Remove SD

doc/sphinx/features.rst

@@ -138,16 +138,6 @@ integrator or thermostat:
 -  Enables the implicit calculation of electro-hydrodynamics for charged
    particles and salt ions in an electric field.
 
--  enable Stokesian Dynamics.
-
--  disable periodic boundary conditions in Stokesian Dynamics.
-
--  use float instead of double in Stokesian Dynamics.
-
--  disable nearfield of Stokesian Dynamics.
-
--  enable debug Stokesian Dynamics.
-
 Interactions
 ------------
 
@@ -279,6 +269,3 @@ looking directly at the code.
 -  Causes to try to provoke a core dump when exiting unexpectedly.
 
 -  Causes to try this even with MPI errors.
-
--  Causes to check more things in the Stokesian Dynamics code. If is
-   larger 1, SD prints more informations.

doc/sphinx/introduction.rst

@@ -183,8 +183,6 @@ report so to the developers.
 +--------------------------------+------------------------+------------------+
 | Tunable Slip Boundary          | Single                 | Single           |
 +--------------------------------+------------------------+------------------+
-| Stokesian Dynamics             | Single                 | Single           |
-+--------------------------------+------------------------+------------------+
 |                             **Analysis**                                   |
 +--------------------------------+------------------------+------------------+
 | uwerr                          | None                   | Good             |

doc/sphinx/setup.rst

@@ -132,19 +132,6 @@ re-used.
 (int) if non-zero (default), some warnings are printed out. Set this to
 zero if you get annoyed by them.
 
-(double) viscosity of the fluid for the Stokesian Dynamics simulation.
-
-(double) hydrodynamic radius of the particles used in Stokesian
-Dynamics.
-
-(int[2]) seed of the Stokes Dynamics random number generator.
-
-(int[2]) offset of the random number generator. Together with the seed,
-the state of the random number generator is well defined.
-
-(double) precision used for the approximation of the square root of the
-mobility. Sometimes higher accuracy can speedup the simulation.
-
 Setting global variables in Python
 ----------------------------------
 
@@ -508,17 +495,6 @@ unit operating at a constant temperature.
 
 Be aware that this thermostat requires to be given in Kelvin.
 
-Stokesian Dynamics thermostat
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-.. todo::
-    This is not implemented yet for the python interface
-
-This thermostat should be used together with the Stokesian Dynamics
-implementation. No other thermostat is able to thermalize SD correctly.
-The precision of the farfield contribution of the thermostat can be
-tuned with
-
 .. _nemd:
 
 ``nemd``: Setting up non-equilibrium MD

doc/ug/features.tex

@@ -146,13 +146,6 @@ \section{General features}
 \item \newfeature{LB_ELECTROHYDRODYNAMICS} Enables the implicit
   calculation of electro-hydrodynamics for charged particles and salt
   ions in an electric field.
-\item \newfeature{SD} enable Stokesian Dynamics.
-\item \newfeature{SD_NOT_PERIODIC} disable periodic boundary conditions in
-  Stokesian Dynamics.
-\item \newfeature{SD_USE_FLOAT} use float instead of double in Stokesian
-  Dynamics.
-\item \newfeature{SD_FF_ONLY} disable nearfield of Stokesian Dynamics.
-\item \newfeature{SD_DEBUG} enable debug Stokesian Dynamics.
 
