-Infinity values for free energy

604 views
Skip to first unread message

osmar

unread,
Apr 7, 2010, 7:12:57 AM4/7/10
to PLUMED users
Hi users,
I performed a simulation (plumed-gromacs) and found the following
values in the fes.dat file, after used the command: ./sum_hills.x -
file HILLS -out fes.dat -ndim 1 -ndw 1
...
2.4356800970 -4.0681835016
2.5794811071 -183.4157691710
2.7232821172 -Infinity
2.8670831274 -Infinity
3.0108841375 -Infinity
3.1546851476 -Infinity
3.2984861577 -Infinity
3.4422871679 -Infinity
3.5860881780 -246.4541845266
3.7298891881 -167.2058965406
...
How can I solve this?

Thanks
Osmair
Brazil

Massimiliano Bonomi

unread,
Apr 7, 2010, 7:29:09 AM4/7/10
to plumed...@googlegroups.com
Dear Osmair,

I'm afraid the sum_hills code has a problem when a large amount of bias is added in one point.
How many Gaussians have you deposited so far?
Which height did you use for the Gaussian?

Regards,
Max

> --
> You received this message because you are subscribed to the Google Groups "PLUMED users" group.
> To post to this group, send email to plumed...@googlegroups.com.
> To unsubscribe from this group, send email to plumed-users...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/plumed-users?hl=en.
>

Max

unread,
Apr 7, 2010, 8:13:15 AM4/7/10
to PLUMED users
PS: A possible workaround could be to use a fake high temperature with
sum_hills:

./sum_hills.x -file HILLS -out fes.dat -ndim 1 -ndw 1 -kt 1000.0

Be careful, you cannot use a fake temperature when you are integrating
out one of your metadynamics CVs.
This is allowed:

1CV META: ./sum_hills.x -file HILLS -out fes.dat -ndim 1 -ndw 1 -
kt 1000.0
2CV META: ./sum_hills.x -file HILLS -out fes.dat -ndim 2 -ndw 1 2 -kt
1000.0

This IS NOT and will give you a wrong one-dimensional free energy:

./sum_hills.x -file HILLS -out fes.dat -ndim 2 -ndw 1 -kt 1000.0

Regards,
Max

On Apr 7, 1:29 pm, Massimiliano Bonomi <massimiliano.bon...@gmail.com>
wrote:

YAGYA CHAUDHARY

unread,
Jan 20, 2021, 7:36:40 AM1/20/21
to PLUMED users
I am performing PTMETAD-WTE in two-dimension on a protein. I have received infinite value and negative value for the projection while getting free energy wrt to CV1. I have tried using high temperature in --kt but still did not work for me? How to solve this?
I have used this command -> plumed sum_hills --hills HILLS_PTMetaDWTE.0,HILLS_PTMetaDWTE.1,HILLS_PTMetaDWTE.2,HILLS_PTMetaDWTE.3,HILLS_PTMetaDWTE.4,HILLS_PTMetaDWTE.5,HILLS_PTMetaDWTE.6,HILLS_PTMetaDWTE.7,HILLS_PTMetaDWTE.8,HILLS_PTMetaDWTE.9,HILLS_PTMetaDWTE.10,HILLS_PTMetaDWTE.11 --idw d1 --sigma 0.03 --outfile fes1.dat --kt 1000.0


Also, I am getting negative value for CVs while summing up the hills. 

I have used 12 replicas from 290K-620K.
For energy space, gaussian width was 800kj/mol
while restarting, gaussian width was 0.03 for cv1 and 1.0 for cv2.

TIA
Yagya

Massimiliano Bonomi

unread,
Jan 20, 2021, 7:39:28 AM1/20/21
to plumed...@googlegroups.com
Hi!

First use just one HILLS file at a time. Second, you should set the right temperature corresponding to the replica you are using. Finally, negative values of free energy and even inifnity (if you have never visited that region of the CV space) are normal.

Sent from my iPhone

On 20 Jan 2021, at 13:36, YAGYA CHAUDHARY <yagya1...@gmail.com> wrote:


To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/8bec8eb3-fcb1-4317-883c-89068765413cn%40googlegroups.com.

YAGYA CHAUDHARY

unread,
Jan 20, 2021, 7:44:50 AM1/20/21
to PLUMED users
Hi Max,

Thanks for your reply.

Getting infinite and negative value for projection is also fine? And why CVs value(like no. of contacts) is going negative? What could be the reason?

Regards
Yagya

Massimiliano Bonomi

unread,
Jan 20, 2021, 8:07:05 AM1/20/21
to plumed...@googlegroups.com
Negative value for a cv which is defined positive might happen depending on the value of the metadynamics Gaussian width. Infinity values are weird!

