dD Range and Segment Trees

*Gabriele Neyer*

This chapter presents the CGAL range tree and segment tree data structures.

A one-dimensional segment tree is a binary search tree as well, but with
*one-dimensional interval data* as input data.
One-dimensional interval data is a pair (i.e., 2-tuple) $$*(a,b)*, where $$*a*
and $$*b* are one-dimensional point data of the same type and $$*a< b*.
The pair $$*(a,b)* represents a half open interval $$*[a,b)*.
Analogously, a $$*d*-dimensional interval is represented by a $$*d*-tuple of
one-dimensional intervals.

The *input data type* for a $$*d*-dimensional tree is a container
class consisting of a $$*d*-dimensional point data type, interval data type
or a mixture of both, and optionally a *value type*, which
can be used to store arbitrary data.
E.g., the $$*d*-dimensional bounding box of a $$*d*-dimensional polygon
may define the interval data of a $$*d*-dimensional segment tree and
the polygon itself can be stored as its value.
An *input data item* is an instance of an input data type.

The range and segment tree classes are fully generic in the sense that they
can be used to define *multilayer trees*.
A multilayer tree of dimension (number of layers) $$*d* is a simple tree in
the $$*d*-th layer, whereas the $$*k*-th layer, $$*1 k d-1*, of the tree
defines a tree where each (inner) vertex contains a multilayer tree of
dimension $$*d-k+1*.
The $$*k-1*-dimensional tree which is nested in the $$*k*-dimensional tree
($$*T*) is called the *sublayer tree* (of $$*T*).
For example, a $$*d*-dim tree can be a range tree on the first layer,
constructed with respect to the first dimension of $$*d*-dimensional data
items.
On all the data items in each subtree, a $$*(d-1)*-dimensional tree is built,
either a range or a segment tree, with respect to the second dimension of
the data items.
And so on.
The figures in Sections 50.5 and 50.6
illustrate the means of a sublayer tree graphically.

After creation of the tree, further insertions or deletions of data items
are disallowed.
The tree class does neither depend on the type of data nor on the concrete
physical representation of the data items.
E.g., let a multilayer tree be a segment tree for which each vertex
defines a range tree.
We can choose the data items to consist of intervals of type *double*
and the point data of type *integer*.
As value type we can choose *string*.

For this generality we have to
define what the tree of each dimension looks like and how the
input data is organized.
For dimension $$*k*, $$*1 k 4*, CGAL provides ready-to-use
range and segment trees that can store k-dimensional keys
(intervals resp.).
Examples illustrating the use of these classes are given in
Sections 50.5.1
and 50.6.1.
The description of the functionality of these classes as well as
the definition of higher dimensional trees and mixed multilayer
trees is given in the reference manual.

In the following two sections we give short definitions of the version of the range tree and segment tree implemented here together with some examples. The presentation closely follows [dBvKOS97].

In order to be able to define a multilayer tree we first
designed the range and segment tree to have a template argument
defining the type of the sublayer tree. With this sublayer tree
type information the sublayers could be created. This approach lead to nested
template arguments, since the sublayer tree can again have a template
argument defining the sublayer. Therefore, the internal class and function
identifiers got longer than a compiler-dependent limit.
This happend already for $$*d=2*.

Therefore, we chose another, object oriented,
design. We defined a pure
virtual base class called *Tree_base* from which we derived
the classes *Range_tree_d* and *Segment_tree_d*.
The constructor of these classes expects an argument called
*sublayer_prototype* of type *Tree_base*.
Since class *Range_tree_d* and class
*Segment_tree_d* are derived from class *Tree_base*, one can
use an instantiation of class *Range_tree_d* or class
*Segment_tree_d* as constructor argument.
This argument defines the sublayer tree of the tree. E.g., you
can construct a *Range_tree_d* with an
instantiation of class *Segment_tree_d* as constructor
argument. You then have defined a range tree with a segment tree
as sublayer tree. Since both classes *Range_tree_d* and
*Segment_tree_d* expect a sublayer tree in their constructor
we had to derive a third class called *Tree_anchor* from
class *Tree_base* which
does not expect a constructor argument. An instantiation of this
class is used as constructor argument of class *Range_tree_d* or
*Segment_tree_d* in order to stop the recursion.

