Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP
Hi Mark and Justin, I did investigate on the rvdw=rlist issue. But, I found even if I make rvdw rlist and still keep time step 0.002, the average temperature goes to 303 K. Only reducing the time step does help ( even if rvdw=rlist) to get average temperature to 300K. Sanku From: Mark Abraham mark.abra...@anu.edu.au To: Discussion list for GROMACS users gmx-users@gromacs.org Sent: Tue, August 2, 2011 11:46:21 PM Subject: Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP On 3/08/2011 2:23 PM, Sanku M wrote: Finally, I achieved the desired 300K average temperature using sd integrator . I had to reduce the time step from 0.002 ps to 0.001 ps and then running the NVT could produce 300K as average temperature. Still Not sure why md integrator still could 300K average temperature using 0.002 ps as time-step. So your integration step was too large to conserve energy. There have been two suggestions made about why that might be true. Have you investigated them? Mark From: Justin A. Lemkul jalem...@vt.edu To: Discussion list for GROMACS users gmx-users@gromacs.org Sent: Tue, August 2, 2011 10:16:17 PM Subject: Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP Sanku M wrote: Hi Justin, I did calculate the average temperature using last 2 ns of the 8ns data. It is still same 303K. Going back to the original .mdp file, I can see that you're using some incorrect settings. You've set rlist=rvdw even though you're using a shifted potential for van der Waals interactions. Surely grompp should have warned you that this would lead to poor energy conservation. Your solvent is probably heating up as a result and leading to the incorrect temperature. -Justin *From:* Justin A. Lemkul jalem...@vt.edu *To:* Discussion list for GROMACS users gmx-users@gromacs.org *Sent:* Tue, August 2, 2011 9:55:40 PM *Subject:* Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP Sanku M wrote: OKK, But, before, I have tried using md as integrator where I start the initial velocity as 280K ( which makes sense as the minimized configuration is essentially at low temperature) and have used Nose-hoover at 300K and I never had problem in getting the desired 300K after running a short NVT simulation as equilabration. The problem arises when sd integrator is used. In fact, I just tried to generate the initial velocity at 300K along with sd integrator. I got the 303 K as average temperature again. The same simulation using md integrator always gives rise to 300K as average temperature( when used in conjunction with Nose-Hoover ) , even if I generate velocity at 280 K or 300K . I guess that the initial velocity-generation temperature should not be an issue here. The sd integrator in contrast to md integrator may be the problem. I have never had such a problem when using sd. I can always achieve the target temperature within 0.5 K, if not exactly, so I do not believe there is an inherent problem with sd. Your first post indicated that your statistics were collected over the entire timeframe, which will not be correct for reasons I have already described. The initial frames will represent unequilibrated data that may skew the averages. What if you analyze only the last half of the time? The last 1 ns? -Justin -- Justin A. Lemkul Ph.D. Candidate ICTAS Doctoral Scholar MILES-IGERT Trainee Department of Biochemistry Virginia Tech Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin -- gmx-users mailing listgmx-users@gromacs.org mailto:gmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org mailto:gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists -- Justin A. Lemkul Ph.D. Candidate ICTAS Doctoral Scholar MILES-IGERT Trainee Department of Biochemistry Virginia Tech Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080
Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP
On 4/08/2011 1:51 AM, Sanku M wrote: Hi Mark and Justin, I did investigate on the rvdw=rlist issue. But, I found even if I make rvdw rlist and still keep time step 0.002, the average temperature goes to 303 K. Only reducing the time step does help ( even if rvdw=rlist) to get average temperature to 300K. ... and the tau_t issue I mentioned? Mark Sanku *From:* Mark Abraham mark.abra...@anu.edu.au *To:* Discussion list for GROMACS users gmx-users@gromacs.org *Sent:* Tue, August 2, 2011 11:46:21 PM *Subject:* Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP On 3/08/2011 2:23 PM, Sanku M wrote: Finally, I achieved the desired 300K average temperature using sd integrator . I had to reduce the time step from 0.002 ps to 0.001 ps and then running the NVT could produce 300K as average temperature. Still Not sure why md integrator still could 300K average temperature using 0.002 ps as time-step. So your integration step was too large to conserve energy. There have been two suggestions made about why that might be true. Have you investigated them? Mark *From:* Justin A. Lemkul jalem...@vt.edu *To:* Discussion list for GROMACS users gmx-users@gromacs.org *Sent:* Tue, August 2, 2011 10:16:17 PM *Subject:* Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP Sanku M wrote: Hi Justin, I did calculate the average temperature using last 2 ns of the 8ns data. It is still same 303K. Going back to the original .mdp file, I can see that you're using some incorrect settings. You've set rlist=rvdw even though you're using a shifted potential for van der Waals interactions. Surely grompp should have warned you that this would lead to poor energy conservation. Your solvent is probably heating up as a result and leading to the incorrect temperature. -Justin *From:* Justin A. Lemkul jalem...@vt.edu mailto:jalem...