Page **1** of **1**

### temperature in GW calculation

Posted:

**Mon Oct 15, 2018 2:57 am**
by **Weiqing Zhou**

Hi all,

Is there a way to introduce temperature in the calculations of Green's function/self-energy?

Best,

Weiqing

### Re: temperature in GW calculation

Posted:

**Mon Oct 15, 2018 7:59 am**
by **Daniele Varsano**

Dear Weiqing,

what exactly you are interested in with temperature?

Do you mean electronic temperature in occupation number?, temperature effects via el-ph coupling?

These two features is possible to take into accounts.

Best,

Daniele

### Re: temperature in GW calculation

Posted:

**Mon Apr 01, 2019 10:15 am**
by **Weiqing Zhou**

Dear Daniele,

Now, I want to obtain imaginary part of self-energy operator from GW calculation.

1. First of all, I want to ensure that the column Width[meV] in o. file is imaginary part of self-energy, Right?

2. I want to check the temperature-dependent self-energy, should I recalculate GW calculation at the very begining if I change ElecTemp? Can I use previous emld and dipole file?

Best,

Weiqing

### Re: temperature in GW calculation

Posted:

**Mon Apr 01, 2019 10:37 am**
by **Daniele Varsano**

Dear Weiqing,

Note that in order to have access to the imaginary part of the self energy the plasmon-pole approximation is not suitable and you need to calculate the GW quasiparticle energy (and lifetimes) using a sull frequency approach.

1. First of all, I want to ensure that the column Width[meV] in o. file is imaginary part of self-energy, Right?

Right

I want to check the temperature-dependent self-energy, should I recalculate GW calculation at the very begining if I change ElecTemp? Can I use previous emld and dipole file?

Note that the electronic temperature only affects occupation number of your system. Beside electron-electron interaction, also electron-phonon interaction affects the temperature dependence of the lifetime. see e.g.

https://journals.aps.org/prb/abstract/1 ... .89.085202https://arxiv.org/pdf/1603.06965.pdfBest,

Daniele

### Re: temperature in GW calculation

Posted:

**Fri Apr 05, 2019 3:04 pm**
by **Weiqing Zhou**

Dear Daniele,

Many thanks!

I do some G0W0 calculations with varying ElecTemp setting to check the temperature-dependent self energy and Electron-electron lifetime. You can check the input file in attachment. But it is very weird that lifetimes are almost same with different ElecTemp. I think e-e lifetime should decrease as temperature increase due to stronger scattering. what's problem?

Best,

Weiqing

### Re: temperature in GW calculation

Posted:

**Fri Apr 05, 2019 3:11 pm**
by **Daniele Varsano**

Dear Weiqing,

as I was telling you in the previous post, the electronic temperature here enters in the occupation number only (Fermi distribution function). I do not know what system you are dealing with. For instance, in the case of a semiconductor, the electronic temperature does not affect the occupation number (hence the result of the calculation) until you have temperature sufficiently large, compared with the gap, to populate the conduction band. Different is the case with metals.

Best,

Daniele

### Re: temperature in GW calculation

Posted:

**Fri Apr 05, 2019 4:07 pm**
by **Weiqing Zhou**

Dear Daniele,

Thanks for patience!

My system is metal. You can check the report and output file in the attachment. From o.qp, you can see self energy is almost same for 50K and 300K. And I understand I can calculate temperature-dependent E-P lifetime by Yambo. Now I want to make clear that can I calculate temperature-dependent E-E lifetime by yambo, in principle?

Best,

Weiqing

### Re: temperature in GW calculation

Posted:

**Mon Apr 08, 2019 10:46 am**
by **Daniele Varsano**

Dear Weiking,

Now I want to make clear that can I calculate temperature-dependent E-E lifetime by yambo, in principle?

Please have a look to the new yambo paper:

https://arxiv.org/pdf/1902.03837.pdfand reference therein.

In section IVB it is described what yambo does for the e-e contribution to the lifetime.

About your results, I've no experience on this calculations so do not have the feeling on how much it should change from 50K to 300K and I cannot tell you if the small variation you are seeing are what is expected.

Best,

Daniele