Max

Sent from my iPhone

On 20 Jan 2021, at 13:39, Massimiliano Bonomi <massimili...@gmail.com> wrote:

Hi!

YAGYA CHAUDHARY

unread,
Jan 20, 2021, 8:29:33 AM1/20/21
to PLUMED users

I have used 12 replicas from 290K-620K.
For energy space, gaussian width was 800kj/mol
while restarting, gaussian width was 0.03 for cv1 and 1.0 for cv2.

The projection is in negative with a very high value.

When I am summing up the hills for the 2D free energy values, CV is coming negative. What does it mean?

Can we restrict the values for CV?

YAGYA

Massimiliano Bonomi

unread,
Jan 20, 2021, 8:32:04 AM1/20/21
to plumed...@googlegroups.com
Can you please paste the sumhills command you used for the 1D and 2D free energies? 

Max

Sent from my iPhone

On 20 Jan 2021, at 13:39, Massimiliano Bonomi <massimili...@gmail.com> wrote:

Hi!

YAGYA CHAUDHARY

unread,
Jan 20, 2021, 8:36:00 AM1/20/21
to PLUMED users
for 1D  -> plumed sum_hills  --hills HILLS_PTMetaDWTE.11 --idw cc --sigma 1.0 --outfile 11_1.dat --kt 620.0
for 2D -> plumed sum_hills  --hills HILLS_PTMetaDWTE.11 --outfile 11.dat 

Also how to combine all the hills data to get a complete free energy profile for a 2D plot?

Massimiliano Bonomi

unread,
Jan 20, 2021, 8:37:56 AM1/20/21
to plumed...@googlegroups.com

> for 1D -> plumed sum_hills --hills HILLS_PTMetaDWTE.11 --idw cc --sigma 1.0 --outfile 11_1.dat --kt 620.0

Ok for replica at 620K, no need to specify sigma here. If you want to impose limits on the output grid for the free energy, just add:

--min - the lower bounds for the grid
--max - the upper bounds for the grid

> for 2D -> plumed sum_hills --hills HILLS_PTMetaDWTE.11 --outfile 11.dat

Ok, here also you can specify limits:

--min - the lower bounds for the grid
--max - the upper bounds for the grid

Max


Massimiliano Bonomi

unread,
Jan 20, 2021, 8:42:29 AM1/20/21
to plumed...@googlegroups.com
Sorry, I made a mistake. The temperature should be specified in energy unit (it is actually kbT). You can convert 620 K to energy unit with:

plumed kt -- temp 620

Sent from my iPhone

> On 20 Jan 2021, at 14:37, Massimiliano Bonomi <massimili...@gmail.com> wrote:
>
> 

YAGYA CHAUDHARY

unread,
Jan 20, 2021, 9:14:48 AM1/20/21
to PLUMED users
What does the negative value of projection signify? Is it ok to get them?
So these are the individual hills data. How should I combine all the hills data into one?

And for the 2D plot, cv values are still negative. Any suggestions?

Massimiliano Bonomi

unread,
Jan 20, 2021, 9:17:28 AM1/20/21
to plumed...@googlegroups.com
As I said, negative cvs values are ok. I explained you the reason: finite width of the Gaussian!

You don’t combine all the HILLS file together, these are at different temperatures.
> --
> 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 view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/85f63119-975f-45f6-a691-bcfe8e16fc42n%40googlegroups.com.

YAGYA CHAUDHARY

unread,
Jan 20, 2021, 9:21:16 AM1/20/21
to PLUMED users
What does this command do-over using just one hills file?

plumed sum_hills --hills PATHTOMYHILLSFILE1,PATHTOMYHILLSFILE2,PATHTOMYHILLSFILE3

Massimiliano Bonomi

unread,
Jan 20, 2021, 9:23:08 AM1/20/21
to plumed...@googlegroups.com
As I said already, you must use only one HILLS file with plumed sum_hills.
Please have a *careful* look at the previous emails!

Max
> To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/13c94306-6534-457d-907a-0c11d2c771dan%40googlegroups.com.

YAGYA CHAUDHARY

unread,
Jan 20, 2021, 9:27:28 AM1/20/21
to PLUMED users
Thank you very much for your responses and help.

YAGYA CHAUDHARY

unread,
Jan 22, 2021, 2:15:12 AM1/22/21
to PLUMED users
Hi,

I am running PTMETAD-WTE in 2D for a protein. The simulation is 200 ns long. How should I check in between whether my system is going in the right direction or not?  As it will be hard if we get to know the values aren't coming right after a whole run.