@vt.edu *To:* Discussion list for GROMACS users gmx-users@gromacs.org mailto:gmx-users@gromacs.org *Sent:* Tue, August 2, 2011 9:55:40 PM *Subject:* Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP Sanku M wrote: OKK, But, before, I have tried using md as integrator where I start the initial velocity as 280K ( which makes sense as the minimized configuration is essentially at low temperature) and have used Nose-hoover at 300K and I never had problem in getting the desired 300K after running a short NVT simulation as equilabration. The problem arises when sd integrator is used. In fact, I just tried to generate the initial velocity at 300K along with sd integrator. I got the 303 K as average temperature again. The same simulation using md integrator always gives rise to 300K as average temperature( when used in conjunction with Nose-Hoover ) , even if I generate velocity at 280 K or 300K . I guess that the initial velocity-generation temperature should not be an issue here. The sd integrator in contrast to md integrator may be the problem. I have never had such a problem when using sd. I can always achieve the target temperature within 0.5 K, if not exactly, so I do not believe there is an inherent problem with sd. Your first post indicated that your statistics were collected over the entire timeframe, which will not be correct for reasons I have already described. The initial frames will represent unequilibrated data that may skew the averages. What if you analyze only the last half of the time? The last 1 ns? -Justin -- Justin A. Lemkul Ph.D. Candidate ICTAS Doctoral Scholar MILES-IGERT Trainee Department of Biochemistry Virginia Tech Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin -- gmx-users mailing list gmx-users@gromacs.org mailto:gmx-users@gromacs.org mailto:gmx-users@gromacs.org mailto:gmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org mailto:gmx-users-requ...@gromacs.org mailto:gmx-users-requ...@gromacs.org mailto:gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists -- Justin A. Lemkul Ph.D. Candidate ICTAS Doctoral Scholar MILES-IGERT Trainee Department of Biochemistry Virginia Tech Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin --
Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP
Yes. I did check the tau_t issue. I used tau_t=2ps as recommended in the manual. Again here, the simulation with 0.002 ps step gives rise to 303K. But, from now on, as per your( and Justin's) recommendation, I am going to use the tau_t=2ps and rvdwrlist in conjunction with 0.001 ps time step when using sd integrator from my production/equilibration run for all lambda. From: Mark Abraham mark.abra...@anu.edu.au To: Discussion list for GROMACS users gmx-users@gromacs.org Sent: Wed, August 3, 2011 11:00:30 AM Subject: Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP On 4/08/2011 1:51 AM, Sanku M wrote: Hi Mark and Justin, I did investigate on the rvdw=rlist issue. But, I found even if I make rvdw rlist and still keep time step 0.002, the average temperature goes to 303 K. Only reducing the time step does help ( even if rvdw=rlist) to get average temperature to 300K. ... and the tau_t issue I mentioned? Mark-- gmx-users mailing listgmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP
Sanku M wrote: Hi, I am trying to calculate the solvation free energy using thermodynamic integration(TI) method . I am using gromacs 4.0.7 . But, I am having a problem in getting accurate average temperature . The following is what I did: I found that when doing TI, grompp recommends using 'sd' ( stochastic dynamics ) as integrator ( provides an warning that for decoupled system, it is better to use sd in stead of md ). Also I went through Justin Lemkul's tutorial which also recommends using 'sd' for TI method. So, I tried to use sd integrator. The tutorial uses BAR for calculation free energy, not TI. Though the protocols are similar, there is a difference, FYI. I wanted to run NVT simulation at 300 K. I also found from manual and also from Justin's website on FEP, sd integrator implicitly controls temperature and so there is no need to specify a thermostat. So, I did not specify any thermostat ( I wrote tcoupl = No ). But, unfortunately, I found that after a long simulation , the average temperature actually goes to 303 K in stead of 300 K( the desired temperature). This happened for all Lambda values where the average temperature turned out to be 303 K. The .mdp file is pasted at the end of the email and I felt I am using reasonable cutoffs and PME as electrostatics. Here is the output from g_energy command to get the temperature. Statistics over 401 steps [ 0. thru 8000.0005 ps ], 1 data sets All averages are exact over 401 steps Energy Average RMSD Fluct. Drift Tot-Drift --- Temperature 303.3334.780844.78082 5.46937e-06 0.043755 Heat Capacity Cv: 12.4764 J/mol K (factor = 0.00024841) I tried three further tests: a) I removed tcoupl = No option . But, it is giving same 303K as average temperature when using sd. a) I tried to specify the Nose-Hoover thermostat along with sd. But still, I found that when using sd, the average temperature is going to 303 K in stead of 300 K. From the manual description of the sd integrator: The parameter tcoupl is ignored. This explains (a) and (b). c) Finally, I tried to overlook grompp warning and went ahead and used 'md' along with Nose-Hoover thermostat. This time I found the right average temperature 300 K is being achieved.B But, I understand that for a decoupled system like here, I need to use a method like stochastic dynamics. But , I was wondering why it is reaching 303 K in stead of 300 K when using sd but 300K is achieved when using md. I looked at the mailing list where people had issues with sd and temperature control. But, I could not find a good solution. So, any help on the right protocol will be really appreciated. What type of equilibration did you do prior to the data collection? If your system isn't sampling the desired ensemble, then you shouldn't proceed. -Justin Here is my .mdp file. ; RUN CONTROL PARAMETERS integrator = sd ; Start time and timestep in ps tinit= 0.0 dt = 0.002 nsteps = 400 ; For exact run continuation or redoing part of a run ; Part index is updated automatically on checkpointing (keeps files separate) simulation_part = 1 init_step= 0 ; mode for center of mass motion removal comm-mode= Linear ; number of steps for center of mass motion removal nstcomm = 1 ; group(s) for center of mass motion removal