I'm getting ellipses as level curves of a fit dataset. After selecting a particular ellipse, I would like to report it as a center point, semi-major and minor axes lengths, and a rotation angle. In other words, I would like to transform (using mathematica) my ellipse equation from the form:
Ax^2 + By^2 + Cx + Dy + Exy + F = 0
to a more standard form:
((xCos[alpha] - ySin[alpha] - h)^2)/(r^2) + ((xSin[alpha] + yCos[alpha] - k)^2)/(s^2) = 1
where (h,k) is the center, alpha is the rotation angle, and r and s are the semi-axes
The actual equation I'm attempting to transform is
1.68052 x - 9.83173 x^2 + 4.89519 y - 1.19133 x y - 9.70891 y^2 + 6.09234 = 0
I know the center point is the fitted maximum, which is:
{0.0704526, 0.247775}
I posted a version of this answer on Math SE since it benefits a lot from proper mathematical typesetting. The example there is simpler as well, and there are some extra details.
The following description follows the German Wikipedia article Hauptachsentransformation. Its English counterpart, according to inter-wiki links, is principal component analysis. I find the former article a lot more geometric than the latter. The latter has a strong focus on statistical data, though, so it might be useful for you nevertheless.
Rotation
Your ellipse is described as
[A E/2] [x] [x]
[x y] * [E/2 B] * [y] + [C D] * [y] + F = 0
First you identify the rotation. You do this by identifying the eigenvalues and eigenvectors of this 2×2 matrix. These eigenvectors will form an orthogonal matrix describing your rotation: its entries are the Sin[alpha] and Cos[alpha] from your formula.
With your numbers, you get
[A E/2] [-0.74248 0.66987] [-10.369 0 ] [-0.74248 -0.66987]
[E/2 B] = [-0.66987 -0.74248] * [ 0 -9.1715] * [ 0.66987 -0.74248]
The first of the three factors is the matrix formed by the eigenvectors, each normalized to unit length. The central matrix has the eigenvalues on the diagonal, and the last one is the transpose of the first. If you multiply the vector (x,y) with that last matrix, then you will change the coordinate system in such a way that the mixed term vanishes, i.e. the x and y axes are parallel to the main axes of your ellipse. This is just what happens in your desired formula, so now you know that
Cos[alpha] = -0.74248 (-0.742479398678 with more accuracy)
Sin[alpha] = 0.66987 ( 0.669868899516)
Translation
If you multiply the row vector [C D] in the above formula with the first of the three matrices, then this effect will exactly cancel the multiplication of (x, y) by the third matrix. Therefore in that changed coordinate system, you use the central diagonal matrix for the quadratic term, and this product for the linear term.
[-0.74248 0.66987]
[1.68052, 4.89519] * [-0.66987 -0.74248] = [-4.5269 -2.5089]
Now you have to complete the square independently for x and y, and you end up with a form from which you can read the center coordinates.
-10.369x² -4.5269x = -10.369(x + 0.21829)² + 0.49408
-9.1715y² -2.5089y = -9.1715(y + 0.13677)² + 0.17157
h = -0.21829 (-0.218286476695)
k = -0.13677 (-0.136774259156)
Note that h and k describe the center in the already rotated coordinate system; to obtain the original center you'd multiply again with the first matrix:
[-0.74248 0.66987] [-0.21829] [0.07045]
[-0.66987 -0.74248] * [-0.13677] = [0.24778]
which fits your description.
Scaling
The completed squares above contributed some more terms to the constant factor F:
6.09234 + 0.49408 + 0.17157 = 6.7580
Now you move this to the right side of the equation, then divide the whole equation by this number so that you get the = 1 from your desired form. Then you can deduce the radii.
