ERROR: the system is looking for a value outside the grid along the 1 index

2,567 views
Skip to first unread message

liyi....@gmail.com

unread,
Jun 14, 2017, 11:37:37 PM6/14/17
to PLUMED users
Hi erveryone,

I use PLUMED2.3 with Gromacs.
I want to limit the the range of CVs

UNITS LENGTH=nm

c_b2: COM ATOMS=723,730,739
c_b3: COM ATOMS=1353,1361
c_b20: COM ATOMS=3075,3083,3091
c_b21: COM ATOMS=3150,3159,3164

d_b2_21: DISTANCE ATOMS=c_b2,c_b21
d_b3_21: DISTANCE ATOMS=c_b3,c_b21

# state    b2_21 b3_21
# apo      0.842 0.473
# liganded 0.490 0.973

uwall: UPPER_WALLS ARG=d_b2_21,d_b3_21 AT=1.0,1.0 KAPPA=5000,5000 EPS=50.0,50.0
lwall: LOWER_WALLS ARG=d_b2_21,d_b3_21 AT=0.0,0.0 KAPPA=5000,5000 EPS=50.0,50.0

meta: METAD ARG=d_b2_21,d_b3_21 HEIGHT=0.001 SIGMA=0.001,0.001 PACE=50 GRID_MIN=0,0 GRID_MAX=1,1 GRID_SPACING=0.01,0.01 FILE=HILLS BIASFACTOR=10 TEMP=300.0

# monitor the two variables and the metadynamics bias potential
PRINT STRIDE=10 ARG=* FILE=COLVAR

I have tried the different values of KAPPA, EPS HIGHT, SIGMA. But it still not working.
Why the system is always looking the outside values?

How to make it runs?
Thanks.

Li Yi

Massimiliano Bonomi

unread,
Jun 15, 2017, 2:05:20 AM6/15/17
to plumed...@googlegroups.com
Hello,

does your simulation always stop right at the beginning?
Have you checked the initial values of your CVs?

As side notes:

1) I would remove:

> GRID_SPACING=0.01,0.01

and let plumed choose this parameter for you.
Consider that this must be smaller than SIGMA

2) Be careful in choosing these parameters

> HEIGHT=0.001 SIGMA=0.001,0.001 PACE=50

In biological applications, HEIGHT is typically of the order of 1.0 kjoule/mol,
SIGMA seems too small (see previous posts and the literature about how to
choose this parameter), PACE is typically of the order of 500 (time steps).

3) you can extend GRID_MAX a bit more beyond your UPPER_WALLS,
since it is not a hard wall.

Max
> --
> You received this message because you are subscribed to the Google Groups "PLUMED users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users...@googlegroups.com.
> To post to this group, send email to plumed...@googlegroups.com.
> Visit this group at https://groups.google.com/group/plumed-users.
> To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/406d6e8b-9f4a-45ea-a0c6-3085ae21d0a5%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

Giovanni Bussi

unread,
Jun 15, 2017, 2:37:17 AM6/15/17
to plumed...@googlegroups.com
Some comments (in addition to the ones from Max):

1. You are computing distances, so the lower wall at zero is useless.
2. With your settings (EPS=50 KAPPA=5000) the penalty for a violation of 0.1 nm is going to be 5000 * (0.1/50)**2=0.02 kj/mol. A violation of 1 nm will result in 2 kj/mol. Why did you choose this large EPS?
3. If you set the upper wall at 1, even with a large kappa / small eps you should expect some violation. Violations will be even larger if you do METAD on that distance. So, set GRID_MAX to something like 1.5 to be sure METAD does not crash for out of grid access.

Giovanni


> To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users+unsubscribe@googlegroups.com.

> To post to this group, send email to plumed...@googlegroups.com.
> Visit this group at https://groups.google.com/group/plumed-users.
> To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/406d6e8b-9f4a-45ea-a0c6-3085ae21d0a5%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "PLUMED users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users+unsubscribe@googlegroups.com.

To post to this group, send email to plumed...@googlegroups.com.
Visit this group at https://groups.google.com/group/plumed-users.
Message has been deleted

liyi....@gmail.com

unread,
Jun 15, 2017, 9:55:40 PM6/15/17
to PLUMED users
Hi Giovanni,

Thanks.

