Added documentation for ReweightMetad and peformance for multicolvar

[makedoc]

src/bias/ReweightMetad.cpp

@@ -24,9 +24,42 @@
 
 //+PLUMEDOC REWEIGHTING REWEIGHT_METAD
 /*
+Calculate the weights configurations should contribute to the histogram in a simulation in which a metadynamics bias acts upon the system.
+
+This command allows you to use the reweighting algorithm discussed in \cite PratyushReweighting when constructing a histogram
+of the configurations visited during a metadynamics simulation.
 
 \par Examples
 
+In the following example there is a metadynamics bias acting on the distance between atoms 1 and 2.  Clearly, this 
+bias will have an effect on the region of phase space that will be sampled when an MD simulation is
+run using this variable.  Consequently, when the histogram as a function of the angle, \f$a\f$, is accumulated,
+we use reweighting into order to discount the effect of the bias from our final histogram.  We do not use 
+\ref REWEIGHT_BIAS here, however, as the bias changes with time.  We thus use the reweighting algorithm for 
+metadynamics instead.  Notice also that we have to specify how often we would like to calculate the c(t) reweighting
+factor and the grid over which we calculate c(t) in the input to the METAD command.
+
+\verbatim
+a: ANGLE ATOMS=1,2,3
+x: DISTANCE ATOMS=1,2
+METAD ARG=x PACE=100 SIGMA=0.1 HEIGHT=1.5 BIASFACTOR=5 GRID_MIN=0 GRID_MAX=10 GRID_BIN=100 REWEIGHTING_NGRID=100 REWEIGHTING_NHILLS=50 
+
+as: REWEIGHT_METAD TEMP=300
+
+HISTOGRAM ...
+  ARG=a
+  GRID_MIN=0.0
+  GRID_MAX=pi
+  GRID_BIN=100
+  BANDWIDTH=0.1
+  LOGWEIGHTS=bias
+  LABEL=hB
+... HISTOGRAM
+
+DUMPGRID GRID=hB FILE=histoB STRIDE=1 FMT=%8.4f
+
+\endverbatim
+
 */
 //+ENDPLUMEDOC
 

user-doc/Performances.txt

@@ -114,8 +114,17 @@ See also \subpage EFFECTIVE_ENERGY_DRIFT.
 
 \page Multicolvar Multicolvar
 
-TO DO
+Whenever you have a multicolvar action such as:
 
+\verbatim
+COORDINATIONNUMBER SPECIES=1-100 SWITCH={RATIONAL R_0=1. D_MAX=3.0} MORE_THAN={RATIONAL R_0=6.0 NN=6 MM=12 D_0=0}
+\endvarbatim
+
+You will get a collosal speedup by specifying the D_MAX keyword in all switching functions that act on distances.
+D_MAX tells PLUMED that the switching function is strictly zero if the distance is greater than this value.  As a result
+PLUMED knows that it does not need to calculate these zero terms in what are essentially sums with a very lage number of terms.
+In fact when D_MAX is set PLUMED uses linked lists when calculating these coordination numbers, which is what 
+gives you such a dramatic increase in performance.
 
 \page Neighbour Neighbour Lists
 

user-doc/bibliography.bib

@@ -2270,3 +2270,26 @@ year = {2015}
   year={2015},
   publisher={Oxford University Press}
 }
+
+@article{PratyushReweighting,
+author = {Pratyush Tiwary and Michele Parrinello},
+title = {A Time-Independent Free Energy Estimator for Metadynamics},
+journal = {The Journal of Physical Chemistry B},
+volume = {119}, 
+number = {3},
+pages = {736-742},
+year = {2015},
+doi = {10.1021/jp504920s},
+    note ={PMID: 25046020},
+        
+URL = { 
+        http://dx.doi.org/10.1021/jp504920s
+        
+},      
+eprint = {
+        http://dx.doi.org/10.1021/jp504920s
+
+}
+
+}
+