1 -10.369
-- = ------- = 1.5344
r² -6.7580
1 -9.1715
-- = ------- = 1.3571
s² -6.7580
r = 0.80730 (0.807304599162099)
s = 0.85840 (0.858398019487315)
Verifying the result
Now let's check that we didn't make any mistakes. With the parameters we found, you can piece together the equation
((-0.74248*x - 0.66987*y + 0.21829)^2)/(0.80730^2)
+ (( 0.66987*x - 0.74248*y + 0.13677)^2)/(0.85840^2) = 1
Move the 1 to the left side, and multiply by -6.7580, and you should end up with the original equation. Expanding that (with the extra precision versions printed in parentheses), you'll get
-9.8317300000 x^2
-1.1913300000 x y
+1.6805200000 x
-9.7089100000 y^2
+4.8951900000 y
+6.0923400000
which is a perfect match for your input.
If you have h and k, you can use Lagrange Multipliers to maximize / minimize the function (x-h)^2+(y-k)^2 subject to the constraint of being on the ellipse. The maximum distance will be the major radius, the minimum distance the minor radius, and alpha will be how much they are rotated from horizontal.
Related
I have been trying to understand Jacobian Determinant.
I hope someone is able to give me a pointer.
Most material that I found on Internet didn't provide
derivation of Jacobian Determinant.
One such web site is:
http://tutorial.math.lamar.edu
(Which I find quite good, otherwise.)
I spent a lot of time trying to deepen my understanding of
Jacobian Determinant.
I played with Transformations that define uv-axes and
how integration of a function over a Region/area would work
with the Transformations.
For example, when I started with simple Transformations of:
u = ( x - y )/√2
v = ( x + y )/2√2
which is uv-axes rotated -45° from Cartesian xy-axes,
and with v-axis at 2 times the scale,
that is, v = 1 maps to 2 units length in xy-coords.
So, I say that uscale = 1, vscale = 2,
for the above transformations.
With this uv-axes, I can simplify a 10x20 rectangle Region
which is rotated at 45° from x-axis,
such that the longer dimension points at 45° from x-axis.
With such examples, I begin to develop intuition
how Jacobian Determinant works.
I understand Jacobian Determinant to be a Scaling Factor
to convert area measurement in uv-axes to xy-dimensions.
Area measurement in uv-axes is given simply by formula
Δu x Δv, where Δu = 10, Δv = 10, because vscale = 2).
Jacobian Determinant Scaling Factor = uscale x vscale
(quite intuitively).
Area in xy-dimensions = Δu x Δv x (uscale x vscale)
= 10 x 10 x 1 x 2 = 200.
Integration of volume over such a simpler uv Square,
could be easier than over the same xy Region,
appearing at an angle.
With the above initial understanding,
I am trying to work out how Jacobian Determinant is derived.
Deriving from the above Transformations formula:
dx/du = √2 / 2
dx/dv = √2
dy/du = -√2 / 2
dy/dv = √2
I can also derive from Geometry that:
dx/du = uscale cos Θ
dy/du = uscale sin Θ
dx/dv = vscale cos (90° - Θ)
dy/dv = vscale sin (90° - Θ)
I could get:
areaInXY / areaInUV = uscale x vscale
which matches my understanding.
However, Jacobian Determinant formula is:
∂(x, y) / ∂(u, v) = ∂x/∂u ∂y/∂v - ∂x/∂v ∂y/∂u
= uscale * vscale * cos 2Θ
This leaves me quite puzzled why I have the extra cos 2Θ factor
which isn't making intuitive sense -- why would the
area Scaling Factor depends on how the rectangle is rotated
and thus how uv-axes are rotated?!
Anybody can see where my reasoning went wrong above?