I have used 12 replicas from 290K-620K.
For energy space, gaussian width was 800kj/mol
while restarting, gaussian width was 0.03 for cv1 and 1.0 for cv2.




Regards
Yagya

Massimiliano Bonomi

unread,
Jan 22, 2021, 4:11:07 AM1/22/21
to plumed...@googlegroups.com
Hi,

I don’t understand what you mean by this:

> the values aren't coming right after a whole run.

Max

YAGYA CHAUDHARY

unread,
Jan 22, 2021, 5:52:30 AM1/22/21
to PLUMED users
What I meant: Is there any way to check the free energy values when we on the halfway to complete a 200ns run in PTMETAD-WTE?  Waiting for completing 200ns and getting the incorrect values will be time-consuming?

Also, do we consider free energy values generated from every HILLS file?

Massimiliano Bonomi

unread,
Jan 22, 2021, 7:49:08 AM1/22/21
to plumed...@googlegroups.com
If you mean check the estimated free energy at intermediate times during the simulation, check the sumhills manual for the corresponding option

Best
Max

Sent from my iPhone

On 22 Jan 2021, at 11:52, YAGYA CHAUDHARY <yagya1...@gmail.com> wrote:


--
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.

YAGYA CHAUDHARY

unread,
Jan 22, 2021, 8:29:48 AM1/22/21
to PLUMED users
Thanks for your response.

1.Is there any way to get completely free energy profile in 2D along CV1 and CV2 for PTMETAD-WTE.
2. What does the value of free energy in different HILLS file signify?
3. How to analyse different fes files generates after summing HILLS file?

Massimiliano Bonomi

unread,
Jan 22, 2021, 8:41:37 AM1/22/21
to plumed...@googlegroups.com
Hi!

> 1.Is there any way to get completely free energy profile in 2D along CV1 and CV2 for PTMETAD-WTE.

Yes, you have to use a so-called “re-weighting” procedure, as explained in the original PBMETAD paper,
which I invite you to read carefully. You can adapt the procedure illustrated here:

https://www.plumed.org/doc-v2.7/user-doc/html/master-_i_s_d_d-2.html

in Exercise 3 for the case of PBMETAD

> 2. What does the value of free energy in different HILLS file signify?

The 1D free-energy as a function of the corresponding CV

> 3. How to analyse different fes files generates after summing HILLS file?

What do you mean by analyze? You should be familiar with the concept of free-energy
and what it means before proceeding with your simulations.

Max


>
>
> On Friday, January 22, 2021 at 6:19:08 PM UTC+5:30 Max wrote:
> If you mean check the estimated free energy at intermediate times during the simulation, check the sumhills manual for the corresponding option
>
> Best
> Max
>
> Sent from my iPhone
>
>> On 22 Jan 2021, at 11:52, YAGYA CHAUDHARY <yagya1...@gmail.com> wrote:
>>
>> 
>> What I meant: Is there any way to check the free energy values when we on the halfway to complete a 200ns run in PTMETAD-WTE? Waiting for completing 200ns and getting the incorrect values will be time-consuming?
>>
>> Also, do we consider free energy values generated from every HILLS file?
>>
>> On Friday, January 22, 2021 at 2:41:07 PM UTC+5:30 Max wrote:
>> Hi,
>>
>> I don’t understand what you mean by this:
>>
>> > the values aren't coming right after a whole run.
>>
>> 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 view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/813636de-559f-4b14-a3f4-933722a78144n%40googlegroups.com.
>
> --
> 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 view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/1c4e504a-0306-4ac9-ac84-9fadf44be815n%40googlegroups.com.

YAGYA CHAUDHARY

unread,
Jan 22, 2021, 8:48:44 AM1/22/21
to PLUMED users

Sorry for putting my question badly.
What I meant from my third ques is:  After summing up all the HILLS files from all the replicas, the output of every summed HILLS file for every replica is different. So what does the different output generated signify in PTMETAD-WTE?

Massimiliano Bonomi

unread,
Jan 22, 2021, 8:53:34 AM1/22/21
to plumed...@googlegroups.com
Well, if you are asking this question you are not understanding what the replicas are.
So, let me give you a few suggestions:

- before using a method, read carefully the original paper and the main references cited within.
For example, in the case of PBMETAD-WTE, you should *at least* read (and fully understand) these papers before proceeding:
- the original Metadynamics paper
- the original PBMETAD paper
- the original WTE paper
- a paper/review describing Replica Exchange / Parallel Tempering
- If there are general concepts you don’t understand (like what is a free-energy), go back to a basic statistical mechanics textbook
- Before running your simulations, play with simple examples on the PLUMED tutorials page and refer to the manual when some
commands are not clear or you need advanced options

Max
> To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/bac79b95-d2dc-499f-b8ea-8dabb6d0eae7n%40googlegroups.com.