在 2017年6月15日星期四 UTC+8下午2:37:17,Giovanni Bussi写道:
Some comments (in addition to the ones from Max):

1. You are computing distances, so the lower wall at zero is useless.

Yes. The zero is useless for distance. But the plumed need GRID_MAX and GRID_MIN should be either present or absent.
I think I can change the GRID_MIN to 0.1. 

2. With your settings (EPS=50 KAPPA=5000) the penalty for a violation of 0.1 nm is going to be 5000 * (0.1/50)**2=0.02 kj/mol. A violation of 1 nm will result in 2 kj/mol. Why did you choose this large EPS?

I try to decrease KAPPA and increase EPS to reduce the violations like
uwall: UPPER_WALLS ARG=d_b2_21,d_b3_21 AT=1.0,1.0 KAPPA=150.0,150.0 EPS=5000,5000
lwall: LOWER_WALLS ARG=d_b2_21,d_b3_21 AT=0.0,0.0 KAPPA=150.0,150.0 EPS=5000,5000
But it still got hte same error. 

3. If you set the upper wall at 1, even with a large kappa / small eps you should expect some violation. Violations will be even larger if you do METAD on that distance. So, set GRID_MAX to something like 1.5 to be sure METAD does not crash for out of grid access.

I have changed the  GRID_MAX to 1.5. It did NOT work.
 

Giovanni


> To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users...@googlegroups.com.

> To post to this group, send email to plumed...@googlegroups.com.
> Visit this group at https://groups.google.com/group/plumed-users.
> To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/406d6e8b-9f4a-45ea-a0c6-3085ae21d0a5%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "PLUMED users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users...@googlegroups.com.

To post to this group, send email to plumed...@googlegroups.com.
Visit this group at https://groups.google.com/group/plumed-users.

liyi....@gmail.com

unread,
Jun 15, 2017, 10:03:55 PM6/15/17
to PLUMED users
Hi Max,

Thanks.

在 2017年6月15日星期四 UTC+8下午2:05:20,Max写道:
Hello,

does your simulation always stop right at the beginning?

It indeed stop at the beginning.
Have you checked the initial values of your CVs?

I can not find the initial value in the md.log file.
PLUMED: Action COM
PLUMED:   with label c_b2
PLUMED:   serial associated to this virtual atom is 119401
PLUMED:   of atoms 723 730 739
PLUMED:   PBC will be ignored
PLUMED: Action COM
PLUMED:   with label c_b3
PLUMED:   serial associated to this virtual atom is 119402
PLUMED:   of atoms 1353 1361
PLUMED:   PBC will be ignored
PLUMED: Action COM
PLUMED:   with label c_b20
PLUMED:   serial associated to this virtual atom is 119403
PLUMED:   of atoms 3075 3083 3091
PLUMED:   PBC will be ignored
PLUMED: Action COM
PLUMED:   with label c_b21
PLUMED:   serial associated to this virtual atom is 119404
PLUMED:   of atoms 3150 3159 3164
PLUMED:   PBC will be ignored
PLUMED: Action DISTANCE
PLUMED:   with label d_b2_21
PLUMED:   between atoms 119401 119404
PLUMED:   using periodic boundary conditions
PLUMED: Action DISTANCE
PLUMED:   with label d_b3_21
PLUMED:   between atoms 119402 119404
PLUMED:   using periodic boundary conditions
PLUMED: Action UPPER_WALLS
PLUMED:   with label uwall
PLUMED:   with arguments d_b2_21 d_b3_21
PLUMED:   added component to this action:  uwall.bias 
PLUMED:   at 1.000000 1.000000
PLUMED:   with an offset 0.000000 0.000000
PLUMED:   with force constant 150.000000 150.000000
PLUMED:   and exponent 2.000000 2.000000
PLUMED:   rescaled 5000.000000 5000.000000
PLUMED:   added component to this action:  uwall.force2 
PLUMED: Action LOWER_WALLS
PLUMED:   with label lwall
PLUMED:   with arguments d_b2_21 d_b3_21
PLUMED:   added component to this action:  lwall.bias 
PLUMED:   at 0.000000 0.000000
PLUMED:   with an offset 0.000000 0.000000
PLUMED:   with force constant 150.000000 150.000000
PLUMED:   and exponent 2.000000 2.000000
PLUMED:   rescaled 5000.000000 5000.000000
PLUMED:   added component to this action:  lwall.force2 
PLUMED: Action METAD
PLUMED:   with label meta
PLUMED:   with arguments d_b2_21 d_b3_21
PLUMED:   added component to this action:  meta.bias 
PLUMED:   Binsize not specified, 1/5 of sigma will be be used
PLUMED:   Gaussian width  0.100000 0.100000  Gaussian height 1.000000
PLUMED:   Gaussian deposition pace 50
PLUMED:   Gaussian file HILLS
PLUMED:   Well-Tempered Bias Factor 10.000000
PLUMED:   Hills relaxation time (tau) 2.244905
PLUMED:   KbT 2.494339
PLUMED:   Grid min 0.1 0.1
PLUMED:   Grid max 1.5 1.5
PLUMED:   Grid bin 70 70
PLUMED:   Grid uses spline interpolation
PLUMED:   added component to this action:  meta.work 
PLUMED:   Bibliography [2][3]
PLUMED: Action PRINT
PLUMED:   with label @10
PLUMED:   with stride 10
PLUMED:   with arguments d_b2_21 d_b3_21 uwall.bias uwall.force2 lwall.bias lwall.force2 meta.bias meta.work
PLUMED:   on file COLVAR
PLUMED:   with format  %f
PLUMED: END FILE: plumed.dat
PLUMED: Timestep: 0.002000
PLUMED: KbT: 2.494339
PLUMED: Relevant bibliography:
PLUMED:   [1] Tribello, Bonomi, Branduardi, Camilloni, and Bussi, Comput. Phys. Commun. 185, 604 (2014)
PLUMED:   [2] Laio and Parrinello, PNAS 99, 12562 (2002)
PLUMED:   [3] Barducci, Bussi, and Parrinello, Phys. Rev. Lett. 100, 020603 (2008)
PLUMED: Please read and cite where appropriate!
PLUMED: Finished setup
Started mdrun on rank 0 Fri Jun 16 09:49:04 2017
           Step           Time         Lambda
              0        0.00000        0.00000
And the COLVAR file is empty.
 