Let me try to explain what basically the Jacobian determinant does. This is true in general for smooth functions mapping from R^n to R^n, but for the sake of simplicity, assume we are working on R^2. Let F(x,y) a smooth R^2 to R^2 function. Then we can say that F(x,y) sends the x coordinate to f1(x,y) and the y coordiate to f2(x,y) at point (x,y). Then think about an infinitesimal rectangular area, defined by the points (x,y),(x+dx,y),(x,y+dy) and (x+dx,y+dy). Now, the area of this infinitesimal rectangle is dxdy. What happens to this rectangle when it goes through the F(x,y) transformation? We apply F(x,y) to each of the four coordinates and obtain the following points:
A:(x,y)->(f1(x,y),f2(x,y))
B:(x+dx,y) -> (f1(x+dx,y),f2(x+dx,y)) (approx.)= (f1(x,y) + (∂f1/∂x)dx,f2(x,y) + (∂f2/∂x)dx)
C:(x,y+dy) -> (f1(x,y+dy),f2(x,y+dy)) (approx.)= (f1(x,y) + (∂f1/∂y)dy,f2(x,y) + (∂f2/∂y)dy)
D:(x+dx,y+dy) -> (f1(x+dx,y+dy),f2(x+dx,y+dy)) (approx.)=(f1(x,y) + (∂f1/∂x)dx + (∂f1/∂y)dy,f2(x,y) + (∂f2/∂x)dx + (∂f2/∂y)dy)
The equalities are approximately equal and exactly hold in the limit where dx and dy goes to 0, they are the best linear approximation to the function F at new points. (We obtain these from the first order parts of the Taylor approximation of the functions f1 and f2).
If we look to the new (approximated) area under the transformation F(x,y), we see the new distance vectors between the transformed points a:
B-A:((∂f1/∂x)dx,(∂f2/∂x)dx)
C-A:((∂f1/∂y)dy,(∂f2/∂y)dy)
D-C:((∂f1/∂x)dx,(∂f2/∂x)dx)
D-B:((∂f1/∂y)dy,(∂f2/∂y)dy)
As you can see, the newly transformed infinitesimal area is a parallelogram. Let:
u=((∂f1/∂x)dx,(∂f2/∂x)dx)
v=((∂f1/∂y)dy,(∂f2/∂y)dy)
These vectors constitute the edges of our parallelogram. It can be shown with the help of the cross product between u and v, that the area of the parallelogram is:
area^2 = (u1v2 - u2v1)^2 = ((∂f1/∂x)(∂f2/∂y)dxdy - (∂f2/∂x)(∂f1/∂y)dxdy)^2
area^2 = ((∂f1/∂x)(∂f2/∂y) - (∂f2/∂x)(∂f1/∂y))^2 (dxdy)^2
area = |(∂f1/∂x)(∂f2/∂y) - (∂f2/∂x)(∂f1/∂y)|dxdy (dx and dy are positive)
area = |det([∂f1/∂x, ∂f1/∂y],[∂f2/∂x, ∂f2/∂y])|dxdy
So, the matrix we are going to take the determinant of is simply the Jacobian matrix. Like I said in the beginning, this derivation can be extended to arbitrary dimensions of n,given the coordinate transformation function F is smooth and the Jacobian matrix is hence invertible, with non-zero determinant.
A good visual explanation of this is given at: http://mathinsight.org/double_integral_change_variables_introduction
I need to store 3D normal vectors, that is vectors (x, y, z) such that x^2 + y^2 + z^2 = 1. But due to space constraints I can only use 2 floats to store it.
So by storing only x and y, the third component can be computed as sqrt(1 - x^2 - y^2), i.e. one square root, two products and two subtractions.
What would be the most efficient way to store the vectors, so that reading them is as fast as possible, and if possible without bias towards one spatial direction?
Edit
Now using the values (a, b) with a = x - y and b = x + y.
You could satisfy your space constraint by storing the vectors via spherical coordinates. As is well known, each point on the unit sphere, i.e., each unit vector, has at least one pair of spherical coordinates characterizing it.
Or if you want something convoluted: The complex square function maps the unit disk to a double cover of it. So you could use the left half-disk for the upper half-sphere and the right half-disk for the lower half-sphere.
