KERNEL TRUNCATION/IPAD: Difference between revisions

Revision as of 08:22, 14 October 2025

Default: KERNEL_TRUNCATION/IPAD	= 3	if `KERNEL_TRUNCATION/IDIMENSIONALITY = 0`
	= 2	if `KERNEL_TRUNCATION/IDIMENSIONALITY = 2`

Description: KERNEL_TRUNCATION/IPAD controls the padding strategy used for the Coulomb kernel truncation in reciprocal space. Padding defines how much additional empty space is introduced around the charge density before applying truncation.^[1] This affects both the accuracy of the truncated Coulomb potential and the computational cost.

Setting KERNEL_TRUNCATION/IPAD allows fine control over how much zero-padding is applied along each reciprocal-space direction. Padding ensures that the truncated Coulomb kernel does not artificially interact with its periodic replicas in non-periodic directions.

Typically, increasing KERNEL_TRUNCATION/IPAD improves accuracy at the expense of computational cost.

Mind: KERNEL_TRUNCATION/LTRUNCATE must be set to .TRUE. for KERNEL_TRUNCATION/IPAD to have any effect.

Example

KERNEL_TRUNCATION {
     LTRUNCATE       = T
     IDIMENSIONALITY = 2
     ISURFACE        = 3
     IPAD            = 2
     FACTOR          = 1
}

In this case an additional empty cell is added along the z direciton as padding. The coulomb interaction is truncated beyond a z length. This ensures maximum usage of the simulation box.

KERNEL_TRUNCATION {
     LTRUNCATE       = T
     IDIMENSIONALITY = 2
     ISURFACE        = 3
     IPAD            = 1
     FACTOR          = 0.5
}

This setup corresponds to truncating the Coulomb interaction along the surface normal (z-direction) for a 2D material, using no vacuum padding and a truncation length of z/2. In this case, half of the simulation box is effectively unused, but the algorithm remains simpler. This configuration can be useful for debugging purposes.

References

↑ S. Vijay, M. Schlipf, H. Miranda, F. Karsai, M. Marsman, G. Kresse, arxiv (2024).

[vijay:arxiv:2024-1] S. Vijay, M. Schlipf, H. Miranda, F. Karsai, M. Marsman, G. Kresse, arxiv (2024).

[1]

@@ Line 35: / Line 35: @@
 </pre>
-This corresponds to truncation along the surface normal (z-direction) for a 2D material with no padding and z/2 length for the coulomb truncation.
+This setup corresponds to truncating the Coulomb interaction along the surface normal (z-direction) for a 2D material, using no vacuum padding and a truncation length of z/2. In this case, half of the simulation box is effectively unused, but the algorithm remains simpler. This configuration can be useful for debugging purposes.
-This setting corresponds to using coulomb truncation without padding. In this case half of the simulation box is in a sense wasted but the algorithm is simpler. This combination can be useful for debug.
 == Related tags and articles ==