10 - Lec06 - Ch08 Qudratic Form Ellipse Example
Uploaded by
Anonymous Ms8nYcbxY210 - Lec06 - Ch08 Qudratic Form Ellipse Example
Uploaded by
Anonymous Ms8nYcbxY2Quadratic Form Example - Ellipse
Roger Bradshaw, ME 565
Plot the case from the book and slides corresponding to the ellipse
becomes
This is defined for the domain [-8, 8] for y1. Lets plot 160 points in y1-y2 space - note that the y2 points are +/- pairs.
(i.e. if y2 = 1 is a solution then y2 = -1 is also a solution due to the exponent 2 in the ellipse equation).
Rearrange the equation to find y2 given y1 as:
2 2
y2 ( y1) 2 1
y1
y2a ( y1) 2 1
y1
For this y1 there is another solution
8 8
Now create points that we can plot i 0 160 y1i 0.1 i 8 y2pi y2 y1i y2m i y2a y1i
161 points 161 y1 points 161 y2 points >= 0 161 y2 points <= 0
We can plot these in y1-y2 space to see it is an ellipse - note that -2 <= y2 <= 2 (used below)
Plot Of Solution In y1-y2 Space
3
y2p i
y2
0
y2mi
2
Pos y2 Solution
Neg y2 Solution
3
9 8 7 6 5 4 3 2 1 0 1 2 3 4 5 6 7 8 9
y1i
y1
However, what we really want to see is the shape in the x1-x2 space (how we were given the equation). Recall from the notes
that we have the relationship between the y vector and the x vector is y = XT x. Since X is orthonormal, we can also say that
X-1 = XT so y = X-1 x. Then multiply both sides by X to find x = X y. Hence, we can take the points above as a vector y and
determine where those points are in x.
For example, the 31st point above (arbitrary) on the positive y2 side is:
(note 31st since MathCad indexes first point as subscript 0)
y130 5
y130 5 y2p30 1.561 which we can combine as a vector as yv yv
y2p30 1.561
Where is this point in the x1-x2 space? Define the X matrix as given above and use it to find x:
1 1 1 2.432
X xv X yv xv
2 1 1 4.64
T 5
Alternately if we were given x we could find its associated y using XT: X xv This recovers y above
1.561
Now map every point y1-y2 point above into a corresponding x1-x2 point:
x1pi y1i x1mi y1i
X This is for the positive y2 points X This is for the negative y2 points
x2pi y2pi x2mi y2mi
Now lets also map points that correspond to the y1-y2 axes. For y1 axis, y2 = 0 while for y2 axis y1 = 0. So we have:
Same 161 points that 4 161 points evenly spaced
y1axi y1i we started with for y1 y2axi 2 i along -2 <= y2 <= 2 Nulli 0 0 vector
160
The y1 axis x1y1axi y1axi The y2 axis x1y2axi Nulli
is then given by X is then given by X
(i.e. for [y1 0]T) x2y1axi Nulli (i.e. for [0 y2]T) x2y2axi y2axi
NOTE
Plot Of Solution In x1-x2 Space
7 In this view, I have plotted the graph
6
such that each side is equal length.
5 This leads to the y1 and y2 axes
appearing normal to each other
4 in the x1-x2 space (which they are).
3
We can show this as:
x2p i 2
y1 unit vector mapped to x
x2mi 1
1 0.707
x1y1 X x1y1
x2
x2y1axi 0
0 0.707
x2y2axi 1
y2 unit vector mapped to x
2
0 0.707
3
x1y2 X x1y2
1 0.707
4 Pos y2 Solution T
Neg y2 Solution x1y1 x1y2 0 They are orthogonal
5
y1 Axis Note that y1 unit vector mapped to x
6 y2 Axis is simply the first eigenvector.
7
7 6 5 4 3 2 1 0 1 2 3 4 5 6 7 Similarly the y2 unit vector mapped to x
x1p i x1mi x1y1axi x1y2axi is simply the second eigenvector.
x1 Hence the eigenvectors are the y axes