SphereFromDisk(a,b)
a2=a*a; b2=b*b; r2=a2+b2; // assert r2 <= 1
x = a2 - b2;
y = 2*a*b
z = sqrt(1-r2*r2)
if(a<0 or (a=0 and b<0) z=-z
return (x,y,z)
Hey so I'm reading this article by Chris Hecker where he has an image of a Parabola surrounded by the a vector field of it's derivative:
However he never mentions how exactly he got the vector field equation, and never even states it. He does say he overlayed the vector field of the slopes in Figure 1, by drawing the solution to the slope equation, dy/dx = 2x, as a short vector at each coordinate on the grid.
How do you create a vector field of the slopes of an equation in the vector field syntax of
V = xi + yj
The Figure title would be clearer if it read:
The curve y = x^2, and the vector field dy/dx = 2x for the general case y = x^2 + C
There are three equations at work in the graph above:
y = x^2 - The equation for the parabola drawn - This is the one long solid curve
y = x^2 + C -The equation for all parabolas that fit on the vector field - C is a constant. This is the equation for all parabolas that fit on that vector field
dy/dx = 2x The equation for the slope field. - This is the slope or derivative of the both the curve drawn and all the possible curves that can be drawn with y = x^2 + C for all constant Cs.
Note that C is a constant, since the derivative of y = x^2 + C with any C is 2x. So the vector field shows how to draw all the different parabolas with different Cs.
So there are two ways to calculate the vector field:
Iterate over your desired range of x and y and calculate the slope, dy/dx- 2x independent of y in this case - at each point. This is how the author did it.
Draw a bunch of parabolas by slowly varying C in y = x^2 + C over a desired range of - let's say - x calculating y.
For a differential equation dy/dx = f(x,y) (e.g., dy/dx = 2x in this case, with f(x,y) = 2x), the vector field (F) will be F = i + f(x,y)j (so in your case, F = i + 2x j )
I'm trying to generate some axis vectors from parameters commonly used to specify crystallographic unit cells. These parameters consist of the length of the three axes: a,b,c and the angles between them: alpha,beta,gamma. By convention alpha is the angle between the b and c axes, beta is between a and c, and gamma between a and b.
Now getting vector representations for the first two is easy. I can arbitrarily set the the a axis to the x axis, so a_axis = [a,0,0]. I then need to rotate b away from a by the angle gamma, so I can stay in the x-y plane to do so, and b_axis = [b*cos(gamma),b*sin(gamma),0].
The problem is the third vector. I can't figure out a nice clean way to determine it. I've figured out some different interpretations but none of them have panned out. One is imagining the there are two cones around the axes axis_a and axis_b whose sizes are specified by the angles alpha and beta. The intersection of these cones create two lines, the one in the positive z direction can be used as the direction for axis_c, of length c.
Does someone know how I should go about determining the axis_c?
Thanks.
The angle alpha between two vectors u,v of known length can be found from their inner (dot) product <u,v>:
cos(alpha) = <u,v>/(||u|| ||v||)
That is, the cosine of alpha is the inner product of the two vectors divided by the product of their lengths.
So the z-component of your third can be any nonzero value. Scaling any or all of the axis vectors after you get the angles right won't change the angles, so let's assume (say) Cz = 1.
Now the first two vectors might as well be A = (1,0,0) and B = (cos(gamma),sin(gamma),0). Both of these have length 1, so the two conditions to satisfy with choosing C are:
cos(alpha) = <B,C>/||C||
cos(beta) = <A,C>/||C||
Now we have only two unknowns, Cx and Cy, to solve for. To keep things simple I'm going to just refer to them as x and y, i.e. C = (x,y,1). Thus:
cos(alpha) = [cos(gamma)*x + sin(gamma)*y]/sqrt(x^2 + y^2 + 1)
cos(beta) = x/(sqrt(x^2 + y^2 + 1)
Dividing the first equation by the second (assuming beta not a right angle!), we get:
cos(alpha)/cos(beta) = cos(gamma) + sin(gamma)*(y/x)
which is a linear equation to solve for the ratio r = y/x. Once you have that, substituting y = rx in the second equation above and squaring gives a quadratic equation for x:
cos^2(beta)*((1+r^2)x^2 + 1) = x^2
cos^2(beta) = (1 - cos^2(beta)*(1 + r^2))x^2
x^2 = cos^2(beta)/[(1 - cos^2(beta)*(1 + r^2))]
By squaring the equation we introduced an artifact root, corresponding to choosing the sign of x. So check the solutions for x you get from this in the "original" second equation to make sure you get the right sign for cos(beta).
