On screen information
During the run of the simulation a lot of information is send to the screen. Starting when the system is being initialized, when the simulation is actually running, when files are dumped and when the simulation is finished. This chapter of the manual discusses some of the most common occuring information on screen in the order it should occur during the course of the simulation.
Initializing
PumMa starts of with announcing the current version and whether the simulation is run sequential or in parallel. Next some information on reading input files is given, and after the topology has been parsed, the following lines, depending on your system, are reported
Molecule W has 1 atoms, 0 bonds, 0 bendings, 0 torsions, and 0 impropers. Molecule G4T4T4 has 12 atoms, 11 bonds, 4 bendings, 0 torsions, and 0 impropers.
Next more information on some simulation parameters are reported. The next set of lines that might be interesting inform the user about the system size and could look like
System size information: 9399 particles, double precision Universe: 20.773973 21.323662 28.301245, density 0.749713
This gives the total number of particles in the system, the precision (which is almost always double), and the x-, y- and z-dimensions of the system, and the particle number density.
To increase the computational speed the system is subdivided into a large number of small cells. When the non-bonded pairlist needs to be (re)build, only the surrounding cells of the central cell needs to be considered, instead of the entire system. The information PumMa reports on this cell division looks like
Cell division information: NPX: 1, NPY: 1 NCx: 13, NCy: 13, NCz: 13 Sx: 2, Sy: 2, Sz: 2 cSx: 0, cSyt: 0, cSyb: 0
The precise explanation is beyond the scope of this manual and is, normally, not of interest to the user.
The next step is to read all the actual coordinates and velocities for all the molecules in the system.
Allocated: 26774872 bytes Reading 2928 positions/velocities from file G4T4T4IT25000.pcf Reading 5386 positions/velocities from file WIT25000.pcf Proc: 0 has 8314 particles
This report tells the user the amount of bytes used on the system, the number of lines/particles loaded into the simulation per molecule and the distribution of the particles across the processors.
If fixing is enabled, the next stage is that the file containing the fixed atoms is parsed.
Reading fixed atoms from file specials.pif...
Finally, the initialization is completed and the simulation can be started. The last bit of information is about the total number of particles and the total number of steps to be run.
Starting simulation... System consists of 8314 particles... Running for 35000 iterations...
Most of the reports discussed above are also send to a file called sysdescr.txt, so they can be reviewed after the simulation has finished.
During the simulation
Depending on the settings in the input options file, some or none information is reported to the screen during the simulation. For instance with the switch option silent set to on, no information is reported at all. By default this value is set to off, so quite some information is reported to the screen. Furthermore, the options Erep and Crep determine the amount of information reported. For instance if Erep is set to 5 every 5 iteration information on the total energy of the system is reported and might look like this
IT= 27815 Elj=-16.421252195 Ebnd= 1.248041493 Ebend= 2.103100709 Etors= 0.000000000 Eimps= 0.000000000 Ecb= 0.000000000 Ekin= 3.826995230 Etot=-12.019475641 Rho= 8.240848752 P=28.440201318 T=306.854137206
In order of appearance the iteration number (IT) is reported, followed by the energy arising from the Van der Waals interactions (Lennard-Jones, Elj), the energy arising from the bonds (Ebnd), from the angles (Ebend), the dihedrals (Etors), the impropers (Eimps), the electrostatic interactions (Ecb), the kinetic energy (Ekin) and the sum of all these energies (Etot). Depending on the choice of Epp in your input options file, the energy information is the average per particle (the default setting) or simply the total sum. In the latter case all individual energy contributions sum up to the total energy Etot, but in the former this does not need to be the case, since the energies are averaged per particle, per bond, per angle, and so on. After all the energies are reported on the current iteration, the particle number density (Rho) is reported, and next the pressure of the system (P). The last reported item (T) is the current temperature of the system.
When simulated annealing is used, the final column of the report line gives the current value of the temperature target (Treq) in the Berendsen temperature loose coupling scheme. Furthermore, when steering is applied (steeredMD is on) a column is added to report the amount of work done on the system to steer particles.
At some regular intervals temperatures need to be scaled (due to the use of simulated annealing), the linear momentum has to be corrected (set in the input options file by the option lmrep), or the coordinates and velocities are dumped to disk (depending on the value of Crep). Any of these reports might look like
At IT 30000 requested temperature is reassigned to 301.0 T* Total mass: 9871.624000 Linear momentum: -29.620877 -55.048626 29.514067 Linear momentum correction: -0.003001 -0.005576 0.002990 Writing 2928 positions/velocities to G4T4T4IT30000.pcf Writing 5386 positions/velocities to WIT30000.pcf
When the simulation is performed in an isobaric ensemble (hence the pressure is fixed, but not the volume), the box size is adjusted throughout the duration of the simulation. This adjustment is reported to screen and might look like
Procid 0 Svals are changing from Sx=1 Sy=1 Sz=1 to newSx=2 newSy=1 newSz=1 Procid 0 NCx= 3, NCxp=3, cSx= 0 NCy= 3, NCyp=3, cSyt=0, cSyb=0, NCz=4 procid 0 has new NBcells environment
This information basically tells the user that the cell division is being updated since the box size is adjusted. Normally the user does not need to be concerned about these reports, although sometimes they can lead to an error and causing the simulation to crash.
Finalizing
When the simulation has finished (hence the final iteration as specified by max_IT in the input options file is reached) the final coordinates and velocities are dumped to file and the simulations finishes by informing the user on the time spent in each type of calculation. This timing report looks like
Total timings:
Bonding : 0.64 % <= 3.81 sec.
Bending : 0.50 % <= 2.98 sec.
Torsion : 0.00 % <= 0.02 sec.
Improper : 0.00 % <= 0.00 sec.
Nonbonded int.: 95.03 % <= 565.10 sec.
- make_list : 14.89 % <= 84.17 sec.
- use_list : 85.11 % <= 480.93 sec.
Leap-frog : 2.15 % <= 12.79 sec.
File dumping : 0.07 % <= 0.43 sec.
Macro. Quant. : 1.47 % <= 8.72 sec.
Overhead : 0.13 % <= 0.78 sec.
Total time : 100.00 % <= 594.63 sec.
Average time per step : 0.1189 sec.
Verlet list building counts:
On average the Verlet list has been built every 20.83 steps.
The Verlet list has been built 240 times.
A total of 5000 steps has been performed.
Where the timings of course depend on the type of system under investigation. In this example barely any time was spent on calculating dihedrals and impropers, but almost all time has been used on the non-bonded interactions (as expected). The division in the non-bonded timing into make_list and use_list refer to the different loops used when building the pairlist (make_list) and when using the pairlist (use_list). Depending on the settings (such as vlrep, Rc or Rl) these timings might change.
<< Running parallel | Manual | Possible errors >>