As side notes:

1) I would remove:

> GRID_SPACING=0.01,0.01

and let plumed choose this parameter for you.
Consider that this must be smaller than SIGMA

I removed it as you suggested.

2) Be careful in choosing these parameters

> HEIGHT=0.001 SIGMA=0.001,0.001 PACE=50

In biological applications, HEIGHT is typically of the order of 1.0 kjoule/mol,
SIGMA seems too small (see previous posts and the literature about how to
choose this parameter), PACE is typically of the order of 500 (time steps).

I changed them like this
meta: METAD ARG=d_b2_21,d_b3_21 HEIGHT=1 SIGMA=0.1,0.1 PACE=50 GRID_MAX=1.5,1.5 GRID_MIN=0.1,0.1 FILE=HILLS BIASFACTOR=10 TEMP=300.0


3) you can extend GRID_MAX a bit more beyond your UPPER_WALLS,
   since it is not a hard wall.

It still got the same error after these change.

Massimiliano Bonomi

unread,
Jun 16, 2017, 3:06:51 AM6/16/17
to plumed...@googlegroups.com
Hello,

You can remove the METAD line and just calculate/print the initial values of your CVs to understand if they are in the range you expect.



Sent from my Nexus. Please excuse typos and brevity.

To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users+unsubscribe@googlegroups.com.

To post to this group, send email to plumed...@googlegroups.com.
Visit this group at https://groups.google.com/group/plumed-users.

Giovanni Bussi

unread,
Jun 16, 2017, 3:11:00 AM6/16/17
to plumed...@googlegroups.com
On Fri, Jun 16, 2017 at 3:55 AM, <liyi....@gmail.com> wrote:
Hi Giovanni,

Thanks.

在 2017年6月15日星期四 UTC+8下午2:37:17,Giovanni Bussi写道:
Some comments (in addition to the ones from Max):

1. You are computing distances, so the lower wall at zero is useless.

Yes. The zero is useless for distance. But the plumed need GRID_MAX and GRID_MIN should be either present or absent.
I think I can change the GRID_MIN to 0.1. 

Please read my email again. I am not referring to GRID_MIN but to LOWER_WALLS. GRID_MIN should be zero.
 