Added:
If beta is a right angle, things are simpler than the above. x = 0 is forced, and we have only to solve the first equation for y:
cos(alpha) = sin(gamma)*y/sqrt(y^2 + 1)
Squaring and multiplying away the denominator gives a quadratic for y, similar to what we did before. Remember to check your choice of a sign for y:
cos^2(alpha)*(y^2 + 1) = sin^2(gamma)*y^2
cos^2(alpha) = [sin^2(gamma) - cos^2(alpha)]*y^2
y^2 = cos^2(alpha)/[sin^2(gamma) - cos^2(alpha)]
Actually if one of the angles alpha, beta, gamma is a right angle, it might be best to label that angle gamma (between the first two vectors A,B) to simplify the computation.
Here is a way to find all Cx, Cy, Cz (first two are the same as in the other answer), given that A = (Ax,0,0), B = (Bx, By, 0), and assuming that |C| = 1
1) cos(beta) = AC/(|A||C|) = AxCx/|A| => Cx = |A|cos(beta)/Ax = cos(beta)
2) cos(alpha) = BC/(|B||C|) = (BxCx+ByCy)/|B| => Cy = (|B|cos(alpha)-Bx cos(beta))/By
3) To find Cz let O be the point at (0,0,0), T the point at (Cx,Cy,Cz), P be the projection of T on Oxy and Q be the projection of T on Ox. So P is the point at (Cx,Cy,0) and Q is the point at (Cx,0,0). Thus from the right angle triangle OQT we get
tan(beta) = |QT|/||OQ| = |QT|/Cx
and from the right triangle TPQ we get |TP|^2 + |PQ|^2 = |QT|^2. So
Cz = |TP| = sqrt(|QT|^2 - |PQ|^2) = sqrt( Cx^2 tan(beta)^2 - Cy^2 )
I'm not sure if this is correct but I might as well take a shot. Hopefully I won't get a billion down votes...
I'm too lazy to scale the vectors by the necessary amounts, so I'll assume they are all normalized to have a length of 1. You can make some simple modifications to the calculation to account for the varying sizes. Also, I'll use * to represent the dot product.
A = (1, 0, 0)
B = (cos(g), sin(g), 0)
C = (Cx, Cy, Cz)
A * C = cos(beta) //This is just a definition of the dot product. I'm assuming that the magnitudes are 1, so I can skip that portion, and you said that beta was the angle between A and C.
A * C = Cx //I did this by multiplying each corresponding value, and the Cy and Cz were ignored because they were being multiplied by 0
cos(beta) = Cx //Combine the previous two equations
B * C = cos(alpha)
B * C = Cx*cos(g) + Cy*sin(g) = cos(beta) * cos(g) + Cy*sin(g)
(cos(alpha) - cos(beta) * cos(g))/(sin(g)) = Cy
To be honest, I'm not sure how to get the z component of vector C, but I would expect it to be one more relatively easy step. If I can figure it out, I'll edit my post.
This is a maths problem I am not exactly sure how to do. The vector is not aligned to an axis, so just rotating 90 degrees around x, y or z won't necessarily give me the other axes.
I can think of a couple of different scenarios you might be asking about.
Given: A pre-existing coordinate system
In a 2D system, your axes/basis are always [1,0] and [0,1] -- x and y axes.
In a 3D system, your axes/basis are always [1,0,0], [0,1,0], and [0,0,1] -- x, y, and z.