All classes provide a *clone()* function which returns an
instance (a copy) of the same tree type. The *clone()*
function of the *sublayer_prototype* is called in the
construction of the tree. In case that the sublayer tree again
has a sublayer, it also has a *sublayer_prototype* which is
also cloned and so on. Thus, a call to the *clone()* function
generates a sublayer tree which has the complete knowledge about
its sublayer tree.

The trees allow to perform window queries, enclosing queries, and inverse range queries on the keys. Clearly, an inverse range query makes only sense in the segment tree. In order to perform an inverse range query, a range query of $$ width has to be performed. We prefered not to offer an extra function for this sort of query, since the inverse range query is a special case of the range query. Furthermore, offering an inverse range query in the segment tree class implies offering this function also in the range tree class and having an extra item in the traits class that accesses the inverse range query point.

The trees are templatized with three arguments: *Data, Window*
and *Traits*. Type *Data* defines
the input data type and type *Window* defines the query
window type. The tree uses a well defined set of functions in
order to access data. These functions have to be provided by
class *Traits*.

The design partly follows the *prototype design pattern*
in [GHJV95]. In comparison to our first approach
using templates we want to note the following: In this approach
the sublayer type is defined in
use of object oriented programming at run time, while in the
approach using templates, the sublayer type is defined at compile
time.

The runtime overhead caused in use of virtual member functions in this object oriented design is negligible since all virtual functions are non trivial. The design concept is illustrated in the figure below.

E.g. in order to define a two dimensional multilayer tree, which
consists of a range tree in the first dimension and a segment
tree in the second dimension we proceed as follows: We construct
an object of type *Tree_anchor* which stops the
recursion. Then we construct an object of type *Segment_tree_d*,
which gets as prototype argument our object of type
*Tree_anchor*. After that, we define an object of type
*Range_tree_d* which is constructed with the object of type
*Segment_tree_d* as prototype argument.
The following piece of code illustrates
the construction of the two-dimensional multilayer tree.

int main(){ Tree_Anchor *anchor=new Tree_Anchor; Segment_Tree_d *segment_tree = new Segment_Tree_d(*anchor); Range_Tree_d *range_segment_tree = new Range_Tree_d(*segment_tree); /* let data_items be a list of Data items */ range_segment_tree->make_tree(data_items.begin(),data_items.end()); }

Here, class *Tree_Anchor, Segment_Tree_d*, and
*Range_Tree_d* are defined by *typedef*s:

typedef Tree_anchor<Data,Window> Tree_Anchor; typedef Segment_tree_d<Data,Window,Interval_traits> Segment_Tree_d; typedef Range_tree_d<Data,Window,Point_traits> Range_Tree_d;

Class *Tree_base* and class
*Tree_anchor* get two template arguments: a class
*Data* which defines the type of data that is stored in
the tree, and a class *Window* which defines the type of a query
range.
The derived classes *Range_tree_d* and *Segment_tree_d*
additionally get an argument called
*Tree_traits* which defines the interface between the
*Data* and the tree. Let the *Data* type be a $$*d*-dimensional
tuple, which is either a point data or an interval data in each
dimension. Then, the class *Tree_traits* provides accessors to
the point (resp. interval) data of that tree layer and a compare
function. Remind our example of the two-dimensional tree which
is a range tree in the first dimension and
a segment tree in the second dimension. Then, the
*Tree_traits* class template argument of class
*Segment_tree_d* defines an accessor to the interval data of
the *Data*, and the
*Tree_traits* class template argument of class
*Range_tree_d* defines an accessor to the point data of
*Data*.
An example implementation for these classes is listed below.

