fix adapt with hybrid and multiple sub-styles of same type

Hello,

I'm trying to use 'fix adapt' with 'pair hybrid' using two copies of lj/cut.

Is this possible? I don't know how to the sub-style number to 'fix adapt'.

Thanks in advance,

Ryan

Hello,

I'm trying to use 'fix adapt' with 'pair hybrid' using two copies of
lj/cut.

Is this possible?

​currently not.

pair hybrid has to defer accessing the low level data structure to the
sub-styles ​and at the moment, the pointer to the first match will be
returned.

  I don't know how to the sub-style number to 'fix adapt'.

​it is more complex, we would have to come up with a convention to
communicate how to select a substyle.

one possible way would be to establish a convention that ​you would refer
to a data element of a substyle via:

lj/cut:epsilon
lj/cut:sigma

and for multiple styles using something like this:

lj/cut:1:epsilon
lj/cut:2:epsilon
​lj/cut:1:sigma
lj/cut:2:sigma​

​the the accessor string has no colon, then the old convention can still
apply, and thus backward compatibility can be maintained.​

​but before trying to implement this, i'd like to get steve's opinion or
alternate suggestion on this. this would also require to be made compatible
with suffixes, so it could require some careful coding.

axel.

Hi, Thanks for the reply.

I’m trying to create a system with two sets of atoms that start out identical but then over time in the simulation change the interaction parameters of one set of atoms and keep the others fixed. Say, two halves of a crystal where the natural lattice spacing increases over time while the other half stays fixed.

Is there a better (than ‘fix adapt’) way to do this?

Thanks,

Ryan

I'm trying to use 'fix adapt' with 'pair hybrid' using two copies of

lj/cut.

Is this possible?

​currently not.

Hi, Thanks for the reply.

I'm trying to create a system with two sets of atoms that start out
identical but then over time in the simulation change the interaction
parameters of one set of atoms and keep the others fixed. Say, two halves
of a crystal where the natural lattice spacing increases over time while
the other half stays fixed.

​if only one set needs to be changed, then you can still use fix adapt.
like i wrote, the current implementation will apply fix adapt to the first
matching substyle. it might need a little tinkering to get it to match the
one you need, but that should still work.​

​axel.​

I'm trying to create a system with two sets of atoms that start out
identical but then over time in the simulation change the interaction
parameters of one set of atoms and keep the others fixed. Say, two halves
of a crystal where the natural lattice spacing increases over time while
the other half stays fixed.

​if only one set needs to be changed, then you can still use fix adapt.
like i wrote, the current implementation will apply fix adapt to the first
matching substyle. it might need a little tinkering to get it to match the
one you need, but that should still work.​

Actually, it seems that the current code detects when multiple identical sub-styles exist and errors out, instead of returning the first matching substyle...

Ryan

I'm trying to create a system with two sets of atoms that start out
identical but then over time in the simulation change the interaction
parameters of one set of atoms and keep the others fixed. Say, two halves
of a crystal where the natural lattice spacing increases over time while
the other half stays fixed.

​if only one set needs to be changed, then you can still use fix adapt.
like i wrote, the current implementation will apply fix adapt to the first
matching substyle. it might need a little tinkering to get it to match the
one you need, but that should still work.​

Actually, it seems that the current code detects when multiple identical sub-styles exist and errors out, instead of returning the first matching substyle...

Ryan

I'm trying to create a system with two sets of atoms that start out

identical but then over time in the simulation change the interaction
parameters of one set of atoms and keep the others fixed. Say, two
halves
of a crystal where the natural lattice spacing increases over time while
the other half stays fixed.

​if only one set needs to be changed, then you can still use fix adapt.
like i wrote, the current implementation will apply fix adapt to the first
matching substyle. it might need a little tinkering to get it to match the
one you need, but that should still work.​

Actually, it seems that the current code detects when multiple identical
sub-styles exist and errors out, instead of returning the first matching
substyle...

​can you send me a test input? this must be failing at a different place
than what i was looking at.

axel.​

I'm trying to create a system with two sets of atoms that start out

identical but then over time in the simulation change the interaction
parameters of one set of atoms and keep the others fixed. Say, two
halves
of a crystal where the natural lattice spacing increases over time while
the other half stays fixed.