comm-grps= ; LANGEVIN DYNAMICS OPTIONS ; Friction coefficient (amu/ps) and random seed bd-fric = 0 ld-seed = 1993 ; OUTPUT CONTROL OPTIONS ; OUTPUT CONTROL OPTIONS ; Output frequency for coords (x), velocities (v) and forces (f) nstxout = 1 nstvout = 1 nstfout = 0 ; Output frequency for energies to log file and energy file nstlog = 1000 nstenergy= 10 ; Output frequency and precision for xtc file nstxtcout= 1000 xtc-precision= 1000 ; This selects the subset of atoms for the xtc file. You can ; select multiple groups. By default all atoms will be written. xtc-grps = ; Selection of energy groups energygrps = ; NEIGHBORSEARCHING PARAMETERS ; nblist update frequency nstlist = 10 ; ns algorithm (simple or grid) ns_type = grid ; Periodic boundary conditions: xyz, no, xy pbc = xyz periodic_molecules = no ; nblist cut-off rlist= 1.4 ; OPTIONS FOR ELECTROSTATICS AND VDW ; Method for doing electrostatics coulombtype = pme rcoulomb-switch = 0 rcoulomb = 1.4 ; Relative dielectric constant for the medium and the reaction field epsilon_r= 1
Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP
Hi Justin, I first performed two minimization using steep and l-bfgs as the method on the solvated system for each lambda. Then I ran a 500ps NVT simulation on minimized system ( for each lambda) where essentially same .mdp file was used except the following change: gen-vel = yes gen-temp= 280 gen-seed= -1 So, essentially, I heated the minimized system up from 280 K with ref_t 300K using sd integrator. But, here also I found the average temperature goes to 303 K( instead of 300 K) . One thing I don't understand that why using md as integrator gives the desired average temperature( when using Nose-Hoover thermostat) while the sd integrator does not. Any further suggestion will be appreciated. Jagannath From: Justin A. Lemkul jalem...@vt.edu To: Discussion list for GROMACS users gmx-users@gromacs.org Sent: Tue, August 2, 2011 8:08:20 PM Subject: Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP Sanku M wrote: Hi, I am trying to calculate the solvation free energy using thermodynamic integration(TI) method . I am using gromacs 4.0.7 . But, I am having a problem in getting accurate average temperature . The following is what I did: I found that when doing TI, grompp recommends using 'sd' ( stochastic dynamics ) as integrator ( provides an warning that for decoupled system, it is better to use sd in stead of md ). Also I went through Justin Lemkul's tutorial which also recommends using 'sd' for TI method. So, I tried to use sd integrator. The tutorial uses BAR for calculation free energy, not TI. Though the protocols are similar, there is a difference, FYI. I wanted to run NVT simulation at 300 K. I also found from manual and also from Justin's website on FEP, sd integrator implicitly controls temperature and so there is no need to specify a thermostat. So, I did not specify any thermostat ( I wrote tcoupl = No ). But, unfortunately, I found that after a long simulation , the average temperature actually goes to 303 K in stead of 300 K( the desired temperature). This happened for all Lambda values where the average temperature turned out to be 303 K. The .mdp file is pasted at the end of the email and I felt I am using reasonable cutoffs and PME as electrostatics. Here is the output from g_energy command to get the temperature. Statistics over 401 steps [ 0. thru 8000.0005 ps ], 1 data sets All averages are exact over 401 steps Energy Average RMSD Fluct. Drift Tot-Drift --- Temperature 303.3334.780844.78082 5.46937e-06 0.043755 Heat Capacity Cv: 12.4764 J/mol K (factor = 0.00024841) I tried three further tests: a) I removed tcoupl = No option . But, it is giving same 303K as average temperature when using sd. a) I tried to specify the Nose-Hoover thermostat along with sd. But still, I found that when using sd, the average temperature is going to 303 K in stead of 300 K. From the manual description of the sd integrator: The parameter tcoupl is ignored. This explains (a) and (b). c) Finally, I tried to overlook grompp warning and went ahead and used 'md' along with Nose-Hoover thermostat. This time I found the right average temperature 300 K is being achieved.B But, I understand that for a decoupled system like here, I need to use a method like stochastic dynamics. But , I was wondering why it is reaching 303 K in stead of 300 K when using sd but 300K is achieved when using md. I looked at the mailing list where people had issues with sd and temperature control. But, I could not find a good solution. So, any help on the right protocol will be really appreciated. What type of equilibration did you do prior to the data collection? If your system isn't sampling the desired ensemble, then you shouldn't proceed. -Justin Here is my .mdp file. ; RUN CONTROL PARAMETERS integrator = sd ; Start time and timestep in ps tinit= 0.0 dt = 0.002 nsteps = 400 ; For exact run continuation or redoing part of a run ; Part index is updated automatically on checkpointing (keeps files separate) simulation_part = 1 init_step= 0 ; mode for center of mass motion removal comm-mode= Linear ; number of steps for center of mass motion removal nstcomm = 1 ; group(s) for center of mass motion removal comm-grps= ; LANGEVIN DYNAMICS OPTIONS ; Friction coefficient (amu/ps) and random seed bd-fric = 0 ld-seed = 1993 ; OUTPUT CONTROL OPTIONS ; OUTPUT CONTROL OPTIONS ; Output frequency for coords (x), velocities (v) and forces (f) nstxout =
Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP
