# Z‑dimension of the box decreases **URL:** \ **Category:** User discussions\ **Tags:** mdp-parameters\ **Created:** [July 30, 2026, 3:28am UTC](https://gromacs.bioexcel.eu/t/z-dimension-of-the-box-decreases/13631 "2026-07-30T03:28:40Z")\ **Posts on this page:** 4\ **Page:** 1
**Author:** ![ZXY](https://avatars.discourse-cdn.com/v4/letter/z/f4b2a3/32.png) [@ZXY](https://gromacs.bioexcel.eu/u/ZXY)\ **Post date:** [July 30, 2026, 3:28am UTC](https://gromacs.bioexcel.eu/t/z-dimension-of-the-box-decreases/13631/1 "2026-07-30T03:28:40Z")
GROMACS version:2018 GROMACS modification: Yes/No Here post your question “Dear all, I am using the Martini 3‑IDP force field to simulate a system consisting of a spherical protein condensate and a cell membrane (DIPC/DIPS) with initial box dimensions of 30 x 30 x 30 nm, built using the command insane -f lk.gro -pbc rectangular -salt 0.15 -x 30 -y 30 -z 30 -sol W -o system.gro -u DIPC:50 -u DIPS:50 -l DIPC:50 -l DIPS:50 -center -dm 9.0. During the production MD run, I observe that the Z‑dimension of the box continuously decreases while the XY area increases, despite using Pcoupltype = semiisotropic. I have already tried changing tau\_p from 12.0 to 5.0 and compressibility from 3e‑4 3e‑4 to 4.5e‑5 4.5e‑5, but the problem persists. Could anyone suggest how to resolve this box compression issue? Thank you.” ; ; STANDARD MD INPUT OPTIONS FOR MARTINI 2.x ; Updated 15 Jul 2015 by DdJ ; ; for use with GROMACS 5 ; For a thorough comparison of different mdp options in combination with the Martini force field, see: ; D.H. de Jong et al., Martini straight: boosting performance using a shorter cutoff and GPUs, submitted. title = Martini ; TIMESTEP IN MARTINI ; Most simulations are numerically stable with dt=40 fs, ; however better energy conservation is achieved using a ; 20-30 fs timestep. ; Time steps smaller than 20 fs are not required unless specifically stated in the itp file. integrator = md dt = 0.01 nsteps = 1000000000 nstcomm = 100 comm-grps = nstxout = 0 nstvout = 0 nstfout = 0 nstlog = 1000 nstenergy = 100 nstxout-compressed = 10000 compressed-x-precision = 300 compressed-x-grps = energygrps = system ; NEIGHBOURLIST and MARTINI ; To achieve faster simulations in combination with the Verlet-neighborlist ; scheme, Martini can be simulated with a straight cutoff. In order to ; do so, the cutoff distance is reduced 1.1 nm. ; Neighborlist length should be optimized depending on your hardware setup: ; updating ever 20 steps should be fine for classic systems, while updating ; every 30-40 steps might be better for GPU based systems. ; The Verlet neighborlist scheme will automatically choose a proper neighborlist ; length, based on a energy drift tolerance. ; ; Coulomb interactions can alternatively be treated using a reaction-field, ; giving slightly better properties. ; Please realize that electrostVatic interactions in the Martini model are ; not considered to be very accurate to begin with, especially as the ; screening in the system is set to be uniform across the system with ; a screening constant of 15. When using PME, please make sure your ; system properties are still reasonable. ; ; With the polarizable water model, the relative electrostatic screening ; (epsilon\_r) should have a value of 2.5, representative of a low-dielectric ; apolar solvent. The polarizable water itself will perform the explicit screening ; in aqueous environment. In this case, the use of PME is more realistic. cutoff-scheme = Verlet nstlist = 30 ns\_type = grid pbc = xyz verlet-buffer-tolerance = 0.005 coulombtype = reaction-field rcoulomb = 1.1 epsilon\_r = 15 ; 2.5 (with polarizable water) epsilon\_rf = 0 vdw\_type = cutoff vdw-modifier = Potential-shift-verlet rvdw = 1.1 ; MARTINI and TEMPERATURE/PRESSURE ; normal temperature and pressure coupling schemes can be used. ; It is recommended to couple individual groups in your system separately. ; Good temperature control can be achieved with the velocity rescale (V-rescale) ; thermostat using a coupling constant of the order of 1 ps. Even better ; temperature control can be achieved by reducing the temperature coupling ; constant to 0.1 ps, although with such tight coupling (approaching ; the time step) one can no longer speak of a weak-coupling scheme. ; We therefore recommend a coupling time constant of at least 0.5 ps. ; The Berendsen thermostat is less suited since it does not give ; a well described thermodynamic ensemble. ; ; Pressure can be controlled with the Parrinello-Rahman barostat, ; with a coupling constant in the range 4-8 ps and typical compressibility ; in the order of 10e-4 - 10e-5 bar-1. Note that, for equilibration purposes, ; the Berendsen barostat probably gives better results, as the Parrinello- ; Rahman is prone to oscillating behaviour. For bilayer systems the pressure ; coupling should be done semiisotropic. tcoupl = V-rescale tc-grps = Protein Bilayer SOL tau\_t = 1.0 1.0 1.0 ref\_t = 295 295 295 Pcoupl = Parrinello-rahman Pcoupltype = semiisotropic tau\_p = 12 ;parrinello-rahman is more stable with larger tau-p, DdJ, 20130422 compressibility = 3e-4 3e-4 ref\_p = 1.0 1.0 gen\_vel = no gen\_temp = 295 gen\_seed = -1 ; MARTINI and CONSTRAINTS ; for ring systems and stiff bonds constraints are defined ; which are best handled using Lincs. constraints = none constraint\_algorithm = Lincs ---
**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 30, 2026, 7:16am UTC](https://gromacs.bioexcel.eu/t/z-dimension-of-the-box-decreases/13631/2 "2026-07-30T07:16:03Z")
That is not “despite” semi-isotropic coupling, as that is exactly what allows this. One reason could be that your protein is not well embedded in the membrane, causing large lateral forces. tau\_p=1 is far too short for Parrinello-Rahman. And PR-coupling can be unstable, especially during equilibration. We recommend using C-rescale with tau\_p at least 5 ps. ---
**Author:** ![ZXY](https://avatars.discourse-cdn.com/v4/letter/z/f4b2a3/32.png) [@ZXY](https://gromacs.bioexcel.eu/u/ZXY)\ **Post date:** [August 1, 2026, 2:10am UTC](https://gromacs.bioexcel.eu/t/z-dimension-of-the-box-decreases/13631/3 "2026-08-01T02:10:14Z")
Thank you for your reply. Unfortunately, we uniformly use GROMACS 2018, which does not support C‑rescale, and my tau‑p is set to 12. However, I have now resolved this issue. The reason was that the membrane built by insane is initially too regular and dense, causing the membrane area to expand during the simulation. As a result, the box area in the xy plane also increases, while the box length along the z axis decreases accordingly. The solution was to first simulate the membrane alone in a solvent environment until the membrane had expanded to its maximum extent, and then place the protein and the membrane into a new box, which solved the problem. Finally, the initial state of my system is as follows: ![image](https://europe1.discourse-cdn.com/flex017/uploads/bioexcel1/original/2X/a/a57d72bd3474d53d4d7e080cc4537ac5e5e6bd0b.jpeg) ---
**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:** [August 5, 2026, 10:09am UTC](https://gromacs.bioexcel.eu/t/z-dimension-of-the-box-decreases/13631/4 "2026-08-05T10:09:26Z")
Why use an 8 year old version??