 \end{itemize}
 
@@ -264,8 +257,6 @@ \section{Debug messages}
 \item \newfeature{FORCE\_CORE} Causes \es{} to try to provoke a core dump
   when exiting unexpectedly.
 \item \newfeature{MPI\_CORE} Causes \es{} to try this even with MPI errors.
-\item \newfeature{SD_DEBUG} Causes \es{} to check more things in the Stokesian
-  Dynamics code. If \var{warnings} is larger 1, SD prints more information.
 \end{itemize}
 
 %%% Local Variables: 

doc/ug/introduction.tex

@@ -166,7 +166,6 @@ \section{Available simulation methods}
   DPD                                 & None   & Good   \\
   Shan-Chen Multicomponent Fluid      & Group  & Group  \\
   Tunable Slip Boundary               & Single & Single \\
-  Stokesian Dynamics                  & Single & Single \\
   \multicolumn{3}{|c|}{\textbf{Analysis}} \\
   uwerr                               & None   & Good   \\
   \multicolumn{3}{|c|}{\textbf{Input/Output}} \\

doc/ug/run.tex

@@ -588,78 +588,6 @@ \subsubsection{Details on implementation}
 codes) but so-called Lucy functions. See \cite{marsili09} for more
 details. These avoid the calculation of exponentials.
 
-\section{\texttt{integrate_sd}: Running a stokesian dynamics simulation}
-\newescommand[integrate-sd]{integrate_sd}
-
-\begin{essyntax}
-  \variant{1} integrate_sd \var{steps}
-\end{essyntax}
-
-\es uses in the stokesian dynamics algorithm  the euler integrater for the
-equations of motion. The motion are overdamped. The velocities of the
-particles are related to the displacement in the last timestep. This are, at
-least if the system is thermalized, however not usefull, as the displacements
-don't scale linear with the timestep. The command \texttt{integrate_sd} with
-an integer \var{steps} as parameter integrates the system for
-\var{steps} time steps. This is implemented using CUDA, so \required{CUDA}
-has to be available in the system.
-
-Currently there is no parallel implementation of this integrator, so all
-particles have to be in a single process.
-
-\subsection{Setting up the system}
-Before running a stokesian dynamics simulation, you have to set a view
-constants:
-\begin{globvar}
-\item[sd_radius] The hydrodynamic particle radius of the (sperical)
-  particles. Only one particle size is supported.
-\item[sd_viscosity] The viscisity of the fluid. Remember this is only a
-  scaling of the timestep, so you can set it without problems to one.
-\item[sd_seed] (int[2]) seed of the Stokes Dynamics random number generator.
-  As the generator used for the SD runes on the GPU (cuRAND) it has its own
-  seed and own state.
-\item[sd_random_state] (int[2]) offset of the random number
-  generator. Together with the seed, the state of the random number generator
-  is well defined.
-\item[sd_precision_random] (double) precision used for the approximation of
-  the square root of the mobility. Sometimes higher accuracy can speedup the
-  simulation.
-\end{globvar}
-
-Make sure to only use the stokesian dynamics thermostat. If you made a warmup
-integration without SD, set it with:
-\begin{tclcode}
-thermostat off
-thermostat sd \$temp
-\end{tclcode}
-
-\subsection{Periodicity}
-The Code uses the ewald-summation (as derived in \cite{beenakker86a}) of the
-Rotne-Prager tensor to be usefull in periodic boundary conditions. To disable
-the peridoicity of the hydrodynamic, use the feature
-\feature{SD_NOT_PERIODIC}. 
-
-The ewald summation is required if the system is periodic, as otherwise an
-cutoff is introduced, which can lead to negative eigenvalues of the
-mobility. This leads to a break down of the code, as it is not possible to
-thermalize such a system.
-
-\subsection{Floatingpoint precision}
-It is possible to switch between double and float as datatype in Stokesian
-Dynamics. As GPUs have more single than double precision floating point
-units, the single precision is faster. However this can lead to problems, if
-the eigenvalues are not precise enough and get negative. Therfore the default
-is double precision. If you want to use single precision, use the feature
-\feature{SD_USE_FLOAT}.
-
-\subsection{Farfield only}
-The mobility matrix consists of two contribution, the farfield and the
-nearfield lubrication correction.
-If the nearfield is neglegible, than the feature \feature{SD_FF_ONLY} can be
-used.
-This should speedup the simulation, as the nearfield doesn't need to be
-inverted. Additional the thermal displacements can be directly calculated.
-
 \section{Multi-timestepping}
 \newescommand[multitimestepping]{multitimestepping}
 