2. With your settings (EPS=50 KAPPA=5000) the penalty for a violation of 0.1 nm is going to be 5000 * (0.1/50)**2=0.02 kj/mol. A violation of 1 nm will result in 2 kj/mol. Why did you choose this large EPS?

I try to decrease KAPPA and increase EPS to reduce the violations like
uwall: UPPER_WALLS ARG=d_b2_21,d_b3_21 AT=1.0,1.0 KAPPA=150.0,150.0 EPS=5000,5000
lwall: LOWER_WALLS ARG=d_b2_21,d_b3_21 AT=0.0,0.0 KAPPA=150.0,150.0 EPS=5000,5000
But it still got hte same error. 

Please read the manual. If you increase EPS the restraint will be weaker, not stronger

Giovanni

 
To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users+unsubscribe@googlegroups.com.

To post to this group, send email to plumed...@googlegroups.com.
Visit this group at https://groups.google.com/group/plumed-users.

liyi....@gmail.com

unread,
Jun 16, 2017, 3:46:30 AM6/16/17
to PLUMED users
Hi Max,

I printed it
#! FIELDS time d_b2_21 d_b3_21
 0.000000 2.999888 6.129205

Both CVs are greater than the upper walls.
I change the wall setting. It still did not work.
uwall: UPPER_WALLS ARG=d_b2_21,d_b3_21 AT=1.5,1.5 KAPPA=50000,50000 EPS=0.1,0.1

Li Yi


在 2017年6月16日星期五 UTC+8下午3:06:51,Max写道:

liyi....@gmail.com

unread,
Jun 16, 2017, 3:50:56 AM6/16/17
to PLUMED users
Hi Giovanni,

My mistakes. I got it wrong.

I change the EPS to make the force enough large?
uwall: UPPER_WALLS ARG=d_b2_21,d_b3_21 AT=1.5,1.5 KAPPA=50000,50000 EPS=0.1,0.1

But it sitll did not run.

Li Yi

在 2017年6月16日星期五 UTC+8下午3:11:00,Giovanni Bussi写道:


Massimiliano Bonomi

unread,
Jun 16, 2017, 4:11:42 AM6/16/17
to plumed...@googlegroups.com
Your starting configuration is clear way beyond the values you set for UPPER WALLS and GRID MAX. You should either increase both or choose another starting configuration 

Max


Sent from my Nexus. Please excuse typos and brevity.
To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users+unsubscribe@googlegroups.com.

To post to this group, send email to plumed...@googlegroups.com.
Visit this group at https://groups.google.com/group/plumed-users.

liyi....@gmail.com

unread,
Jun 19, 2017, 2:34:06 AM6/19/17
to PLUMED users
Hi Max,
Thanks.

I changed the configuration and checked the atoms again.
The plumed found the right values at this time.

#! FIELDS time d_b2_21 d_b3_21 meta.bias meta.work
 0.000000 1.103512 0.719011 0.000000 0.000000

According to distance in static structures, I want to increase b2_21 and decrease b3_21.
# state   b2_21 b3_21
# apo    0.842 0.473
# liganded 0.490 0.973

But this two CVs increase during the simulation.

Li Yi


在 2017年6月16日星期五 UTC+8下午4:11:42,Max写道:

Massimiliano Bonomi

unread,
Jun 19, 2017, 2:41:54 AM6/19/17
to plumed...@googlegroups.com
Hello,

you cannot really control with metadynamics in which direction your CVs will go.
Under the effect of the metadynamics bias potential, sampling will be accelerated
along the CVs. So eventually you will visit your apo and holo states.

If you haven’t done it yet, I suggest to play with the metadynamics tutorials on our website.
Particularly important is the new tutorial about error analysis and assessing convergence:

http://plumed.github.io/doc-master/user-doc/html/trieste-4.html

I would make sure to be familiar with the concepts explained in the tutorial before
proceeding with more complex cases, like yours.

Max
> To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/9f752bd5-4c2a-467a-8a08-91cd4431c2c9%40googlegroups.com.

liyi....@gmail.com

unread,
Jun 19, 2017, 2:47:24 AM6/19/17
to PLUMED users
Hi Max,

Ok, Thanks.
I will try it as your suggestion.

Li Yi

在 2017年6月19日星期一 UTC+8下午2:41:54,Max写道:
Reply all
Reply to author
Forward
0 new messages