# Temperature coupling in Gromacs membrane tutorials

**URL:** <https://gromacs.bioexcel.eu/t/temperature-coupling-in-gromacs-membrane-tutorials/9393>\
**Category:** User discussions\
**Tags:** mdp-parameters\
**Created:** [June 14, 2024, 3:50pm UTC](https://gromacs.bioexcel.eu/t/temperature-coupling-in-gromacs-membrane-tutorials/9393 "2024-06-14T15:50:22Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![Dulaj](https://dub1.discourse-cdn.com/flex017/user_avatar/gromacs.bioexcel.eu/dulaj/32/3408_2.png) [@Dulaj](https://gromacs.bioexcel.eu/u/Dulaj)\
**Post date:** [June 14, 2024, 3:50pm UTC](https://gromacs.bioexcel.eu/t/temperature-coupling-in-gromacs-membrane-tutorials/9393/1 "2024-06-14T15:50:22Z")

</div>

GROMACS version: 2024.2  
GROMACS modification: Yes/No  
Here post your question

I have heard that temperature coupling is done separately for multiple sub-groups.  
For example;

```auto
tcoupl = berendsen
tc_grps = SOLU MEMB SOLV
tau_t = 1.0 1.0 1.0
ref_t = 303.15 303.15 303.15

```

However, in the Gromacs tutorials found on https://tutorials.gromacs.org/membrane-protein.html, temperature coupling have been done in this way,

```auto
;;Temperature coupling 
tcoupl = v-rescale ; Modified version of Berendsen thermostat, Temperature coupling using velocity rescaling with a stochastic term 
tc_grps = SYSTEM ; Coupling groups 
tau_t = 1.0 ; Time constant [ps]
ref_t = 300 ; Reference temperature [K]

```

I know that V-rescale is better than Berendsen.  
However, is it accurate enough to couple the whole system into one group with even V-rescale thermostat? Would this be accepted in the publication level?

---

<div class="post-metadata">

**Author:** ![jalemkul](https://dub1.discourse-cdn.com/flex017/user_avatar/gromacs.bioexcel.eu/jalemkul/32/18_2.png) [@jalemkul](https://gromacs.bioexcel.eu/u/jalemkul)\
**Post date:** [June 14, 2024, 6:48pm UTC](https://gromacs.bioexcel.eu/t/temperature-coupling-in-gromacs-membrane-tutorials/9393/2 "2024-06-14T18:48:57Z")

</div>

There is no reason to couple groups separately. And don’t use Berendsen; it’s provably invalid as far as maintaining a canonical ensemble.