Sanku M wrote: Hi Justin, I first performed two minimization using steep and l-bfgs as the method on the solvated system for each lambda. Then I ran a 500ps NVT simulation on minimized system ( for each lambda) where essentially same .mdp file was used except the following change: gen-vel = yes gen-temp= 280 gen-seed= -1 So, essentially, I heated the minimized system up from 280 K with ref_t 300K using sd integrator. But, here also I found the average temperature goes to 303 K( instead of 300 K) . Unless you did simulated annealing in between, you didn't heat the system, you suddenly jolted the temperature, which is probably not stable. Maybe the integrator/Langevin thermostat does not react well to these conditions. Equilibrate under the conditions you wish to collect data. Your data collection uses NPT, so running NVT at a different temperature and suddenly changing the temperature and introducing pressure coupling is not an appropriate procedure. If nothing else, your data will be unreliable, even if your temperature was correct. One thing I don't understand that why using md as integrator gives the desired average temperature( when using Nose-Hoover thermostat) while the sd integrator does not. I can only hazard a guess that for some reason Nose-Hoover can arrive at the correct temperature more quickly, but that seems uncharacteristic for it given the nature of its fluctuations. Perhaps someone with more knowledge of the algorithms and code can comment. But until you've established a properly equilibrated system under appropriate conditions, it's not worth investigating these differences much further. -Justin Any further suggestion will be appreciated. Jagannath *From:* Justin A. Lemkul jalem...@vt.edu *To:* Discussion list for GROMACS users gmx-users@gromacs.org *Sent:* Tue, August 2, 2011 8:08:20 PM *Subject:* Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP Sanku M wrote: Hi, I am trying to calculate the solvation free energy using thermodynamic integration(TI) method . I am using gromacs 4.0.7 . But, I am having a problem in getting accurate average temperature . The following is what I did: I found that when doing TI, grompp recommends using 'sd' ( stochastic dynamics ) as integrator ( provides an warning that for decoupled system, it is better to use sd in stead of md ). Also I went through Justin Lemkul's tutorial which also recommends using 'sd' for TI method. So, I tried to use sd integrator. The tutorial uses BAR for calculation free energy, not TI. Though the protocols are similar, there is a difference, FYI. I wanted to run NVT simulation at 300 K. I also found from manual and also from Justin's website on FEP, sd integrator implicitly controls temperature and so there is no need to specify a thermostat. So, I did not specify any thermostat ( I wrote tcoupl = No ). But, unfortunately, I found that after a long simulation , the average temperature actually goes to 303 K in stead of 300 K( the desired temperature). This happened for all Lambda values where the average temperature turned out to be 303 K. The .mdp file is pasted at the end of the email and I felt I am using reasonable cutoffs and PME as electrostatics. Here is the output from g_energy command to get the temperature. Statistics over 401 steps [ 0. thru 8000.0005 ps ], 1 data sets All averages are exact over 401 steps Energy Average RMSDFluct. Drift Tot-Drift --- Temperature303.3334.780844.78082 5.46937e-06 0.043755 Heat Capacity Cv: 12.4764 J/mol K (factor = 0.00024841) I tried three further tests: a) I removed tcoupl = No option . But, it is giving same 303K as average temperature when using sd. a) I tried to specify the Nose-Hoover thermostat along with sd. But still, I found that when using sd, the average temperature is going to 303 K in stead of 300 K. From the manual description of the sd integrator: The parameter tcoupl is ignored. This explains (a) and (b). c) Finally, I tried to overlook grompp warning and went ahead and used 'md' along with Nose-Hoover thermostat. This time I found the right average temperature 300 K is being achieved.B But, I understand that for a decoupled system like here, I need to use a method like stochastic dynamics. But , I was wondering why it is reaching 303 K in stead of 300 K when using sd but 300K is achieved when using md. I looked at the mailing list where people had issues with sd and temperature control. But, I could not find a good solution. So, any help on the right protocol will be really appreciated. What type of
Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP
Justin A. Lemkul wrote: Sanku M wrote: Hi Justin, I first performed two minimization using steep and l-bfgs as the method on the solvated system for each lambda. Then I ran a 500ps NVT simulation on minimized system ( for each lambda) where essentially same .mdp file was used except the following change: gen-vel = yes gen-temp= 280 gen-seed= -1 So, essentially, I heated the minimized system up from 280 K with ref_t 300K using sd integrator. But, here also I found the average temperature goes to 303 K( instead of 300 K) . Unless you did simulated annealing in between, you didn't heat the system, you suddenly jolted the temperature, which is probably not stable. Maybe the integrator/Langevin thermostat does not react well to these conditions. Equilibrate under the conditions you wish to collect data. Your data collection uses NPT, so running NVT at a different temperature and suddenly changing the temperature and introducing pressure coupling is not an appropriate procedure. If nothing else, your data will be unreliable, even if your temperature was correct. Sorry, misread the previous .mdp settings. You're still using NVT, but the point about the temperature is still valid here. -Justin -- Justin A. Lemkul Ph.D. Candidate ICTAS Doctoral Scholar MILES-IGERT Trainee Department of Biochemistry Virginia Tech Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin -- gmx-users mailing listgmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP
Hi Justin, So, do you suggest that after minimization, I should generate the velocity at 300K instead ? i.,e for equilibration,should following the set-up? gen-vel = yes gen-temp= 300 gen-seed= -1 From: Justin A. Lemkul jalem...@vt.edu To: Discussion list for GROMACS users gmx-users@gromacs.org Sent: Tue, August 2, 2011 8:26:16 PM Subject: Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP Justin A. Lemkul wrote: Sanku M wrote: Hi Justin, I first performed two minimization using steep and l-bfgs as the method on the solvated system for each lambda. Then I ran a 500ps NVT simulation on minimized system ( for each lambda) where essentially same .mdp file was used except the following change: gen-vel = yes gen-temp= 280 gen-seed= -1 So, essentially, I heated the minimized system up from 280 K with ref_t 300K using sd integrator. But, here also I found the average temperature goes to 303 K( instead of 300 K) . Unless you did simulated annealing in between, you didn't heat the system, you suddenly jolted the temperature, which is probably not stable. Maybe the integrator/Langevin thermostat does not react well to these conditions. Equilibrate under the conditions you wish to collect data. Your data collection uses NPT, so running NVT at a different temperature and suddenly changing the temperature and introducing pressure coupling is not an appropriate procedure. If nothing else, your data will be unreliable, even if your temperature was correct. Sorry, misread the previous .mdp settings. You're still using NVT, but the point about the temperature is still valid here. -Justin -- Justin A. Lemkul Ph.D. Candidate ICTAS Doctoral Scholar MILES-IGERT Trainee Department of Biochemistry Virginia Tech Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin -- gmx-users mailing listgmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists -- gmx-users mailing listgmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP
Sanku M wrote: Hi Justin, So, do you suggest that after minimization, I should generate the velocity at 300K instead ? i.,e for equilibration,should following the set-up? You should always equilibrate under the desired conditions. Never make sudden changes, or else you're going to damage the stability of the system. -Justin gen-vel = yes gen-temp= 300 gen-seed= -1 *From:* Justin A. Lemkul jalem...@vt.edu *To:* Discussion list for GROMACS users gmx-users@gromacs.org *Sent:* Tue, August 2, 2011 8:26:16 PM *Subject:* Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP Justin A. Lemkul wrote: Sanku M wrote: Hi Justin, I first performed two minimization using steep and l-bfgs as the method on the solvated system for each lambda. Then I ran a 500ps NVT simulation on minimized system ( for each lambda) where essentially same .mdp file was used except the following change: gen-vel = yes gen-temp= 280 gen-seed= -1 So, essentially, I heated the minimized system up from 280 K with ref_t 300K using sd integrator. But, here also I found the average temperature goes to 303 K( instead of 300 K) . Unless you did simulated annealing in between, you didn't heat the system, you suddenly jolted the temperature, which is probably not stable. Maybe the integrator/Langevin thermostat does not react well to these conditions. Equilibrate under the conditions you wish to collect data. Your data collection uses NPT, so running NVT at a different temperature and suddenly changing the temperature and introducing pressure coupling is not an appropriate procedure. If nothing else, your data will be unreliable, even if your temperature was correct. Sorry, misread the previous .mdp settings. You're still using NVT, but the point about the temperature is still valid here. -Justin -- Justin A. Lemkul Ph.D. Candidate ICTAS Doctoral Scholar MILES-IGERT Trainee Department of Biochemistry Virginia Tech Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin -- gmx-users mailing listgmx-users@gromacs.org mailto:gmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org mailto:gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists -- Justin A. Lemkul Ph.D. Candidate ICTAS Doctoral Scholar MILES-IGERT Trainee Department of Biochemistry Virginia Tech Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin -- gmx-users mailing listgmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP
OKK, But, before, I have tried using md as integrator where I start the initial velocity as 280K ( which makes sense as the minimized configuration is essentially at low temperature) and have used Nose-hoover at 300K and I never had problem in getting the desired 300K after running a short NVT simulation as equilabration. The problem arises when sd integrator is used. In fact, I just tried to generate the initial velocity at 300K along with sd integrator. I got the 303 K as average temperature again. The same simulation using md integrator always gives rise to 300K as average temperature( when used in conjunction with Nose-Hoover ) , even if I generate velocity at 280 K or 300K . I guess that the initial velocity-generation temperature should not be an issue here. The sd integrator in contrast to md integrator may be the problem. From: Justin A. Lemkul jalem...@vt.edu To: Discussion list for GROMACS users gmx-users@gromacs.org Sent: Tue, August 2, 2011 8:40:21 PM Subject: Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP Sanku M wrote: Hi Justin, So, do you suggest that after minimization, I should generate the velocity at 300K instead ? i.,e for equilibration,should following the set-up? You should always equilibrate under the desired conditions. Never make sudden changes, or else you're going to damage the stability of the system. -Justin gen-vel = yes gen-temp= 300 gen-seed= -1 *From:* Justin A. Lemkul jalem...@vt.edu *To:* Discussion list for GROMACS users gmx-users@gromacs.org *Sent:* Tue, August 2, 2011 8:26:16 PM *Subject:* Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP Justin A. Lemkul wrote: Sanku M wrote: Hi Justin, I first performed two minimization using steep and l-bfgs as the method on the solvated system for each lambda. Then I ran a 500ps NVT simulation on minimized system ( for each lambda) where essentially same .mdp file was used except the following change: gen-vel = yes gen-temp= 280 gen-seed = -1 So, essentially, I heated the minimized system up from 280 K with ref_t 300K using sd integrator. But, here also I found the average temperature goes to 303 K( instead of 300 K) . Unless you did simulated annealing in between, you didn't heat the system, you suddenly jolted the temperature, which is probably not stable. Maybe the integrator/Langevin thermostat does not react well to these conditions. Equilibrate under the