Given: One axis in an arbitrary-basis 2D coordinate system
If you have one axis in an arbitrary-basis 2D coordinate system, the other axis is the orthogonal vector.
To rotate a vector orthogonally counter-clockwise:
[x_new, y_new] = [ -y_old, x_old]
To rotate a vector orthogonally clockwise:
[x_new, y_new] = [ y_old, -x_old]
To summarize:
Given: x-axis = [ a, b]
Then: y-axis = [-b, a]
Given: y-axis = [ c, d]
Then: x-axis = [ d, -c]
Given: Two axes in an arbitrary-basis 3D coordinate system
To do this, find the cross product.
[a,b,c] x [d,e,f] = [ b*f - c*e, c*d - a*f, a*e - b*d ]
Following these three guidelines:
(x axis) x (y axis) = (z axis)
(y axis) x (z axis) = (x axis)
(z axis) x (x axis) = (y axis)
Given: One axis in an arbitrary-basis 3D coordinate system
There is not enough information to find the unique solution this problem. This is because, if you look at the second case (One axis in an arbitrary-basis 2D coordinate system), you first need to find an orthogonal vector. However, there are an infinite amount of possible orthogonal vectors to a single axis in 3D space!
You can, however, find one of the possible solutions.
One way to find an arbitrary one of these orthogonal vectors by finding any vector [d,e,f] where:
[a,b,c] = original axis
[d,e,f] = arbitrary orthogonal axis (cannot be [0,0,0])
a*d + b*e + c*f = 0
For example, if your original axis is [2,3,4], you'd solve:
2 * d + 3 * e + 4 * f = 0
That is, any value of [d,e,f] that satisfies this is a satisfactory orthogonal vector (as long as it's not [0,0,0]). One could pick, for example, [3,-2,0]:
2 * 3 + 3 *-2 + 4 * 0 = 0
6 + -6 + 0 = 0
As you can see, one "formula" that works to is [d,e,f] = [b,-a,0]...but there are many other ones that can work as well; there are, in fact, an infinite!
Once you find your two axes [a,b,c] and [d,e,f], you can reduce this back to the previous case (case 3), using [a,b,c] and [d,e,f] as your x and y axes (or whatever axes you need them to be, for your specific problem).
Normalization
Note that, as you continually do dot products and cross products, your vectors will begin to grow larger and larger. Depending on what you want, this might not be desired. For example, you might want your basis vectors (your coordinate axes) to all be the same size/length.
To turn any vector (except for [0,0,0]) into a unit vector (a vector with a length of 1, in the same direction as the original vector):
r = [a,b,c]
v = Sqrt(a^2 + b^2 + c^2) <-- this is the length of the original vector
r' = [ a/v , b/v , c/v ]
Where r' represents the unit vector of r -- a vector with length of 1 that points in the same direction as r does. An example:
r = [1,2,3]
v = Sqrt(1^2 + 2^2 + 3^2) = Sqrt(13) = 3.60555 <-- this is the length of the original vector
r' = [0.27735, 0.55470, 0.83205]
Now, if I wanted, for example, a vector in the same direction of r with a length of 5, I'd simply multiply out r' * 5, which is [a' * 5, b' * 5, c' * 5].
Having only one axis isn't enough, since there are still an infinite number of axes that can be in the perpendicular plane.
If you manage to get another axis though, you can use the cross product to find the third.
If you have one vector (x,y,z) you can get one perpendicular vector to it as (y,-x,0) (dot-product is xy-yx+0*z = 0)
Then you take the cross-product of both to get the remaining perpendicular vector:
(x,y,z) × (y,-x,0) = (0y+zx, yz-0x, -x²-y²) = (zx, yz, -x²-y²)
Are you talking about a typical 3coordinate system like the one used in a 3D engine?
With just a vector you can't find the other two, the only information you will have it the plane on which they lay.. but they can be at any angle also if they're perpendicular with the only one vector you have.