X-Git-Url: http://biod.pnpi.spb.ru/gitweb/?a=blobdiff_plain;f=docs%2Fmanual%2Fforcefield.tex;h=c406e054a6e1bdfbf751395bf3e13cb428aded88;hb=513813b51406baae3a1b6835daf50017d0f9cb7f;hp=70829ad13e07568f4187669fcc171ec40c4bbb55;hpb=37573afb8aea3688b3bb2fe30919300efe198ac0;p=alexxy%2Fgromacs.git diff --git a/docs/manual/forcefield.tex b/docs/manual/forcefield.tex index 70829ad13e..c406e054a6 100644 --- a/docs/manual/forcefield.tex +++ b/docs/manual/forcefield.tex @@ -1810,27 +1810,42 @@ after taking the derivative, we {\em can} insert \ve{p} = m\ve{v}, such that: \subsubsection{Constraints} \label{subsubsec:constraints} -\newcommand{\clam}{C_{\lambda}} The constraints are formally part of the Hamiltonian, and therefore they give a contribution to the free energy. In {\gromacs} this can be calculated using the \normindex{LINCS} or the \normindex{SHAKE} algorithm. -If we have a number of constraint equations $g_k$: +If we have $k = 1 \ldots K$ constraint equations $g_k$ for LINCS, then \beq -g_k = \ve{r}_{k} - d_{k} +g_k = |\ve{r}_{k}| - d_{k} \eeq -where $\ve{r}_k$ is the distance vector between two particles and -$d_k$ is the constraint distance between the two particles, we can write -this using a $\LAM$-dependent distance as +where $\ve{r}_k$ is the displacement vector between two particles and +$d_k$ is the constraint distance between the two particles. We can express +the fact that the constraint distance has a $\LAM$ dependency by \beq -g_k = \ve{r}_{k} - \left(\LL d_{k}^A + \LAM d_k^B\right) +d_k = \LL d_{k}^A + \LAM d_k^B \eeq -the contribution $\clam$ -to the Hamiltonian using Lagrange multipliers $\lambda$: + +Thus the $\LAM$-dependent constraint equation is +\beq +g_k = |\ve{r}_{k}| - \left(\LL d_{k}^A + \LAM d_k^B\right). +\eeq + +The (zero) contribution $G$ to the Hamiltonian from the constraints +(using Lagrange multipliers $\lambda_k$, which are logically distinct +from the free-energy $\LAM$) is \bea -\clam &=& \sum_k \lambda_k g_k \\ -\dvdl{\clam} &=& \sum_k \lambda_k \left(d_k^B-d_k^A\right) +G &=& \sum^K_k \lambda_k g_k \\ +\dvdl{G} &=& \frac{\partial G}{\partial d_k} \dvdl{d_k} \\ + &=& - \sum^K_k \lambda_k \left(d_k^B-d_k^A\right) \eea +For SHAKE, the constraint equations are +\beq +g_k = \ve{r}_{k}^2 - d_{k}^2 +\eeq +with $d_k$ as before, so +\bea +\dvdl{G} &=& -2 \sum^K_k \lambda_k \left(d_k^B-d_k^A\right) +\eea \subsection{Soft-core interactions\index{soft-core interactions}} \begin{figure}