struct Data{ int min,max; /* interval data */ double point; /* point data */ }; struct Window{ int min,max; double min_point, max_point; }; class Point_traits{ public: typedef double Key; Key get_key(Data& d){return d.point;} /*key accessor */ Key get_left(Window& w){return w.min_point;} Key get_right(Window& w){return w.max_point;} bool comp(Key& key1, Key& key2){return (key1 < key2);} } class Interval_traits{ public: typedef int Key; Key get_left(Data& d){return d.min;} Key get_right(Data& d){return d.max;} Key get_left_win(Window& w){return w.min;} Key get_right_win(Window& w){return w.max;} bool comp(Key& key1, Key& key2){return (key1 < key2);} }

Now let us have a closer look on how a multilayer tree is built.
In case of creating a $$*d*-dimensional tree, we handle a
sequence of arbitrary data
items, where each item defines a $$*d*-dimensional interval, point
or other object. The tree is constructed with an iterator over
this structure. In the $$*i*-th layer, the tree is
built with respect to the data slot that defines the $$*i*-th
dimension. Therefore, we need to define which data slot
corresponds to which dimension.
In addition we want our tree to work with arbitrary data items.
This requires an
adaptor between the algorithm and the data item. This is resolved
by the use of traits classes, implemented in
form of a traits class using
function objects.
These classes provide
access functions to a specified data slot of a data item.
A $$*d*-dimensional tree is then defined separately for each layer by
defining a traits class for each layer.

A one-dimensional range tree is a binary search tree on one-dimensional
point data.
The point data of the tree is stored in the leaves.
Each inner vertex stores the highest entry of its left subtree.
The version of a range tree implemented here is static, which means that
after construction of the tree, no elements be inserted or deleted.
A $$*d*-dimensional range tree is a binary leaf search tree according to the
first dimension of the $$*d*-dimensional point data, where each vertex contains
a $$*(d-1)*-dimensional search tree of the points in the subtree (sublayer tree)
with respect to the second dimension.
See [dBvKOS97] and [Sam90] for more detailed information.

A $$*d*-dimensional range tree can be used to determine all
$$*d*-dimensional points that lie inside a given $$*d*-dimensional
interval (*window_query*).
The pictures below show a two-dimensional and a $d$-dimensional
range tree.

A two-dimensional range tree. The tree is a binary search tree on the first dimension. Each sublayer tree of a vertex $v$ is a binary search tree on the second dimension. The data items in a sublayer tree of $v$ are all data items of the subtree of $v$ | A d-dimensional range tree. For each layer of the tree, one sublayer tree is illustrated. |

The tree can be built in $$*O(n*log$$* ^{d-1} n)* time and
needs $$

The following example program uses the predefined * Range_tree_2* data structure together with the predefined traits
class *Range_tree_map_traits_2* which has two template
arguments specifying the
type of the point data in each dimension
(*CGAL::Cartesian<double>*) and the value type of the
2-dimensional point data (*char*). Therefore the * Range_tree_2* is defined on 2-dimensional point data each of which is
associated with a character.
Then, a few data items are created and put into a list. After
that the tree is constructed according to that list, a window
query is performed, and the query elements are given out.

#include <CGAL/Cartesian.h> #include <CGAL/Range_segment_tree_traits.h> #include <CGAL/Range_tree_k.h> typedef CGAL::Cartesian<double> K; typedef CGAL::Range_tree_map_traits_2<K, char> Traits; typedef CGAL::Range_tree_2<Traits> Range_tree_2_type; int main() { typedef Traits::Key Key; typedef Traits::Interval Interval; std::vector<Key> InputList, OutputList; InputList.push_back(Key(K::Point_2(8,5.1), 'a')); InputList.push_back(Key(K::Point_2(1,1.1), 'b')); InputList.push_back(Key(K::Point_2(3,2.1), 'c')); Range_tree_2_type Range_tree_2(InputList.begin(),InputList.end()); Interval win(Interval(K::Point_2(4,8.1),K::Point_2(5,8.2))); std::cout << "\n Window Query:\n "; Range_tree_2.window_query(win, std::back_inserter(OutputList)); std::vector<Key>::iterator current=OutputList.begin(); while(current!=OutputList.end()){ std::cout << (*current).first.x() << "," << (*current).first.y() << ":" << (*current++).second << std::endl; } }

This example illustrates the use of the range tree on 2-dimensional point data (no value is associated to a data item). After the definition of the tree, some input data items are created and the tree is constructed according to the input data items. After that, a window query is performed and the query elements are given to standard out.

