Fix reinitialization issues when using total forces in NAMD#887

Merged

HanatoK merged 25 commits intomasterfrom

remove_unused_member

Dec 17, 2025

Commits on Dec 12, 2025

cleanup: Remove unused flag

giacomofiorin
authored and
HanatoK
committed
fix: try fixing the crash
HanatoK
committed
opt: Do not invalidate NAMD GlobalMaster arrays when not needed

giacomofiorin
authored and
HanatoK
committed
fix: Update atoms_map in NAMD proxy after requesting new atoms

giacomofiorin
authored and
HanatoK
committed
test: force the boundary checks to reveal the bug
HanatoK
committed
fix: reset the atoms_map to -1 instead of clearing it
HanatoK
committed
test: try the 003_reinitatoms test on ARM64
HanatoK
committed
test: remove the skip_test.sh and try the test with ARM64 again
HanatoK
committed
test: revert the previous commits to disable ARM64 test again
HanatoK
committed
fix: Track when total forces from the engine are valid or not

giacomofiorin
authored and
HanatoK
committed
test: Update regtest reference file to remove nan

giacomofiorin
authored and
HanatoK
committed
test: Re-enable 003_reinitatoms test in ARM64

giacomofiorin
authored and
HanatoK
committed
doc: Set bug error flag upon bug occurrence

giacomofiorin
authored and
HanatoK
committed
fix: Backport changes from LAMMPS repository

giacomofiorin
authored and
HanatoK
committed
test: Try downloading container a couple of times before raising error

giacomofiorin
authored and
HanatoK
committed
fix: Fix container name in CI job

giacomofiorin
authored and
HanatoK
committed
fix: Access colvarproxy_lammps object from correct rank

giacomofiorin
authored and
HanatoK
committed
fix: skip the total force calc in the GPU code path if unavailable
HanatoK
committed
refactor: Invalidate colvar total force from a function used also by GPU scheduler

giacomofiorin
authored and
HanatoK
committed
refactor: Define accessor to reset the colvar total force

giacomofiorin
authored and
HanatoK
committed
fix: Apply consistent check between total-force compute / collect steps

giacomofiorin
authored and
HanatoK
committed
refactor: When atomic total forces are invalid, ensure colvar total force is zero after collection

giacomofiorin
authored and
HanatoK
committed
fix: Ensure that reported colvar total force is zeroed immediately

giacomofiorin
authored and
HanatoK
committed
fix: ensure that total forces are reset before calc_colvar_properties if
HanatoK
committed
fix: fix the typos in previous rebase
HanatoK
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fix reinitialization issues when using total forces in NAMD#887

Fix reinitialization issues when using total forces in NAMD#887
HanatoK merged 25 commits intomasterfrom
remove_unused_member

Commits on Dec 12, 2025

cleanup: Remove unused flag

fix: try fixing the crash

opt: Do not invalidate NAMD GlobalMaster arrays when not needed

fix: Update atoms_map in NAMD proxy after requesting new atoms

test: force the boundary checks to reveal the bug

fix: reset the atoms_map to -1 instead of clearing it

test: try the 003_reinitatoms test on ARM64

test: remove the skip_test.sh and try the test with ARM64 again

test: revert the previous commits to disable ARM64 test again

fix: Track when total forces from the engine are valid or not

test: Update regtest reference file to remove nan

test: Re-enable 003_reinitatoms test in ARM64

doc: Set bug error flag upon bug occurrence

fix: Backport changes from LAMMPS repository

test: Try downloading container a couple of times before raising error

fix: Fix container name in CI job

fix: Access colvarproxy_lammps object from correct rank

fix: skip the total force calc in the GPU code path if unavailable

refactor: Invalidate colvar total force from a function used also by GPU scheduler

refactor: Define accessor to reset the colvar total force

fix: Apply consistent check between total-force compute / collect steps

refactor: When atomic total forces are invalid, ensure colvar total force is zero after collection

fix: Ensure that reported colvar total force is zeroed immediately

fix: ensure that total forces are reset before calc_colvar_properties if

fix: fix the typos in previous rebase