.de TQ
.  br
.  ns
.  TP \\$1
..
.TH GVMAP 1 "3 March 2011"
.SH NAME
gvmap \- find clusters and create a geographical map highlighting clusters.
.SH SYNOPSIS
.B gvmap
[
.I options
]
[
.BI \-o
.I outfile
]
[ 
.I files
]
.SH DESCRIPTION
.B gvmap
takes as input a graph in DOT format, finds node clusters and produces a rendering of the graph as a geographic-style map, with clusters highlighted
as countries, in xdot format.
.P
In the input graph, each node must have position, width and height information (pos, width and height attributes, 
respectively) defined, and nodes must not overlap.
.P
By default, 
.B gvmap
will generate the clusters from the data. If desired, the input graph can specify
cluster information by giving every node a \fIcluster\fP attribute whose value is
a small positive integer. (It is works best if \fIcluster\fP values are all integers
in the interval [1,K] for some K. Nodes sharing the same \fIcluster\fP attribute value will 
be put into the same cluster. \fBN.B.\fP For the \fIcluster\fP attribute to be used,
all nodes must have a valid value.
.P
If the \fI\-D\fP flag is used, 
.B  gvmap
will use the top-level cluster subgraphs to determine the clustering. Any nodes not in
such a cluster will be put into a single catch-all cluster.
.P
If the input specifies the desired clustering as described above, it can also
specify a desired coloring by having some node in each cluster provide a
\fIclustercolor\fP attribute. \fBN.B.\fP Unless one specifies \fI\-c0\fP, only the \fIclustercolor\fP
of the last node in a cluster has an effect. In addition, unless one uses \fI\-O\fP, \fBgvmap\fP may
permute the given colors.
.SH OPTIONS
The following options are supported:
.TP
.BI \-a " k"
The integer k specifies the average number of artificial points added along 
the bounding box of the labels. Such artificial points are added to avoid 
a country boundary cutting through the boundary box of the labels. Computing 
time is proportional to k; hence, for large graphs, a small value of k is 
suggested. If k = -1, a suitable value of k is automatically selected based on 
the graph size. By default k = -1.
.TP
.BI \-b " v"
The real number v specifies the line width used to draw the polygon boundaries, with v < 0 for no line. By default v = 0.
.TP
.BI \-c " k"
The integer k specifies color scheme used to color the countries. By default k = 1.
.TP
   Acceptable values are:
   0 : no polygons
   1 : pastel
   2 : blue to yellow
   3 : white to red
   4 : light grey to red
   5 : primary colors
   6 : sequential single hue red
   7 : sequential single hue lighter red
   8 : light grey
.TP
.BI \-c_opacity= xy
Specifies a two-character hexadecimal string specifying the opacity of the polygons.
.TP
.BI \-C " d"
The integer d specifies the maximum number of clusters (countries) allowed. By default d = 0, which means that there is no limit.
.TP
.BI \-d " d"
The integer d specifies the random seed used during color assignment optimization that maximize color difference between neighboring countries.
.TP
.BI \-D
If specified, the graph's cluster subgraphs are used to specify the clustering.
.TP
.BI \-e
If specified, edges will be included in the final output.
.TP
.BI \-g " c"
Specifies the bounding box color. If not specified, a bounding box is not drawn. 
.TP
.BI \-h " k"
The number of artificial points added to maintain a bridge between endpoints. By default, this is zero.
.TP
.BI \-highlight= k
Only draw cluster \fIk\fP. By default, all clusters are drawn.
.TP
.BI \-k 
If specified, increases the randomness of outer boundary.
.TP
.BI \-l " s"
Use the string \fIs\fP as a label for the drawing.
.TP
.BI \-m " v"
Generate a margin of \fIv\fP points around the drawing. By default, this is determined by \fBgvmap\fP.
.TP
.BI \-O  
Do NOT do color assignment optimization that maximizes color differences between neighboring countries
.TP
.BI \-o <file> 
Put output in <file>. Default output is stdout
.TP
.BI \-p " k"
Indicates what level of points should be shown. By default, no points are shown.
.TP
   Acceptable values are:
   0 : no points
   1 : all points
   2 : label points
   3 : random/artificial points
.TP
.BI \-Q
Use modularity quality for clustering rather than the default modularity clustering.
.TP
.BI \-r " k"
The number of random points k (integer) used to define sea and lake boundaries. If 0, auto assigned. By default v = 0
.TP
.BI \-s " v"
The real number v specifies the depth of the sea and lake shores in points. If 0, auto assigned. By default v = 0.
.TP
.BI \-t " n"
Make \fIn\fP attempts to improve cluster contiguity.
.TP
.BI \-v 
Verbose mode.
.TP
.BI \-z " c" 
Specified the polygon line color. Default is black.
.TP
.BI \-? 
Print usage and exit.
.SH EXAMPLES
.PP
Given a graph foo.gv, one way to generate a layout and highlight the clusters 
is to first select a layout engine with a suitable overlap removal method, then 
feed the output to gvmap, and finally render the map using specific graphics 
format. For example, the following pipeline
creates a map with edges in semi-transparent light gray and nodes laid 
out using sfdp:
.PP
sfdp \-Goverlap=prism foo.gv | gvmap \-e | neato \-n2 \-Ecolor=#55555522 \-Tpng > foo.png
.PP
The shell script gvmap.sh provides a shorthand for such pipelines. For example, the
above pipeline can be achieved using
.PP
gvmap.sh \-Ae \-Ecolor=#55555522 \-Tpng foo.gv > foo.png


.SH AUTHOR
Yifan Hu <yifanhu@yahoo.com>
.SH "SEE ALSO"
.PP
gvmap.sh(1), sfdp(1), neato(1), gvpr(1)
.PP
E. R. Gansner, Y. Hu, S. G. Kobourov, "GMap: Visualizing graphs and clusters as maps," Proc. Pacific Vis. 2010, pp. 201\(hy208.