#include <CGAL/Cartesian.h> #include <CGAL/Range_segment_tree_traits.h> #include <CGAL/Range_tree_k.h> typedef CGAL::Cartesian<double> K; typedef CGAL::Range_segment_tree_set_traits_2<K> Traits; typedef CGAL::Range_tree_2<Traits> Range_tree_2_type; int main() { typedef Traits::Key Key; typedef Traits::Interval Interval; std::vector<Key> InputList, OutputList; std::vector<Key>::iterator first, last, current; InputList.push_back(Key(8,5.1)); InputList.push_back(Key(1,1.1)); InputList.push_back(Key(3,2.1)); Range_tree_2_type Range_tree_2(InputList.begin(),InputList.end()); Interval win=Interval(Key(4,8.1),Key(5,8.2)); std::cout << std::endl << "Window Query: lower left point: (4.0,5.0),"; std::cout << "upper right point: (8.1,8.2)" << std::endl; Range_tree_2.window_query(win, std::back_inserter(OutputList)); current=OutputList.begin(); while(current!=OutputList.end()){ std::cout << (*current).x()<< "-" << (*current).y() << std::endl; current++; } }

A segment tree is a static binary search tree for a given set of
coordinates. The set of coordinates is defined by the endpoints
of the input data intervals. Any two adjacent coordinates
build an elementary interval. Every leaf corresponds to an
elementary interval.
Inner vertices
correspond to the union of the subtree intervals of the vertex.
Each vertex or leaf $$*v* contains a sublayer type (or a
list, if it is one-dimensional) that will contain all intervals $$*I*, such that
$$*I* contains the interval of vertex $$*v* but not the interval
of the parent vertex of $$*v*.

A $$*d*-dimensional segment tree can be used to solve the following problems:

- Determine all $$
*d*-dimensional intervals that contain a $$*d*-dimensional point. This query type is called ``inverse range query''. - Determine all $$
*d*-dimensional intervals that enclose a given $$*d*-dimensional interval (*enclosing_query*). - Determine all $$
*d*-dimensional intervals that partially overlap or are contained in a given $$*d*-dimensional interval (*window_query*).

An example of a one-dimensional segment tree and an example of a two-dimensional segment tree are shown below.

A one-dimensional segment tree. The segments and the corresponding elementary intervals are shown below the tree. The arcs from the nodes point to their subsets. | A two-dimensional segment tree. The first layer of the tree is built according to the elementary intervals of the first dimension. Each sublayer tree of a vertex $v$ is a segment tree according to the second dimension of all data items of $v$. |

The tree can be built in $$*O(n*log$$* ^{d} n)* time and
needs $$

One possible application of a two-dimensional segment tree is the following. Given a set of convex polygons in two-dimensional space (CGAL::Polygon_2), we want to determine all polygons that intersect a given rectangular query window. Therefore, we define a two-dimensional segment tree, where the two-dimensional interval of a data item corresponds to the bounding box of a polygon and the value type corresponds to the polygon itself. The segment tree is created with a sequence of all data items, and a window query is performed. The polygons of the resulting data items are finally tested independently for intersections.

