# Query about pullf.xvg plot **URL:** \ **Category:** User discussions\ **Tags:** mdrun, umbrella-sampling, pull-code, domain-decomposition\ **Created:** [July 22, 2025, 12:33pm UTC](https://gromacs.bioexcel.eu/t/query-about-pullf-xvg-plot/12423 "2025-07-22T12:33:07Z")\ **Posts on this page:** 5\ **Page:** 1
**Author:** ![pratyush](https://avatars.discourse-cdn.com/v4/letter/p/cc9497/32.png) [@pratyush](https://gromacs.bioexcel.eu/u/pratyush)\ **Post date:** [July 22, 2025, 12:33pm UTC](https://gromacs.bioexcel.eu/t/query-about-pullf-xvg-plot/12423/1 "2025-07-22T12:33:07Z")
GROMACS version: 2024.2 GROMACS modification: Yes Based on the pull code below, I’m trying to determine whether the resulting force profile is physically reasonable. The force increases steadily in the first 800 ps, shows two more resistance peaks (around 2400 ps and 3000 ps), and then stabilizes around 500 kJ/mol/nm for the remaining part of the 5 ns simulation. There’s no sharp drop to zero, and the pulling distance increases smoothly throughout. Would this be considered a reasonable force profile for a loop-loop SMD simulation? This is my first time encountering such profile but I am unsure because I have only ever done SMD for systems having more than 2 entities but here, I am trying to pull two different parts of the same entity (protein). ; Pull code pull = yes pull\_ncoords = 1 pull\_ngroups = 2 pull\_group1\_name = a\_loop pull\_group2\_name = c\_loop pull\_coord1\_type = umbrella pull\_coord1\_geometry = distance pull\_coord1\_groups = 1 2 pull\_coord1\_start = yes pull\_coord1\_dim = N N Y pull\_coord1\_rate = 0.002 pull\_coord1\_k = 1000 pull-group1-pbcatom = 2803 pull-group2-pbcatom = 2304 pull-pbc-ref-prev-step-com = yes ![Screenshot 2025-07-22 171629](https://europe1.discourse-cdn.com/flex017/uploads/bioexcel1/original/2X/5/5f3ea141d657d5e809b2bd86553e9245d53dc56d.png) ---
**Author:** ![hess](https://dub1.discourse-cdn.com/flex017/user_avatar/gromacs.bioexcel.eu/hess/32/416_2.png) [@hess](https://gromacs.bioexcel.eu/u/hess)\ **Post date:** [July 22, 2025, 1:05pm UTC](https://gromacs.bioexcel.eu/t/query-about-pullf-xvg-plot/12423/2 "2025-07-22T13:05:35Z")
2 nm/ns is a very high pull rate. I would expect that your results are far from equilibrium. If I didn’t make a mistake, the integral of the plot above results in feeding 5000 kJ/mol into the system in those 5 ns. This is an enormous number. ---
**Author:** ![pratyush](https://avatars.discourse-cdn.com/v4/letter/p/cc9497/32.png) [@pratyush](https://gromacs.bioexcel.eu/u/pratyush)\ **Post date:** [July 23, 2025, 10:37am UTC](https://gromacs.bioexcel.eu/t/query-about-pullf-xvg-plot/12423/3 "2025-07-23T10:37:07Z")
Thank you for replying. The 0.002 (2nm/ns) is something I have reduced from 0.01 (10nm.ns), given in the supplied mdp file in [http://www.mdtutorials.com/gmx/umbrella/Files/md\_pull.mdp](http://www.mdtutorials.com/gmx/umbrella/Files/md_pull.mdp) . This was because I wanted to reduce the pull rate. If I have to go down further, do you suggest going beyond 0.001 (i.e. 1 nm/ns)? Do you also mean that the 5000 ps is a substantially high timescale for SMD and is unnecessary? So I should decrease that to 2000 ps or so? ---
**Author:** ![hess](https://dub1.discourse-cdn.com/flex017/user_avatar/gromacs.bioexcel.eu/hess/32/416_2.png) [@hess](https://gromacs.bioexcel.eu/u/hess)\ **Post date:** [July 23, 2025, 1:47pm UTC](https://gromacs.bioexcel.eu/t/query-about-pullf-xvg-plot/12423/4 "2025-07-23T13:47:50Z")
Your issue, which is a very common issue, is that you often need to pull extremely slow with SMD to get reasonable results. You would need to pull one or more orders of magnitude slower than 0.002 nm/ps. I don’t know what you are after. If you want a free-energy profile, umbrella sampling or AWH are better approaches. But even those might not converge within reasonable time. ---
**Author:** ![Sasha](https://avatars.discourse-cdn.com/v4/letter/s/54ee81/32.png) [@Sasha](https://gromacs.bioexcel.eu/u/Sasha)\ **Post date:** [July 23, 2025, 9:54pm UTC](https://gromacs.bioexcel.eu/t/query-about-pullf-xvg-plot/12423/5 "2025-07-23T21:54:25Z")
For what it’s worth, just to provide some scale context, when we absolutely need to pull something, the rate is typically ~1 nm over 1 microsecond. For proper energy estimates etc, we usually do umbrella sampling, just like hess is suggesting.