SETUP

NAME
SETUP - This is the basic command to decide which kind of simulation, Hamiltonian SGE simulation or SGE simulation in the space of collective coordinates, one wants to carry out. This command also defines the number of ensembles, the scaling options and the restart option.

SYNOPSIS
SETUP $nstates$ [$scale_1$ $scale_2$ $scale_3$] $irest$

DESCRIPTION
Hamiltonian SGE simulations.
If the parameters $scale_1$, $scale_2$ and $scale_3$ (real numbers) are specified in the SETUP command, then a Hamiltonian SGE simulation with total or partial scaling of the potential energy is performed (simulated-tempering and solute-tempering like simulations, respectively). In such a case the SETUP command is used to define the number of ensembles (nstates; integer number) and the lowest scaling factor (i.e the highest temperature) of the last ensemble. The number of replicas in the SGE simulations is equal to the number of processors passed to the MPI routines ($nprocs$). At variance with REM, $nprocs$ may be not equal to $nstates$. The restart option of a SGE simulation is controlled by irest (integer number). The three parameters, $scale_1$, $scale_2$ and $scale_3$, can be different and refer to scaling features of different parts of the potential energy. $scale_1$ refers to the bending, stretching and improper torsional potentials, $scale_2$ to the (proper) torsional potential and to the 1-4 non-bonded interactions and $scale_3$ refers to the non bonded potential. IMPORTANT NOTE: when the Ewald summation is used together with the command SEGMENT(&SGE), $scale_3$ scales only the direct (short-ranged) part of the electrostatic interactions and the (long-ranged) reciprocal part has a scaling factor of 1 (i.e. these interactions are not scaled). If $scale_1 = scale_2 = scale_3$, then an equal scaling is applied to all parts of the potential (it corresponds to a simulated tempering simulation). If $irest = 0$, the run is restarted from a previous one. This implies that the directories PARXXXX are present and are equal in number to $nprocs$, i.e. the number of replicas. If $irest \ne 0$ then the run refers to a cold start from scratch and

• $irest=1$: the scaling factors associated with the intermediate ensembles are derived according to a geometric progression, namely $scale_i(m)=scale_i^{(m-1)/(nstates -1)}$, where $scale_i(m)$ is the scaling factor for the potential $i$ of the ensemble $m$ with $1\le m \le nstates$. For example, if $scale_i = 0.6$ and $nstates = 4$, then $scale_i(1)=1$, $scale_i(2)= 0.843433$, $scale_i(3) = 0.711379$ and $scale_i(4) = scale_i = 0.6$. The $nprocs$ replicas are initially distributed as described in Section 6.3.2 (note: we assume $\Lambda_1$ to correspond to $m=1$, i.e., to the unscaled ensemble).

• $irest = 2$: the scaling factors are read from an auxiliary file called ``SGE.set'' that must be present in the directory from which the program is launched using the mpiexec/mpirun command. This ASCII file has two comment lines on the top and then as many lines as the number of ensembles ($nstates$). On each line there are four numbers in the following order: (1) the number of the ensemble; (2) the scaling factor for bonding and bending potential related to that ensemble; (3) the scaling factor for torsions and 1-4 interaction potentials related to that ensemble; (4) the related scaling factor for non-bonding potential related to that ensemble.

SGE simulations in the space of collective coordinates.
If the parameters $scale_1$, $scale_2$ and $scale_3$ are not specified in the SETUP command, then a SGE simulation in the space of collective coordinates is performed. In such a case the SETUP command is used to define the number of ensembles ($nstates$) and the restart option ($irest$). Their meaning has been explained above. The collective coordinates are defined using the ADD_STR_BONDS (bond coordinates), ADD_STR_BENDS (bending coordinates) and ADD_STR_TORS (torsional coordinates). These commands are defined in the &POTENTIAL environment and must be used in the following form
ADD_STR_BONDS $iat1$ $iat2$ $k_s$ $r_i$ $r_f$
ADD_STR_BENDS $iat1$ $iat2$ $iat3$ $k_b$ $\alpha_i$ $\alpha_f$
ADD_STR_TORS $iat1$ $iat2$ $iat3$ $iat4$ $k_t$ $\theta_i$ $\theta_f$
These expressions define the additional harmonic potential entering into Eq. 6.24. For example, if we perform a SGE simulation in the space of a distance between two atoms, then ADD_STR_BONDS must be used. The parameters $iat1$ and $iat2$ are the atom numbers, $k_s$ corresponds to $k$ of Eq. 6.24 and $r_i$ and $r_f$ define the intermediate ensembles as follows: $\lambda_n = r_i + ( n - 1 ) (r_f - r_i) / (nstates-1)$, where $\lambda_n$ is the parameter characteristic of the ensemble $n$ with $n = 1, 2, \dots, nstates$ (see Eq. 6.24).

EXAMPLES
SETUP 5 1. 1. 0.6 1
A Hamiltonian SGE simulation is performed. The non bonded potential (direct part) is scaled using a geometric progression, while the other potential terms are unscaled. The number of ensembles is 5.
SETUP 4 1
ADD_STR_BONDS 22 143 1. 10. 14.5
ADD_STR_BENDS 25 33 67 2. 100. 130.
A SGE simulation in the space of collective coordinates is performed using 4 ensembles. The collective coordinates are one bond and one bending. The bond is related to the atoms 22 and 143. The bending is defined by the atoms 25, 33 and 67. The ensembles are defined by 2 parameters, $\Lambda_n = (\lambda_n^{bond}, \lambda_n^{bend})$, where the bond related parameters are $\lambda^{bond}_1 = 10$, $\lambda^{bond}_2 = 11.5$, $\lambda^{bond}_3 = 13$, $\lambda^{bond}_4 = 14.5$ (in Å) and the bending related parameters is $\lambda^{bend}_1 = 100$, $\lambda^{bend}_2 = 110$, $\lambda^{bend}_3 = 120$, $\lambda^{bend}_4 = 130$ (in degrees). Therefore the transition of a replica from the ensemble $\Lambda_n$ to the ensemble $\Lambda_{n+1}$ involves a synchronous change of both parameters, i.e. $\lambda_n^{bond} \rightarrow \lambda_{n+1}^{bond}$ and $\lambda_n^{bend} \rightarrow \lambda_{n+1}^{bend}$. Finally, the harmonic force constants (see Eq. 6.24) are 1 and 2 kcal mol$^{-1}$ for bond and bending, respectively.