Information Technology Reference
In-Depth Information
s
ʷ
6
=
s
μ
belongs to the line with slope
g
2
passing through
s
ʷ
5
. We now rotate
ʓ
ʷ
1
so
that the segment
s
ʷ
1
t
ʷ
1
uses the green slope
g
2
and attach it to
ʓ
ʷ
5
and
ʓ
ʷ
6
. Finally, the
edge corresponding to
ʷ
4
is drawn as the segment
s
μ
s
ʷ
3
.Invariant
Ia.
holds because the
drawings
ʓ
ʷ
1
,
ʓ
ʷ
2
,
ʓ
ʷ
3
,
ʓ
ʷ
4
,
ʓ
ʷ
5
,and
ʓ
ʷ
6
do not intersect each other except at common
endpoints. Aboutthis,let
˄
be the triangle defined by the three vertices
s
μ
,
s
ʷ
3
,and
s
ʷ
5
; it is easy to see that
ʓ
ʷ
2
is completely contained inside
˄
except for the segment
2
+
s
ʷ
3
s
ʷ
5
that
ʓ
ʷ
2
shares with
˄
. Namely the angle inside
˄
at
s
ʷ
3
is
ʸ
, while the angle
is at least
4
<
4
). Since
inside
˄
at
s
ʷ
5
(because the angle inside
˄
at
s
μ
is
ʸ
and 2
ʸ
ʲ
ʷ
2
<
4
ʳ
ʷ
2
<
4
, the triangle
except for the vertical side
shared by the two triangles. Concerning Invariant
Ib.
, we observe that
and
˄
ʷ
2
is completely inside
˄
ʓ
ʷ
1
,
ʓ
ʷ
2
,
ʓ
ʷ
3
,
ʓ
ʷ
4
,and
ʓ
ʷ
5
are rotated by an angle that is a multiple of
ʸ
and therefore
Ib.
holds by
construction for each of them. We now show that the slope
ˆ
of the edge corresponding
to
ʷ
4
is in fact either a green slope or a yellow one (refer to Figure 3(f)). Let
ʴ
x
1
be the
horizontal distance between
s
ʷ
3
and
t
μ
and let
ʴ
x
2
be the horizontal distance between
s
μ
and
s
ʷ
3
.Bysimpletrigonometry we have
ʴ
x
1
tan(
g
4ʔ
)=
ʴ
x
2
tan(
ˆ
) and
ʴ
x
1
tan(
g
j
)=
ʴ
x
2
tan(
g
3
),where
g
j
is the slope of the segment representing the edge corresponding
to
ʷ
5
(and therefore
j
= 4
ʔ
−
ʔ
(
t
ʷ
3
)). From the two previousequations we obtain
tan(ˆ)=
tan(
g
4ʔ
)tan(
g
3
)
tan(
g
j
)
. Notice that 1
≤
ʔ(
t
ʷ
3
)
≤
ʔ
and therefore 3
ʔ
≤
j
≤
4
ʔ
−
1. If
j
= 4
ʔ
−
1, then tan(
g
3
)=
−
tan(
g
j
) and tan(
ˆ
)=
−
tan(
g
4ʔ
)=tan(
g
2
), hence
ˆ
=
g
2
,
= arctan
tan(
g
4ʔ
)tan(
g
3
)
tan(
g
j
)
= and therefore
i.e.,
ˆ
is a green slope. Otherwise
ˆ
ˆ
is the
yellow slope
y
1
,
j
(recall that
g
1
= 0). Concerning Invariant
Ic.
,wehavethat
ʔ
(
s
μ
)=
ʸ
≤
−
ʔ
(
s
ʷ
1
)+2and
ʔ
(
t
μ
)=
ʔ
(
t
ʷ
3
)+2. Moreover,
ʲ
μ
=
ʲ
ʷ
1
+ 2
(
ʔ
(
s
ʷ
1
)
1)
ʸ
+ 2
ʸ
=
−
ʸ
≤
−
−
(
ʔ
(
s
μ
)
1)
ʸ
. Finally,
ʳ
μ
=
ʳ
ʷ
3
+ 2
(
ʔ
(
t
ʷ
3
)
1)
ʸ
+ 2
ʸ
=(
ʔ
(
t
μ
)
1)
ʸ
.
Lemma 11.
Let
ˁ
be the root of T andlet
ʾ
be its uniquechild. GraphG
=
G
ˁ
∪
G
ʾ
admits a straight-linedrawing
ʓ
that respects Invariants
Ia.
,
Ib.
and
Ic
.
ByLemmas7,8,9,10,and11,wecanprovethefollowing lemma.
Lemma 12.
Let G be a biconnected outer
1
-plane graphwithnvertices and withmax-
imum degree
2
ʔ
. Gadmits a planarstraight-linedrawing that uses at most
4
ʔ
−
4
ʔ
slopes. Also, this drawing can be computed in O
(
n
)
time.
The result above can be extended to simply connected outer 1-planar graph with the
same technique described in Section 3. We obtain the following theorem.
Theorem 2.
Let G be an outer
1
-plane graphwithnvertices and withmaximum degree
ʔ
2
+ 12
. Gadmits a planarstraight-linedrawing that uses at most
4
ʔ
ʔ
+ 8
slopes.
Also, this drawing can be computed in O
(
n
)
time.
5
Open Problems
An interesting open problem motivated by ourresult of Section 3 is whether the 1
-
planar slope number
of 1-planar straight-line drawable graphs (not all 1-planar graphs
admit a 1-planar straight-line drawing [12]), is bounded in
or not. A second problem
is whether the quadratic upper bound of Section 4 is tight or not. Finally, it could be
interesting to further explore trade-offs between slopes and crossings, e.g., can we draw
planar partial 3-trees with
o
(
ʔ
5
) slopes and a constant number of crossings per edge?
ʔ