From a64eb330e3a26fab7ae97a0d69e9aae764afee88 Mon Sep 17 00:00:00 2001 From: Axel Kohlmeyer Date: Sun, 11 Sep 2016 14:48:37 -0400 Subject: [PATCH] document how using shrink-wrap boundaries can cause lost atoms in parallel --- doc/src/boundary.txt | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) diff --git a/doc/src/boundary.txt b/doc/src/boundary.txt index e832a74d24..91096589a7 100644 --- a/doc/src/boundary.txt +++ b/doc/src/boundary.txt @@ -54,7 +54,15 @@ allow for lost atoms. For style {s}, the position of the face is set so as to encompass the atoms in that dimension (shrink-wrapping), no matter how far they -move. +move. Note that when the difference between the current box dimensions +and the shrink-wrap box dimensions is large, this can lead to lost +atoms at the beginning of a run when running in parallel. This is due +to the large change in the (global) box dimensions also causing +significant changes in the individual sub-domain sizes. If these +changes are farther than the communication cutoff, atoms will be lost. +This is best addressed by setting initial box dimensions to match the +shrink-wrapped dimensions more closely, by using {m} style boundaries +(see below). For style {m}, shrink-wrapping occurs, but is bounded by the value specified in the data or restart file or set by the