doc/ug/setup.tex

@@ -151,17 +151,6 @@ \section{\texttt{setmd}: Setting global variables in TCL}
   verlet list has been re-used.
 \item[warnings] (int) if non-zero (default), some warnings are printed
   out. Set this to zero if you get annoyed by them.
-\item[sd_viscosity] (double) viscosity of the fluid for the Stokesian Dynamics
-  simulation.
-\item[sd_radius] (double) hydrodynamic radius of the particles used in
-  Stokesian Dynamics.
-\item[sd_seed] (int[2]) seed of the Stokes Dynamics random number generator.
-\item[sd_random_state] (int[2]) offset of the random number
-  generator. Together with the seed, the state of the random number generator
-  is well defined.
-\item[sd_precision_random] (double) precision used for the approximation of
-  the square root of the mobility. Sometimes higher accuracy can speedup the
-  simulation.
 \end{globvar}
 
 
@@ -602,21 +591,6 @@ \subsection{CPU thermostat}
 Be aware that this thermostat requires \var{temperature} to be given
 in Kelvin.
 
-\subsection{Stokesian Dynamics thermostat}
-\label{ssec:SDthermostat}
-\begin{essyntax}
-  thermostat sd \var{temperature}
-  \begin{features}
-    \required{CUDA} and \required{SD}
-  \end{features}
-\end{essyntax}
-
-This thermostat should be used together with the Stokesian Dynamics
-implementation.
-No other thermostat is able to thermalize SD correctly.
-The precision of the farfield contribution of the thermostat can be tuned
-with
-
 \section{\texttt{nemd}: Setting up non-equilibrium MD}
 \newescommand{nemd}
 \label{sec:NEMD}

src/core/global.cpp

@@ -34,7 +34,6 @@
 #include "imd.hpp"
 #include "ghmc.hpp"
 #include "lb.hpp"
-#include "integrate_sd.hpp"
 
 /** This array contains the description of all global variables.
 
@@ -99,19 +98,14 @@ const Datafield fields[] = {
   {&lb_components,      TYPE_INT, 1, "lb_components",     2, false },         /* 51 from ghmc.cpp */
   {&warnings,           TYPE_INT, 1, "warnings",          1, false },         /* 52 from global.cpp */
   {&dpd_ignore_fixed_particles, TYPE_INT, 1, "dpd_ignore_fixed_particles", 1, false },         /* 53 from global.cpp */
-  {&sd_viscosity,            TYPE_DOUBLE, 1, "sd_viscosoity",              4, false },         /* 54 from integrate_sd.cpp */
-  {&sd_radius,               TYPE_DOUBLE, 1, "sd_radius",                  4, false },         /* 55 from integrate_sd.cpp */
-  {&sd_seed,                    TYPE_INT, 2, "sd_seed",                    4, false },         /* 56 from integrate_sd.cpp */
-  {&sd_random_state,            TYPE_INT, 2, "sd_random_state",            4, false },         /* 57 from integrate_sd.cpp */
-  {&sd_random_precision,     TYPE_DOUBLE, 1, "sd_precision_random",        4, false },         /* 58 from integrate_sd.cpp */
-  {&smaller_time_step,TYPE_DOUBLE,1, "smaller_time_step", 5, false },         /* 59 from integrate.cpp */
-  {configtemp,       TYPE_DOUBLE, 2, "configtemp",        1, false },         /* 60 from integrate.cpp */
-  {&langevin_trans,  TYPE_BOOL, 1, "langevin_trans_switch", 1, false },       /* 61 from thermostat.cpp */
-  {&langevin_rotate,  TYPE_BOOL, 1, "langevin_rotate_switch", 1, false },     /* 62 from thermostat.cpp */
+  {&smaller_time_step,TYPE_DOUBLE,1, "smaller_time_step", 5, false },         /* 54 from integrate.cpp */
+  {configtemp,       TYPE_DOUBLE, 2, "configtemp",        1, false },         /* 55 from integrate.cpp */
+  {&langevin_trans,  TYPE_BOOL, 1, "langevin_trans_switch", 1, false },       /* 56 from thermostat.cpp */
+  {&langevin_rotate,  TYPE_BOOL, 1, "langevin_rotate_switch", 1, false },     /* 57 from thermostat.cpp */
 #ifndef ROTATIONAL_INERTIA
-  {&langevin_gamma_rotation,  TYPE_DOUBLE, 1, "gamma_rot",1, false },    /* 63 from thermostat.cpp */
+  {&langevin_gamma_rotation,  TYPE_DOUBLE, 1, "gamma_rot",1, false },    /* 58 from thermostat.cpp */
 #else
-  {langevin_gamma_rotation,  TYPE_DOUBLE, 3, "gamma_rot",1, false },    /* 63 from thermostat.cpp */
+  {langevin_gamma_rotation,  TYPE_DOUBLE, 3, "gamma_rot",1, false },    /* 58 from thermostat.cpp */
 #endif
   { NULL, 0, 0, NULL, 0, false }
 };