conditions you wish to collect data. Your data collection uses NPT, so running NVT at a different temperature and suddenly changing the temperature and introducing pressure coupling is not an appropriate procedure. If nothing else, your data will be unreliable, even if your temperature was correct. Sorry, misread the previous .mdp settings. You're still using NVT, but the point about the temperature is still valid here. -Justin -- Justin A. Lemkul Ph.D. Candidate ICTAS Doctoral Scholar MILES-IGERT Trainee Department of Biochemistry Virginia Tech Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin -- gmx-users mailing listgmx-users@gromacs.org mailto:gmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org mailto:gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists -- Justin A. Lemkul Ph.D. Candidate ICTAS Doctoral Scholar MILES-IGERT Trainee Department of Biochemistry Virginia Tech Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin -- gmx-users mailing listgmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin -- gmx-users mailing listgmx-users@gromacs.org mailto:gmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at
Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP
Sanku M wrote: OKK, But, before, I have tried using md as integrator where I start the initial velocity as 280K ( which makes sense as the minimized configuration is essentially at low temperature) and have used Nose-hoover at 300K and I never had problem in getting the desired 300K after running a short NVT simulation as equilabration. The problem arises when sd integrator is used. In fact, I just tried to generate the initial velocity at 300K along with sd integrator. I got the 303 K as average temperature again. The same simulation using md integrator always gives rise to 300K as average temperature( when used in conjunction with Nose-Hoover ) , even if I generate velocity at 280 K or 300K . I guess that the initial velocity-generation temperature should not be an issue here. The sd integrator in contrast to md integrator may be the problem. I have never had such a problem when using sd. I can always achieve the target temperature within 0.5 K, if not exactly, so I do not believe there is an inherent problem with sd. Your first post indicated that your statistics were collected over the entire timeframe, which will not be correct for reasons I have already described. The initial frames will represent unequilibrated data that may skew the averages. What if you analyze only the last half of the time? The last 1 ns? -Justin -- Justin A. Lemkul Ph.D. Candidate ICTAS Doctoral Scholar MILES-IGERT Trainee Department of Biochemistry Virginia Tech Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin -- gmx-users mailing listgmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP
On 03/08/11, Sanku M msank...@yahoo.com wrote: !-- DIV {margin:0px;} -- OKK, But, before, I have tried using md as integrator where I start the initial velocity as 280K ( which makes sense as the minimized configuration is essentially at low temperature) and have used Nose-hoover at 300K and I never had problem in getting the desired 300K after running a short NVT simulation as equilabration. The problem arises when sd integrator is used. In fact, I just tried to generate the initial velocity at 300K along with sd integrator. I got the 303 K as average temperature again. The same simulation using md integrator always gives rise to 300K as average temperature( when used in conjunction with Nose-Hoover ) , even if I generate velocity at 280 K or 300K . I guess that the initial velocity-generation temperature should not be an issue here. The sd integrator in contrast to md integrator may be the problem. Or your use of it... check out manual 7.3.3 and how the value of ref_t is relevant. Mark From: Justin A. Lemkul jalem...@vt.edu To: Discussion list for GROMACS users gmx-users@gromacs.org Sent: Tue, August 2, 2011 8:40:21 PM Subject: Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP Sanku M wrote: Hi Justin, So, do you suggest that after minimization, I should generate the velocity at 300K instead ? i.,e for equilibration,should following the set-up? You should always equilibrate under the desired conditions. Never make sudden changes, or else you're going to damage the stability of the system. -Justin gen-vel = yes gen-temp = 300 gen-seed = -1 *From:* Justin A. Lemkul jalem...@vt.edu *To:* Discussion list for GROMACS users gmx-users@gromacs.org *Sent:* Tue, August 2, 2011 8:26:16 PM *Subject:* Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP Justin A. Lemkul wrote: Sanku M wrote: Hi Justin, I first performed two minimization using steep and l-bfgs as the method on the solvated system for each lambda. Then I ran a 500ps NVT simulation on minimized system ( for each lambda) where essentially same .mdp file was used except the following change: gen-vel = yes gen-temp = 280 gen-seed = -1 So, essentially, I heated the minimized system up from 280 K with ref_t 300K using sd integrator. But, here also I found the average temperature goes to 303 K( instead of 300 K) . Unless you did simulated annealing in between, you didn't heat the system, you suddenly jolted the temperature, which is probably not stable. Maybe the integrator/Langevin thermostat does not react well to these conditions. Equilibrate under the conditions you wish to collect data. Your data collection uses NPT, so running NVT at a different temperature and suddenly changing the temperature and introducing pressure coupling is not an appropriate procedure. If nothing else, your data will be unreliable, even if your temperature was correct. Sorry, misread the previous .mdp settings. You're still using NVT, but the point about the temperature is still valid here. -Justin -- Justin A. Lemkul Ph.D. Candidate ICTAS Doctoral Scholar MILES-IGERT Trainee Department of Biochemistry Virginia Tech Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin -- gmx-users mailing list gmx-users@gromacs.org mailto:gmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org mailto:gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists -- Justin A. Lemkul Ph.D. Candidate ICTAS Doctoral Scholar MILES-IGERT Trainee Department of Biochemistry Virginia Tech Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin -- gmx-users mailing list gmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists Blacksburg, VA
Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP
Hi Justin, I did calculate the average temperature using last 2 ns of the 8ns data. It is still same 303K. From: Justin A. Lemkul jalem...@vt.edu To: Discussion list for GROMACS users gmx-users@gromacs.org Sent: Tue, August 2, 2011 9:55:40 PM Subject: Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP Sanku M wrote: OKK, But, before, I have tried using md as integrator where I start the initial velocity as 280K ( which makes sense as the minimized configuration is essentially at low temperature) and have used Nose-hoover at 300K and I never had problem in getting the desired 300K after running a short NVT simulation as equilabration. The problem arises when sd integrator is used. In fact, I just tried to generate the initial velocity at 300K along with sd integrator. I got the 303 K as average temperature again. The same simulation using md integrator always gives rise to 300K as average temperature( when used in conjunction with Nose-Hoover ) , even if I generate velocity at 280 K or 300K . I guess that the initial velocity-generation temperature should not be an issue here. The sd integrator in contrast to md integrator may be the problem. I have never had such a problem when using sd. I can always achieve the target temperature within 0.5 K, if not exactly, so I do not believe there is an inherent problem with sd. Your first post indicated that your statistics were collected over the entire timeframe, which will not be correct for reasons I have already described. The initial frames will represent unequilibrated data that may skew the averages. What if you analyze only the last half of the time? The last 1 ns? -Justin -- Justin A. Lemkul Ph.D. Candidate ICTAS Doctoral Scholar MILES-IGERT Trainee Department of Biochemistry Virginia Tech Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin -- gmx-users mailing listgmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists -- gmx-users mailing listgmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP
Sanku M wrote: Hi Justin, I did calculate the average temperature using last 2 ns of the 8ns data. It is still same 303K. Going back to the original .mdp file, I can see that you're using some incorrect settings. You've set rlist=rvdw even though you're using a shifted potential for van der Waals interactions. Surely grompp should have warned you that this would lead to poor energy conservation. Your solvent is probably heating up as a result and leading to the incorrect temperature. -Justin *From:* Justin A. Lemkul jalem...@vt.edu *To:* Discussion list for GROMACS users gmx-users@gromacs.org *Sent:* Tue, August 2, 2011 9:55:40 PM *Subject:* Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP Sanku M wrote: OKK, But, before, I have tried using md as integrator where I start the initial velocity as 280K ( which makes sense as the minimized configuration is essentially at low temperature) and have used Nose-hoover at 300K and I never had problem in getting the desired 300K after running a short NVT simulation as equilabration. The problem arises when sd integrator is used. In fact, I just tried to generate the initial velocity at 300K along with sd integrator. I got the 303 K as average temperature again. The same simulation using md integrator always gives rise to 300K as average temperature( when used in conjunction with Nose-Hoover ) , even if I generate velocity at 280 K or 300K . I guess that the initial velocity-generation temperature should not be an issue here. The sd integrator in contrast to md integrator may be the problem. I have never had such a problem when using sd. I can always achieve the target temperature within 0.5 K, if not exactly, so I do not believe there is an inherent problem with sd. Your first post indicated that your statistics were collected over the entire timeframe, which will not be correct for reasons I have already described. The initial frames will represent unequilibrated data that may skew the averages. What if you analyze only the last half of the time? The last 1 ns? -Justin -- Justin A. Lemkul Ph.D. Candidate ICTAS Doctoral Scholar MILES-IGERT Trainee Department of Biochemistry Virginia Tech Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin -- gmx-users mailing listgmx-users@gromacs.org mailto:gmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org mailto:gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists -- Justin A. Lemkul Ph.D. Candidate ICTAS Doctoral Scholar MILES-IGERT Trainee Department of Biochemistry Virginia Tech Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin -- gmx-users mailing listgmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP
Finally, I achieved the desired 300K average temperature using sd integrator . I had to reduce the time step from 0.002 ps to 0.001 ps and then running the NVT could produce 300K as average temperature. Still Not sure why md integrator still could 300K average temperature using 0.002 ps as time-step. From: Justin A. Lemkul jalem...@vt.edu To: Discussion list for GROMACS users gmx-users@gromacs.org Sent: Tue, August 2, 2011 10:16:17 PM Subject: Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP Sanku M wrote: Hi Justin, I did calculate the average temperature using last 2 ns of the 8ns data. It is still same 303K. Going back to the original .mdp file, I can see that you're using some incorrect settings. You've set rlist=rvdw even though you're using a shifted potential for van der Waals interactions. Surely grompp should have warned you that this would lead to poor energy conservation. Your solvent is probably heating up as a result and leading to the incorrect temperature. -Justin *From:* Justin A. Lemkul jalem...