The following example program uses the predefined * Segment_tree_2* data structure together with the predefined traits
class *Segment_tree_map_traits_2* which has two template arguments
specifying the
type of the point data in each dimension
(*CGAL::Cartesian<double>*) and the value type of the
2-dimensional point data (*char*). Therefore the * Segment_tree_2* is defined on 2-dimensional point data
(*CGAL::Point_2<Cartesian<double> >*) each of which is
associated with a character.
Then, a few data items are created and put into a list. After
that the tree is constructed according to that list, a window
query is performed, and the query elements are given out.

#include <CGAL/Cartesian.h> #include <CGAL/Segment_tree_k.h> #include <CGAL/Range_segment_tree_traits.h> typedef CGAL::Cartesian<double> K; typedef CGAL::Segment_tree_map_traits_2<K, char> Traits; typedef CGAL::Segment_tree_2<Traits > Segment_tree_2_type; int main() { typedef Traits::Interval Interval; typedef Traits::Pure_interval Pure_interval; typedef Traits::Key Key; std::list<Interval> InputList, OutputList1, OutputList2; InputList.push_back(Interval(Pure_interval(Key(1,5), Key(2,7)),'a')); InputList.push_back(Interval(Pure_interval(Key(2,7), Key(3,8)),'b')); InputList.push_back(Interval(Pure_interval(Key(6,9), Key(9,13)),'c')); InputList.push_back(Interval(Pure_interval(Key(1,3), Key(3,9)),'d')); Segment_tree_2_type Segment_tree_2(InputList.begin(),InputList.end()); Interval a=Interval(Pure_interval(Key(3,6), Key(7,12)),'e'); Segment_tree_2.window_query(a,std::back_inserter(OutputList1)); std::list<Interval>::iterator j = OutputList1.begin(); std::cout << "\n window_query (3,6),(7,12)\n"; while(j!=OutputList1.end()){ std::cout << (*j).first.first.x() << "-" << (*j).first.second.x() << " " << (*j).first.first.y() << "-" << (*j).first.second.y() << std::endl; j++; } Interval b=Interval(Pure_interval(Key(6,10),Key(7,11)), 'f'); Segment_tree_2.enclosing_query(b,std::back_inserter(OutputList2)); j = OutputList2.begin(); std::cout << "\n enclosing_query (6,10),(7,11)\n"; while(j!=OutputList2.end()){ std::cout << (*j).first.first.x() << "-" << (*j).first.second.x() << " " << (*j).first.first.y() << "-" << (*j).first.second.y() << std::endl; j++; } return 0; }

This example illustrates the use of the predefined segment tree on 3-dimensional interval data (with no value associated). After the definition of the traits type and tree type, some intervals are constructed and the tree is build according to the intervals. Then, a window query is performed and the query elements are given out.

#include <CGAL/Cartesian.h> #include <CGAL/Segment_tree_k.h> #include <CGAL/Range_segment_tree_traits.h> typedef CGAL::Cartesian<int> K; typedef CGAL::Range_segment_tree_set_traits_3<K> Traits; typedef CGAL::Segment_tree_3<Traits > Segment_tree_3_type; int main() { typedef Traits::Interval Interval; typedef Traits::Key Key; std::list<Interval> InputList, OutputList; InputList.push_back(Interval(Key(1,5,7), Key(2,7,9))); InputList.push_back(Interval(Key(2,7,6), Key(3,8,9))); InputList.push_back(Interval(Key(6,9,5), Key(9,13,8))); InputList.push_back(Interval(Key(1,3,4), Key(3,9,8))); Segment_tree_3_type Segment_tree_3(InputList.begin(),InputList.end()); Interval a(Key(3,6,5), Key(7,12,8)); Segment_tree_3.window_query(a,std::back_inserter(OutputList)); std::list<Interval>::iterator j = OutputList1.begin(); std::cout << "\n window_query (3,6,5),(7,12,8) \n"; while(j!=OutputList.end()){ std::cout << (*j).first.x() << "," << (*j).first.y() << ","; std::cout << (*j).first.z() <<", " << (*j).second.x() << ","; std::cout << (*j).second.y() << "," << (*j).second.z() << std::endl; j++; } }