Fail to constrain a cubic cell with fixed shape

[xiancai]

Dear VASP developers and users,

I performed a FPMD simulation in the NPT ensemble using VASP 6.4.1, during which the simulation cell was constrained as a cubic cell with variable volume but fixed shape through the ICONST file. However, I found that the side lengths were not strictly equal to each other, and the angles among lattice vectors were not strictly constrained either. All input files (i.e., INCAR, ICONST, POSCAR, POTCAR, and KPOINTS) as well as the outputted XDATCAR and CONTCAR files, are provided as attachments. I am wondering if this phenomenon is normal, or if there are any problems in my input files.

Thank you for your time and I am looking forward to your reply.
Sincerely
Hua Tian

[alexey.tal]

Dear Hua Tian,

However, I found that the side lengths were not strictly equal to each other, and the angles among lattice vectors were not strictly constrained either.

I see that the difference in the lattice vectors is <1E-6, which is within the default accuracy.
VASP uses the Shake and Rattle algorithm to impose constraints on the system, as for example the lattice. The accuracy of this method can be set by the SHAKETOL tag.
You can also increase the maximum number of steps of the shake algorithm for better accuracy with SHAKEMAXITER

[xiancai]

Dear Dr. Tal,

Thank you very much for your guidance! I will try using these two keywords to reduce the differences.

Sincerely，
Tian Hua

Dear Hua Tian,

However, I found that the side lengths were not strictly equal to each other, and the angles among lattice vectors were not strictly constrained either.

I see that the difference in the lattice vectors is <1E-6, which is within the default accuracy.
VASP uses the Shake and Rattle algorithm to impose constraints on the system, as for example the lattice. The accuracy of this method can be set by the SHAKETOL tag.
You can also increase the maximum number of steps of the shake algorithm for better accuracy with SHAKEMAXITER