YAGYA CHAUDHARY

unread,
Jan 22, 2021, 8:58:23 AM1/22/21
to PLUMED users
I know pretty well about biases and the method I am using. Actually, my doubt is with the postprocessing of the HILLS files in PTMETAD. 

Thanks

Massimiliano Bonomi

unread,
Jan 22, 2021, 9:01:09 AM1/22/21
to plumed...@googlegroups.com
Well, if you cannot answer yourself to this question:

> So what does the different output generated signify in PTMETAD-WTE?

you are not understanding what you are calculating in each replica and what/why the replicas are different.
Or I have not understood your question :-)

Max



YAGYA CHAUDHARY

unread,
Jan 22, 2021, 9:21:32 AM1/22/21
to PLUMED users

I understand about replicas and what and why they are different. I got my doubt cleared from the paper linked in this tutorial https://www.plumed.org/doc-v2.5/user-doc/html/belfast-7.html

Thanks for your response.

YAGYA CHAUDHARY

unread,
Jan 22, 2021, 12:01:46 PM1/22/21
to PLUMED users
Can mintozero function (used to align the profiles by setting the global minimum to zero) change the projection values from negative to positive also the magnitude has also changed? Why? 

Massimiliano Bonomi

unread,
Jan 22, 2021, 12:02:52 PM1/22/21
to plumed...@googlegroups.com
Minzero can only shift the values of the free energy to set the minimum value to zero

Max

Sent from my iPhone

On 22 Jan 2021, at 18:01, YAGYA CHAUDHARY <yagya1...@gmail.com> wrote:

Can mintozero function (used to align the profiles by setting the global minimum to zero) change the projection values from negative to positive also the magnitude has also changed? Why? 
--
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.

YAGYA CHAUDHARY

unread,
Jan 22, 2021, 12:07:30 PM1/22/21
to PLUMED users

THis is the output when I used --mintozero
 #! FIELDS d1 projection
#! SET min_d1 -0.104785
#! SET max_d1 6.54206
#! SET nbins_d1  628
#! SET periodic_d1 false
   -0.104785000  153.878441342
   -0.094183971  153.869577961
   -0.083582943  153.840905572
   -0.072981914  153.757895709
   -0.062380885  153.541814665
   -0.051779856  153.028733930
   -0.041178828  151.900646045
   -0.030577799  149.605529815
   -0.019976770  145.440376821
   -0.009375742  139.049811462
    0.001225287  130.848881668
    0.011826316  121.768281499
    0.022427344  112.874641550
    0.033028373  105.194621439
    0.043629402   99.443724246
    0.054230431   95.659872340
    0.064831459   93.257431148
    0.075432488   91.497962748
    0.086033517   89.852227115
    0.096634545   88.089107643
    0.107235574   86.070854969
    0.117836603   83.627724542
    0.128437632   80.958942062
    0.139038660   78.641962451



This is the output without--mintozero
#! FIELDS d1 projection
#! SET min_d1 -0.104785
#! SET max_d1 6.54206
#! SET nbins_d1  628
#! SET periodic_d1 false
   -0.104785000  -28.110362934
   -0.094183971  -28.119226315
   -0.083582943  -28.147898704
   -0.072981914  -28.230908567
   -0.062380885  -28.446989611
   -0.051779856  -28.960070346
   -0.041178828  -30.088158231
   -0.030577799  -32.383274462
   -0.019976770  -36.548427455
   -0.009375742  -42.938992814
    0.001225287  -51.139922609
    0.011826316  -60.220522777
    0.022427344  -69.114162727
    0.033028373  -76.794182837
    0.043629402  -82.545080031
    0.054230431  -86.328931936
    0.064831459  -88.731373129
    0.075432488  -90.490841528
    0.086033517  -92.136577161
    0.096634545  -93.899696633
    0.107235574  -95.917949307
    0.117836603  -98.361079734
    0.128437632 -101.029862214
    0.139038660 -103.346841825
Is this possible?

Massimiliano Bonomi

unread,
Jan 22, 2021, 12:10:27 PM1/22/21
to plumed...@googlegroups.com
Please, before posting, try to do your homework. Are the two profiles shifted by a constant value? If not, there is something wrong. Do not expect other people to work for you by posting here the raw data.

Max

Sent from my iPhone

Reply all
Reply to author
Forward
0 new messages