The general bivariate quadratic curve can be written
(1)
|
Define the following quantities:
(2)
| |||
(3)
| |||
(4)
| |||
(5)
|
Then the quadratics are classified into the types summarized in the following table (Beyer 1987). The real (nondegenerate) quadratics (the ellipse and its special case the circle, hyperbola, and parabola) correspond to the curves which can be created by the intersection of a plane with a (two-nappes) cone, and are therefore known as conic sections.
curve | ||||
coincident lines | 0 | 0 | 0 | |
ellipse (imaginary) | ||||
ellipse (real) | ||||
hyperbola | ||||
intersecting lines (imaginary) | 0 | |||
intersecting lines (real) | 0 | |||
parabola | 0 | |||
parallel lines (imaginary) | 0 | 0 | ||
parallel lines (real) | 0 | 0 |
It is always possible to eliminate the cross term by a suitable rotation
of the axes. To see this, consider rotation by an arbitrary angle
. The rotation matrix
is
(6)
| |||
(7)
|
so
(8)
| |||
(9)
| |||
(10)
| |||
(11)
| |||
(12)
|
Plugging these into (◇) and grouping terms gives
(13)
|
Comparing the coefficients with (◇) gives an equation of the form
(14)
|
where the new coefficients are
(15)
| |||
(16)
| |||
(17)
| |||
(18)
| |||
(19)
| |||
(20)
|
The cross term
can therefore be made to vanish by setting
(21)
| |||
(22)
|
For
to be zero, it must be true that
(23)
|
The other components are then given with the aid of the identity
(24)
|
by defining
(25)
|
so
(26)
| |||
(27)
|
Rotating by an angle
(28)
|
therefore transforms (◇) into
(29)
|
(30)
|
(31)
|
Defining ,
, and
gives
(32)
|
If ,
then divide both sides by
. Defining
and
then gives
(33)
|
Therefore, in an appropriate coordinate system, the general conic section can be written (dropping the primes) as
(34)
|
Consider an equation of the form where
. Re-express this using
and
in the form
(35)
|
Therefore, rotate the coordinate system
(36)
|
so
(37)
|
and
(38)
| |||
(39)
| |||
(40)
|
Therefore,
(41)
| |||
(42)
| |||
(43)
| |||
(44)
| |||
(45)
|
(46)
|
the same angle as before. But
(47)
| |||
(48)
| |||
(49)
|
so
(50)
|
Rewriting and copying (◇),
(51)
| |||
(52)
| |||
(53)
|
(54)
| |||
(55)
|
Note that these roots can also be found from
(56)
|
(57)
| |
(58)
| |
(59)
| |
(60)
|
The original problem is therefore equivalent to looking for a solution to
(61)
|
(62)
|
which gives the simultaneous equations
(63)
|
Let
be any point
with old coordinates and
be its new coordinates. Then
(64)
|
and
(65)
| |||
(66)
|
If
and
are both
,
the curve is an ellipse. If
and
are both
, the curve is empty. If
and
have opposite signs, the curve
is a hyperbola. If either is 0, the curve is a parabola.
To find the general form of a quadratic curve in polar coordinates (as given, for example, in Moulton 1970), plug and
into (◇) to obtain
(67)
|
(68)
|
Define .
For
,we
can divide through by
,
(69)
|
Applying the quadratic formula gives
(70)
|
where
(71)
| |||
(72)
|
Using the trigonometric identities
(73)
| |||
(74)
|
it follows that
(75)
| |||
(76)
| |||
(77)
|
Defining
(78)
| |||
(79)
| |||
(80)
| |||
(81)
| |||
(82)
|
then gives the equation
(83)
|
(Moulton 1970). If ,
then (◇) becomes instead
(84)
|
Therefore, the general form of a quadratic curve in polar coordinates is given by
(85)
|