​if only one set needs to be changed, then you can still use fix adapt.
like i wrote, the current implementation will apply fix adapt to the first
matching substyle. it might need a little tinkering to get it to match the
one you need, but that should still work.​

Actually, it seems that the current code detects when multiple identical
sub-styles exist and errors out, instead of returning the first matching
substyle...

​here is an idea for a workaround: install the OPT package and use
lj/cut/opt for the style you don't want to change and lj/cut for the style
you do want to change. that should avoid the "two styles with the same
name" problem.

axel.​

Actually, it seems that the current code detects when multiple identical
sub-styles exist and errors out, instead of returning the first matching
substyle...

​can you send me a test input? this must be failing at a different place
than what i was looking at.

Hi Axel,

Here is an input file that fails with this error:

ERROR: Fix adapt pair style does not exist (../fix_adapt.cpp:306)

This comes from the Aug 13 git repo at commit 36315f3a96a745eca6bd904e05a7d62ec2ed6f7e

(Also, a heads up that git.lammps.org has been unresponsive for about a day now...)

Thanks,

Ryan

in.LJ_2 (1.47 KB)

Actually, it seems that the current code detects when multiple identical

sub-styles exist and errors out, instead of returning the first matching
substyle...

​can you send me a test input? this must be failing at a different place
than what i was looking at.

Hi Axel,

Here is an input file that fails with this error:

ERROR: Fix adapt pair style does not exist (../fix_adapt.cpp:306)

This comes from the Aug 13 git repo at commit
36315f3a96a745eca6bd904e05a7d62ec2ed6f7e

(Also, a heads up that git.lammps.org has been unresponsive for about a
day now...)

​i don't see any problems with git.lammps.org. i can access it from on- and
off-campus​ via git and http protocol.

axel.

Actually, it seems that the current code detects when multiple identical

sub-styles exist and errors out, instead of returning the first matching
substyle...

​can you send me a test input? this must be failing at a different place
than what i was looking at.

Hi Axel,

Here is an input file that fails with this error:

ERROR: Fix adapt pair style does not exist (../fix_adapt.cpp:306)

​i came up with a similar test. try the workaround i mentioned in the other
mail using lj/cut/opt to remove the ambiguity.

axel.​

(Also, a heads up that git.lammps.org has been unresponsive for about a
day now...)

​i don't see any problems with git.lammps.org. i can access it from on- and
off-campus​ via git and http protocol.

Hu.... All I get is time-out errors...

<1 Ryans-MacBook-Pro-100:lammps-ro >ping git.lammps.org
PING cms.icms.temple.edu (155.247.99.12): 56 data bytes
Request timeout for icmp_seq 0
Request timeout for icmp_seq 1
Request timeout for icmp_seq 2
Request timeout for icmp_seq 3
Request timeout for icmp_seq 4
^C
--- cms.icms.temple.edu ping statistics ---
6 packets transmitted, 0 packets received, 100.0% packet loss

(Also, a heads up that git.lammps.org has been unresponsive for about a

day now...)

​i don't see any problems with git.lammps.org. i can access it from on-
and
off-campus​ via git and http protocol.

Hu.... All I get is time-out errors...

<1 Ryans-MacBook-Pro-100:lammps-ro >ping git.lammps.org
PING cms.icms.temple.edu (155.247.99.12): 56 data bytes
Request timeout for icmp_seq 0
Request timeout for icmp_seq 1
Request timeout for icmp_seq 2
Request timeout for icmp_seq 3
Request timeout for icmp_seq 4
^C
--- cms.icms.temple.edu ping statistics ---
6 packets transmitted, 0 packets received, 100.0% packet loss

[[email protected] Downloads]$ ping git.lammps.org
PING cms.icms.temple.edu (155.247.99.12) 56(84) bytes of data.
64 bytes from cms.icms.temple.edu (155.247.99.12): icmp_seq=1 ttl=55
time=14.7 ms
64 bytes from cms.icms.temple.edu (155.247.99.12): icmp_seq=2 ttl=55
time=17.8 ms
64 bytes from cms.icms.temple.edu (155.247.99.12): icmp_seq=3 ttl=55
time=39.6 ms
64 bytes from cms.icms.temple.edu (155.247.99.12): icmp_seq=4 ttl=55
time=15.2 ms