From: Mayne, Christopher G (cmayne2_at_illinois.edu)
Date: Tue Jun 11 2013 - 15:32:40 CDT

Surya,

Please retain the cc to vmd-l_at_ks.uiuc.edu<mailto:vmd-l_at_ks.uiuc.edu> on these email correspondences so that everyone can gain from the discussion.

I have previously tried to reproduce the error you mention here using the input provided from your dropbox. Using the debugging log I was able to reconstruct your input but was unable to reproduce your error. Please make sure that you are using the latest ffTK code by downloading the most recent VMD 1.9.2 alpha build. I suggest that you pare down your input to the smallest example that still generates the error, and then clearly define that input so that I can reproduce the error. Only then can i be more helpful.

Regards,
Christopher Mayne

On Jun 11, 2013, at 11:13 AM, surya narayanan chandrasekaran wrote:

thank you. it worked but i am getting this below error many times. I am tired of it. even i logout of system by closing the terminal window. Even making new folders every time still i am getting this error. the real frustrating thing is the error message pop-up after all the iterations for an hour or so.

in the terminal window it shows

Simulated annealing:
Initial temperature T0=25.0
Number of T cycles Tsteps=20
Simulated annealing step 0: T=25.0
Already converged after T=25.0
Simulated annealing step 1: T=0
Simulated annealing converged.

expected floating-point number but got ""
expected floating-point number but got ""
    while executing
"format %1.3f [lindex $finalCharges $i]"

On Tue, Jun 11, 2013 at 3:42 PM, Mayne, Christopher G <cmayne2_at_illinois.edu<mailto:cmayne2_at_illinois.edu>> wrote:
When you load the QM target data in ffTK tries to automatically parse the atom name out of the filename. For the 120A/B log files you need to make sure that you change the O1A-120A to O1A.

Christopher Mayne

On Jun 11, 2013, at 6:47 AM, surya narayanan chandrasekaran wrote:

so this the error: what i did was i loaded psf, pdb and the sp log files then O1A three log files (o1A, o1A-120A, o1A-120b) then when i click run simulation this what i get.

 atomselect: cannot parse selection text: name O1A-120a and resname BCL
atomselect: cannot parse selection text: name O1A-120a and resname BCL
    while executing
"atomselect $refmolid "name [lindex $atomList $i] and resname $resName" frame $i"
    (procedure "::ForceFieldToolKit::ChargeOpt::optimize" line 213)
    invoked from within
"::ForceFieldToolKit::ChargeOpt::optimize"
    (procedure "::ForceFieldToolKit::gui::coptRunOpt" line 55)
    invoked from within
"::ForceFieldToolKit::gui::coptRunOpt "
    invoked from within
".fftk_gui.hlf.nb.chargeopt.runOpt invoke "
    invoked from within
".fftk_gui.hlf.nb.chargeopt.runOpt instate {pressed !disabled} { .fftk_gui.hlf.nb.chargeopt.runOpt state !pressed; .fftk_gui.hlf.nb.chargeopt.runOpt in..."
    (command bound to event)

On Tue, Jun 11, 2013 at 12:32 AM, Mayne, Christopher G <cmayne2_at_illinois.edu<mailto:cmayne2_at_illinois.edu>> wrote:
ffTK intentionally generates three input files for carbonyls. interactions around carbonyls are very different between QM and MM levels of theory; therefore, the current best practice (as we understand it) is to fit BOTH the QM position (120 degrees) and the MM position (180 degrees). so, the generation of multiple QM input files is intentional.

looking back over some of the preceding email thread, I was unable to reproduce your error using the input you provided.

Christopher Mayne

On Jun 10, 2013, at 5:24 PM, surya narayanan chandrasekaran wrote:

Hi,

I was asking about the multiple file created by fftk for oxygen molecule. For eg for atom name o1A it creates 120a and 120b additional files probably due to the atom type it places additional water molecule and if I load the optimized log file to fftk then I am getting multiple bond error. I am not bothered about the visualization state.

