Randomization During Molecular Dynamics

Aproblem in searching conformational space using molecular dynamics simulations is repeating a trajectory that generates the same structures. To reduce this possibility, you can randomize the velocities of the atoms. 

To increase the possibly of finding new conformers, try one of these techniques  

---

Let us compare the kinetics of the selective-solvent-induced collapse of protein-like copolymers with the collapse of random and random-block copolymers [18]. Several kinetic criteria were examined using Langevin molecular dynamics simulations. There are some general results, which seem to be independent of the nature of interactions or the kinetic criteria monitored during the collapse. Here, we restrict our analysis to the evolution of the characteristic ratio f = (Rgp/Rg ) that combines the partial mean-square radii of gyration calculated separately for hydrophobic and hydrophilic beads, k2n and Rg . This ratio takes into account both the properties of compactness and solubility for a heteropolymer globule [70] (compactness is directly related to the mean size of the hydrophobic core, whereas solubility should be dependent on the size of the hydrophilic shell). 

Whereas Bernal had to rely primarily on mechanically produced random sphere packings in his work on liquid structure, the molecular dynamics and Monte Carlo computer simulation techniques have, during the past three decades, provided researchers with new and powerful experimental tools enabling a much closer look into the structure of the liquid—one has available the trajectory of every atom. Despite this, computer simulation has been used principally to calculate liquid-state correlation functions. This situation, to quote Lumsden and Wilson [6], ... appears to stem in part from a peculiar and fundamental relation that has always existed between experiment and theory in science the importance of experimental data is judged by the theory to which it is applied. As the physicist Arthur Eddington said half seriously, no fact should be accepted as true until it has been confirmed by theory. Unless an attractive theory exists that decrees certain kinds of information to be important, few scientists will set out to acquire the information. Thus, it is only infrequently that computer simulations have been used to characterize liquid structure in ways other than those dictated by the prevailing liquid theory. 

However, single molecule molecular dynamics for liquid crystal molecules can often be problematic. Many liquid crystal systems have torsional energy barriers in excess of 12 kJ mol separating conformations of similar energy (see Sec. 3.2.1). Such barriers can be difficult to cross during the course of a short MD simulation, and this can result in molecules becoming periodically trapped in regions of phase space. This has led to the development of stochastic dynamics techniques where random noise added to the equations of motion is de-... 

---