src/core/global.hpp

@@ -185,26 +185,16 @@ extern const Datafield fields[];
 #define FIELD_WARNINGS            52
 /** DPD_IGNORE_FIXED_PARTICLES */
 #define FIELD_DPD_IGNORE_FIXED_PARTICLES 53
-/** index of \ref sd_viscosity in \ref #fields */
-#define FIELD_SD_VISCOSITY        54
-/** index of \ref sd_radius in \ref_#fields */
-#define FIELD_SD_RADIUS           55
-/** index of \ref sd_seed in \ref #fields */
-#define FIELD_SD_SEED             56
-/** index of \ref sd_random_state i_ \ref #fields */
-#define FIELD_SD_RANDOM_STATE     57
-/** index of \ref sd_precision_random in \ref #fields */
-#define FIELD_SD_RANDOM_PRECISION 58
 /** index of \ref smaller_timestep in \ref #fields */
-#define FIELD_SMALLERTIMESTEP     59
+#define FIELD_SMALLERTIMESTEP     54
 /** index of \ref configtemp in \ref #fields */
-#define FIELD_CONFIGTEMP          60
+#define FIELD_CONFIGTEMP          55
 /** index of \ref langevin_trans in \ref #fields */
-#define FIELD_LANGEVIN_TRANS_SWITCH 61
+#define FIELD_LANGEVIN_TRANS_SWITCH 56
 /** index of \ref langevin_rotate in \ref #fields */
-#define FIELD_LANGEVIN_ROT_SWITCH 62
+#define FIELD_LANGEVIN_ROT_SWITCH 57
 /** index of \ref langevin_gamma_rotation in  \ref #fields */
-#define FIELD_LANGEVIN_GAMMA_ROTATION 63
+#define FIELD_LANGEVIN_GAMMA_ROTATION 58
 
 /*@}*/
 

src/core/integrate.cpp

@@ -361,15 +361,6 @@ void integrate_vv(int n_steps, int reuse_forces) {
     }
 #endif
 
-#ifdef SD
-    if (thermo_switch & THERMO_SD) {
-      runtimeWarning("Use integrate_sd to use Stokesian Dynamics Thermalizer.");
-    }
-    if (thermo_switch & THERMO_BD) {
-      runtimeWarning("Use integrate_sd to use Brownian Dynamics Thermalizer.");
-    }
-#endif
-
     /* Integration Steps: Step 1 and 2 of Velocity Verlet scheme:
        v(t+0.5*dt) = v(t) + 0.5*dt * f(t)
        p(t + dt)   = p(t) + dt * v(t+0.5*dt)