@vt.edu *To:* Discussion list for GROMACS users gmx-users@gromacs.org *Sent:* Tue, August 2, 2011 9:55:40 PM *Subject:* Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP Sanku M wrote: OKK, But, before, I have tried using md as integrator where I start the initial velocity as 280K ( which makes sense as the minimized configuration is essentially at low temperature) and have used Nose-hoover at 300K and I never had problem in getting the desired 300K after running a short NVT simulation as equilabration. The problem arises when sd integrator is used. In fact, I just tried to generate the initial velocity at 300K along with sd integrator. I got the 303 K as average temperature again. The same simulation using md integrator always gives rise to 300K as average temperature( when used in conjunction with Nose-Hoover ) , even if I generate velocity at 280 K or 300K . I guess that the initial velocity-generation temperature should not be an issue here. The sd integrator in contrast to md integrator may be the problem. I have never had such a problem when using sd. I can always achieve the target temperature within 0.5 K, if not exactly, so I do not believe there is an inherent problem with sd. Your first post indicated that your statistics were collected over the entire timeframe, which will not be correct for reasons I have already described. The initial frames will represent unequilibrated data that may skew the averages. What if you analyze only the last half of the time? The last 1 ns? -Justin -- Justin A. Lemkul Ph.D. Candidate ICTAS Doctoral Scholar MILES-IGERT Trainee Department of Biochemistry Virginia Tech Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin -- gmx-users mailing listgmx-users@gromacs.org mailto:gmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org mailto:gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists -- Justin A. Lemkul Ph.D. Candidate ICTAS Doctoral Scholar MILES-IGERT Trainee Department of Biochemistry Virginia Tech Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin -- gmx-users mailing listgmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin -- gmx-users mailing listgmx-users@gromacs.org mailto:gmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org mailto:gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists -- Justin A. Lemkul Ph.D. Candidate ICTAS Doctoral
Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP
On 3/08/2011 2:23 PM, Sanku M wrote: Finally, I achieved the desired 300K average temperature using sd integrator . I had to reduce the time step from 0.002 ps to 0.001 ps and then running the NVT could produce 300K as average temperature. Still Not sure why md integrator still could 300K average temperature using 0.002 ps as time-step. So your integration step was too large to conserve energy. There have been two suggestions made about why that might be true. Have you investigated them? Mark *From:* Justin A. Lemkul jalem...@vt.edu *To:* Discussion list for GROMACS users gmx-users@gromacs.org *Sent:* Tue, August 2, 2011 10:16:17 PM *Subject:* Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP Sanku M wrote: Hi Justin, I did calculate the average temperature using last 2 ns of the 8ns data. It is still same 303K. Going back to the original .mdp file, I can see that you're using some incorrect settings. You've set rlist=rvdw even though you're using a shifted potential for van der Waals interactions. Surely grompp should have warned you that this would lead to poor energy conservation. Your solvent is probably heating up as a result and leading to the incorrect temperature. -Justin *From:* Justin A. Lemkul jalem...@vt.edu mailto:jalem...@vt.edu *To:* Discussion list for GROMACS users gmx-users@gromacs.org mailto:gmx-users@gromacs.org *Sent:* Tue, August 2, 2011 9:55:40 PM *Subject:* Re: [gmx-users] Problem with temperature and stochastic dynamics in FEP Sanku M wrote: OKK, But, before, I have tried using md as integrator where I start the initial velocity as 280K ( which makes sense as the minimized configuration is essentially at low temperature) and have used Nose-hoover at 300K and I never had problem in getting the desired 300K after running a short NVT simulation as equilabration. The problem arises when sd integrator is used. In fact, I just tried to generate the initial velocity at 300K along with sd integrator. I got the 303 K as average temperature again. The same simulation using md integrator always gives rise to 300K as average temperature( when used in conjunction with Nose-Hoover ) , even if I generate velocity at 280 K or 300K . I guess that the initial velocity-generation temperature should not be an issue here. The sd integrator in contrast to md integrator may be the problem. I have never had such a problem when using sd. I can always achieve the target temperature within 0.5 K, if not exactly, so I do not believe there is an inherent problem with sd. Your first post indicated that your statistics were collected over the entire timeframe, which will not be correct for reasons I have already described. The initial frames will represent unequilibrated data that may skew the averages. What if you analyze only the last half of the time? The last 1 ns? -Justin -- Justin A. Lemkul Ph.D. Candidate ICTAS Doctoral Scholar MILES-IGERT Trainee Department of Biochemistry Virginia Tech Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin -- gmx-users mailing list gmx-users@gromacs.org mailto:gmx-users@gromacs.org mailto:gmx-users@gromacs.org mailto:gmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org mailto:gmx-users-requ...@gromacs.org mailto:gmx-users-requ...@gromacs.org mailto:gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists -- Justin A. Lemkul Ph.D. Candidate ICTAS Doctoral Scholar MILES-IGERT Trainee Department of Biochemistry Virginia Tech Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin -- gmx-users mailing list gmx-users@gromacs.org mailto:gmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org mailto:gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin -- gmx-users mailing list gmx-users@gromacs.org