numer of g-vectors).
The mHa unit is in some sense less system dependent respect to vacuuum etc..

Cla

[lesheng]

Dear Lesheng,

moreover I advise you to use mHa as unit for the different parameters instead of RL (reciprocal lattice vectors = numer of g-vectors).
The mHa unit is in some sense less system dependent respect to vacuuum etc..

Cla

Dear Cla,

Sorry I just notice your reply, and thank you for your advice, I'll try this way and compare the difference.

Thank you again.

Best,
lesheng