Thnk q
Surya

Thank you
Surya

On Monday, June 10, 2013, Mayne, Christopher G wrote:
the "multiple bond problem" has no effect on the optimization, it is purely an artifact of how VMD guesses bonding patterns when the data is loaded for visualization purposes. ab initio QM calculations don't know about explicit bonds; they only care about spatial arrangements of atoms.

interaction energy = E(cmpd+water) - [E(cmpd) + E(water)]

for each step of the optimization (i.e. for a given set of assigned charges) water is positioned at several points along the interaction vector. The MM interaction energy is computed at each position, and the MME minimum is identified along with the distance at which the minimum occurs. These MM-computed data are compared to the QM optimized energy and distance and applied to the "objective function" which operates via a partial least squares fit and is summed over all interactions in the target data. A cumulative "score" is then returned to the optimizer, and used to generate a new set of test charges for the next iteration.

Christopher Mayne

On Jun 10, 2013, at 3:55 PM, surya narayanan chandrasekaran wrote:

Hi,

If you have any idea or suggestion on how can solve the multiple bond problem for oxygen file created by the fftk kindly let me know.
I also looked at the fftk script. I would like to know the reference of the interaction energy function that is used in the script. If you can also briefly state the optimization procedure carried out in the script with respect to atoms selected and the log files given it will be really helpfull for me.

Thank you.
Surya

On Thursday, May 30, 2013, Mayne, Christopher G wrote:
Suryanarayanan,

I looked over the ChargeOpt.debug.log file and setup an optimization that matched yours as closely as possible using files I downloaded from your dropbox. Optimizations run both in downhill and simulated annealing modes terminated without error--I was unable to reproduce the error you are reporting using the provided input. Unfortunately, if I cannot reproduce the error, I don't have much insight into resolving your problem.

I did not see the ChargeOpt.log file in your dropbox, which may or may not provide additional information or hints.

Regards,
Christopher Mayne

On May 30, 2013, at 2:43 PM, surya narayanan chandrasekaran wrote:

Hi,

Thank you for the reply. I have kept the files in my drop box below link. three files ChargeOptScript.tcl, ChargeOpt.log and ChargeOpt.debug.log https://www.dropbox.com/sh/6klo7fzbgshgtnk/LpfqxDz-Pf/forcefield?n=106188702

On Thu, May 30, 2013 at 2:56 PM, Mayne, Christopher G <cmayne2_at_illinois.edu> wrote:
I missed a section of a previous email that I should probably return to. there is a pretty clear theme to my responses. please see below.

Regards,
Christopher Mayne

>
> also i would like to state. how i overcome the previous error maybe it will be helpfull for others.
> 1) one of the atom name Oxygen has been replaced by FFTK to carbon and after doing the RHF calculation the water molecule just bounces out and i got a trajectory of 50 frames for the step so i removed it.
> 2) Aother error is really stupid like the toolkit creates some file with 120a and 120b for the same atom name when you load the optimized file into the next step it creates an error "atom name" has multiple bonds. so i removed that file too.
>

1. It is the responsibility of the user to ensure that the Gaussian optimization files terminate normally, and yield water positions that represent reasonable and useful energy minima. The entire point of the "Load LOG Files" button in the "Water Int." tab is to aid users in visualizing this and catching such problematic interactions.

2. I would argue that the user that discards target data without a clear justification is "really stupid." The toolkit generates these files for reasons that are explained in the resources linked to on the documentation pages. You should read up on CHARMM and CGenFF. Also, overcoming this error is simple--make sure that these target data are assigned to an appropriate atom name (i.e., just delete the 120x from the automatically parsed filename).

A major contribution to your troubles is that you are trying to optimize charges for the entire substituted porphyrin structure that includes 97 input files of target data. That is overwhelming your ability understand where your charge optimization is going off of the rails.

> since quantum calculation computes charges for all the atoms you dont really need to worry placing water molecule for each atom.

This is an ignorant statement within the context of parameterization of CHARMM compatible force fields. Please read up on how charges are properly derived according to the underlying principles of CHARMM and CGenFF force fields.

> but it would be really helpful if its clearly stated in the journal of FFTK how exactly the final charges are computed from each file.
>

We have submitted a manuscript that covers these details. If you perused more of the CHARMM and CGenFF lit

--
Suryanarayanan C
M.S Chemist
+94-01794019116
--
Suryanarayanan C
M.S Chemist
+94-01794019116
--
Suryanarayanan C
M.S Chemist
+94-01794019116