Reference ensembles & other advanced controls

This page collects the remaining advanced keywords: reference/debugging switches, box and trajectory controls, a couple of chain-handling options, and the experimental gate that unlocks the not-yet-stable features.

Reference ensembles

Sometimes the most useful comparison is the system with its interactions turned off - a well-defined reference state that isolates the effect of excluded volume or of the backbone.

NON_INTERACTING : True: Zero all interaction energies and run a pure excluded-volume reference ensemble. The parameter file is still required (bead types must be defined) but its pairwise and solvation energies are ignored. This is the natural “ideal chain in a box” baseline to compare an interacting run against - anything that differs from the non-interacting ensemble is a genuine consequence of the interactions.
ANGLES_OFF : True: Disable the backbone-angle penalties entirely, so the chains are perfectly flexible. With this set you do not need ANGLE_PENALTY lines in the parameter file. Useful for isolating the role of chain stiffness, or simply for models where stiffness is not wanted.

The two switches are independent and can be combined: NON_INTERACTING : True with ANGLES_OFF : True is the fully ideal, freely-jointed, excluded-volume-only chain.

Energy-consistency checking

PIMMS tracks the total energy incrementally - each accepted move adds its energy change to a running total rather than recomputing the whole Hamiltonian. That is what makes it fast, but it also means a subtle bookkeeping bug would slowly drift the tracked energy away from the truth.

ENERGY_CHECK : <freq>: Every <freq> steps, recompute the total energy from scratch and compare it to the incrementally tracked value. If they disagree the run aborts with an exception and dumps the offending configuration to CONFIG_AT_ENERGY_FAIL.pdb / .xtc so you can inspect it. This is an O(N) check: cheap to run occasionally on a modest system, but expensive if you run it every step on a large one. The default frequency is deliberately infrequent; lower it when you are developing or debugging and want tight verification, raise it (or leave it) for production.

Box and equilibration controls

RESIZED_EQUILIBRATION / EQUILIBRATION_OFFSET

Equilibrate in a smaller box and then grow to the production DIMENSIONS at the end of equilibration (chains are re-centred as the box expands). This lets you condense or assemble a system at high effective concentration and then relax it into the full production volume - often much faster than waiting for assembly to happen at the production density.

RESIZED_EQUILIBRATION gives the equilibration box size (2 or 3 values), and EQUILIBRATION_OFFSET places that smaller box within the full box. Both are constrained: RESIZED_EQUILIBRATION must be <= DIMENSIONS in every dimension, and EQUILIBRATION_OFFSET + RESIZED_EQUILIBRATION must fit inside DIMENSIONS. The equilibration phase is always run under hardwall boundaries (forced internally, so a system is never resized while chains straddle a periodic face); your production HARDWALL setting takes over once the box has grown. The feature is incompatible with RESTART_OVERRIDE_DIMENSIONS and with PBC restart files. See Simulation setup: boundaries and box resizing for the surrounding set-up keywords.

DIMENSIONS            : 60 60 60
RESIZED_EQUILIBRATION : 30 30 30      # equilibrate at 8x the density...
EQUILIBRATION_OFFSET  : 15 15 15      # ...centred in the production box

AUTOCENTER : True

For a single-chain simulation, keep the chain centred in the box on every frame. This stops a lone chain drifting to the edge of a hardwall box or wandering across periodic images, which keeps trajectories tidy for visualisation and analysis.

Chain-handling options

CASE_INSENSITIVE_CHAINS : False: By default (True) chain sequences are upper-cased when the keyfile is read, so a and A are the same bead type. Set this to False to treat case as significant, which effectively doubles the alphabet of bead types available (A and a become distinct). Every bead letter used in a chain must still be defined in the parameter file.
EXTRA_CHAIN : <count> <sequence>: Add chains that were not in the original RESTART_FILE when restarting from a saved configuration. The format matches the CHAIN keyword (<number of chains> <sequence>), multiple EXTRA_CHAIN lines are allowed for different species, and the new chains are inserted at random so as not to overlap anything already present. This is how you take the end-state of one run and continue it with additional material - and it can be repeated as many times as you like. EXTRA_CHAIN requires a RESTART_FILE (there must be an existing configuration to add to). It pairs naturally with a freeze file: freeze the restart configuration and let the extra chains explore around it (as in the star_destroyer demo).

The experimental gate

EXPERIMENTAL_FEATURES : True: Unlocks the remaining not-yet-stable keywords and moves. As of this writing the only gated feature is the MOVE_VMMC collective move (and its VMMC_* tuning keywords). It is not guaranteed to behave correctly in every configuration, so the recommendation is to leave the gate False unless you specifically need it - and to sanity-check the results carefully when you do. (The pull, jump-and-relax and temperature-switch (TSMMC) moves, non-cubic boxes, and the EXTRA_CHAIN, FREEZE_FILE and EQUILIBRATION_OFFSET keywords have all graduated out of the experimental gate and need no special flag.)