Possible errors
However unlikely, it is possible that PumMa stops prematurely due to an error that occured during the simulation. Most of the time these errors are caused by incorrect input options, atoms placed too close to each other, unknown parameters, and so on. They are easily solved be adjusting the input system. Unfortunately, some errors have a different cause and are not so easily solved. This section of the manual discusses some of the most encountered errors when using PumMa and provides some guidelines to solve the problem. If, for any reason, the error does not disappear, the user should contact any of the developers in order to solve the problem.
Unknown variable
Reading options from file input.pif... ERROR from 0 0: Unknown variable: temp
When parsing the input options file all options are checked against all possibilities within the PumMa code. If an option is encountered that does not exist within PumMa, this error is generated and the simulation stops. Removing the unknown option or naming it correctly will remove the error.
Unknown atom type
Molecule G4T4T4 has 12 atoms, 11 bonds, 4 bendings, 0 torsions, and 0 impropers. ERROR from 0 0: Unknown atom type: Q
When in a topology file an atom of a molecule is declared to be of type that is not present in the parameter file, the simulations stops, issueing this error. Renaming the atom, or providing the correct parameters should solve the problem.
Besides an error with atom typing a similar type of error can occur if incorrect bond, angle, dihedral, etc. typing or parameters are supplied. Examining the topology files and parameter file should allow the user to solve the problem.
Unknown atom name
Molecule G4T4T4 has 12 atoms, 11 bonds, 4 bendings, 0 torsions, and 0 impropers. ERROR from 0 0: Unknown atom type: ATOM
Besides atom typing errors, PumMa sometimes crashes with the same error message, although the cause is completely different. This type of error is referred to as the unknown atom name error. This error can be identified since the atom 'type' that PumMa reports to be wrong, matches an atom name (as in the first column of the topology file) instead of an atom type (the second column in the same file).
The most common cause for this error is due to the assignment of the partial charges in the third column of the topology file. When no charges are specified in this file, this unknown atom name error is likely to occur. It can therefore easily be fixed by specifying charges (which could be zero of course).
Array too small
Starting simulation... System consists of 9399 particles... Running for 500000 iterations... ERROR from 0 0: Binding array too small
The occurance of this error means that something went wrong after the topologies and coordinates have been parsed. Most likely, as with the unknown atom name error, the absence of the charges in the topology files is to blame. This error is not restricted to bonds. It can also occur with angles, dihedrals and so on. Assigning charges should solve the problem.
Lost intramolecular structure
Bond: 2669 != 2684 Writing 6471 positions/velocities to WIT14.pcf Writing 2928 positions/velocities to G4T4T4IT14.pcf ERROR from 0 0: Binding(s) lost
During the simulation intramolecular structures, such as bonds, angles, dihedrals, could be extended beyond reasonable values, causing them to break. This occurance is referred to as if the intramoleculare structure is lost. There are two common possible causes for this error.
The first is when the parameters are not matched with the molecular structure under investigation. For instance if two atoms that share a bond are initially placed 0.45 nm apart, but according to the parameter file, the reference value for the bond length is 0.90 nm, the bond is overstretched, and will rupture at any next iteration. Hence, the bond is 'lost'.
The second cause can be the size of the time step. If it is chosen to large, atoms are able to move too fast than would be reasonable, and, consequently, speed up too much. This could result in high velocity collisions in which it is likely that two atoms connected by a bond are overstretched, causing the bond to rupture, and as a result the simulation to crash.
Universe too small
ERROR from 0 0: Universe too small
By far the most difficult error in PumMa is this one. It can have multiple causes and is sometimes not easily removed. The most common cause, however, lies with the Berendsen loose pressure coupling. When this coupling is too loose (hence, muc is too large), the simulation box has the ability to scale rapidly. Unfortunately this may lead to incorrect cell divisions, causing some of the cells to become too small, and making the simulation unstable. The program subsequently crashes with the universe too small error. Most of the time a report that looks 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
has been send to the screen just before the error. The user then knows that the problem probably lies with the value of muc and should change it accordingly in the input options file.
Another possible cause for this error is the system size. If the system initially is too small, the same cell division problem as described above occurs, causing the simulation to crash even before it has started. Increasing the size of the system should solve the problem. For molecules floating in a vacuum this is easily accomplished by increasing the values in the first line of the system configuration file. For periodic systems the best way to deal with this error, is to incorporate one periodic copy within the actual system (increasing the system size by a factor two).
<< On screen information | Manual | Output Files >>