VisTrails Home

User:Tohline/Appendix/Ramblings/ConcentricEllipsodalCoordinates

From VisTrailsWiki

(Difference between revisions)
Jump to: navigation, search
(Speculation4)
(Speculation4)
Line 1,231: Line 1,231:
<table border="1" width="80%" align="center" cellpadding="10"><tr><td align="left">
<table border="1" width="80%" align="center" cellpadding="10"><tr><td align="left">
-
Note that, <math>~\hat{e}_3 \cdot \hat{e}_3 = 1</math>, which means that this also is a properly normalized ''unit'' vector.
+
Note that, <math>~\hat{e}_3 \cdot \hat{e}_3 = 1</math>, which means that this also is a properly normalized ''unit'' vector.  Note, as well, that the dot product between our "first" and "third" unit vectors should be zero if they are indeed orthogonal to each other.  Let's see &hellip;
 +
<table border="0" cellpadding="5" align="center">
 +
 
 +
<tr>
 +
  <td align="right">
 +
<math>~\hat{e}_3 \cdot \hat{e}_1</math>
 +
  </td>
 +
  <td align="center">
 +
<math>~=</math>
 +
  </td>
 +
  <td align="left">
 +
<math>~
 +
(- q^2y \ell_q)x\ell_{3D} + (x\ell_q) q^2y\ell_{3D} = 0 \, .
 +
</math>
 +
  </td>
 +
</tr>
 +
</table>
 +
Q.E.D.
</td></tr></table>
</td></tr></table>
 +
 +
Now, even though we have not yet determined the proper expression for the "second" orthogonal coordinate, <math>~\lambda_2</math>, we should be able to obtain an expression for its associated unit vector from the cross product of the "third" and "first" unit vectors.  Specifically we find,
 +
<table border="0" cellpadding="5" align="center">
 +
 +
<tr>
 +
  <td align="right">
 +
<math>~\hat{e}_3 \times \hat{e}_1</math>
 +
  </td>
 +
  <td align="center">
 +
<math>~=</math>
 +
  </td>
 +
  <td align="left">
 +
<math>~</math>
 +
  </td>
 +
</tr>
 +
</table>
=See Also=
=See Also=

Revision as of 16:21, 30 October 2020

Contents

Concentric Ellipsoidal (T6) Coordinates

Whitworth's (1981) Isothermal Free-Energy Surface
|   Tiled Menu   |   Tables of Content   |  Banner Video   |  Tohline Home Page   |

Background

Building on our general introduction to Direction Cosines in the context of orthogonal curvilinear coordinate systems, and on our previous development of T3 (concentric oblate-spheroidal) and T5 (concentric elliptic) coordinate systems, here we explore the creation of a concentric ellipsoidal (T6) coordinate system. This is motivated by our desire to construct a fully analytically prescribable model of a nonuniform-density ellipsoidal configuration that is an analog to Riemann S-Type ellipsoids.

Orthogonal Coordinates

Primary (radial-like) Coordinate

We start by defining a "radial" coordinate whose values identify various concentric ellipsoidal shells,

~\lambda_1

~\equiv

~(x^2 + q^2 y^2 + p^2 z^2)^{1 / 2} \, .

When ~\lambda_1 = a, we obtain the standard definition of an ellipsoidal surface, it being understood that, ~q^2 = a^2/b^2 and ~p^2 = a^2/c^2. (We will assume that ~a > b > c, that is, ~p^2 > q^2 > 1.)

A vector, ~\bold{\hat{n}}, that is normal to the ~\lambda_1 = constant surface is given by the gradient of the function,

~F(x, y, z)

~\equiv

~(x^2 + q^2 y^2 + p^2 z^2)^{1 / 2} - \lambda_1 \, .

In Cartesian coordinates, this means,

~\bold{\hat{n}}(x, y, z)

~=

~
\hat\imath \biggl( \frac{\partial F}{\partial x} \biggr)
+ \hat\jmath \biggl( \frac{\partial F}{\partial y} \biggr)
+ \hat{k} \biggl( \frac{\partial F}{\partial z} \biggr)

 

~=

~
\hat\imath \biggl[ x(x^2 + q^2 y^2 + p^2 z^2)^{- 1 / 2} \biggr]
+ \hat\jmath \biggl[ q^2y(x^2 + q^2 y^2 + p^2 z^2)^{- 1 / 2} \biggr]
+ \hat{k}\biggl[ p^2 z(x^2 + q^2 y^2 + p^2 z^2)^{- 1 / 2} \biggr]

 

~=

~
\hat\imath \biggl( \frac{x}{\lambda_1} \biggr)
+ \hat\jmath \biggl( \frac{q^2y}{\lambda_1} \biggr)
+ \hat{k}\biggl(\frac{p^2 z}{\lambda_1} \biggr) \, ,

where it is understood that this expression is only to be evaluated at points, ~(x, y, z), that lie on the selected ~\lambda_1 surface — that is, at points for which the function, ~F(x,y,z) = 0. The length of this normal vector is given by the expression,

~[ \bold{\hat{n}} \cdot \bold{\hat{n}} ]^{1 / 2}

~=

~
\biggl[ \biggl( \frac{\partial F}{\partial x} \biggr)^2 + \biggl( \frac{\partial F}{\partial y} \biggr)^2 + \biggl( \frac{\partial F}{\partial z} \biggr)^2 \biggr]^{1 / 2}

 

~=

~
\biggl[ \biggl( \frac{x}{\lambda_1} \biggr)^2
+ \biggl( \frac{q^2y}{\lambda_1} \biggr)^2
+ \biggl(\frac{p^2 z}{\lambda_1} \biggr)^2 \biggr]^{1 / 2}

 

~=

~
\frac{1}{\lambda_1 \ell_{3D}}

where,

~\ell_{3D}

~\equiv

~\biggl[ x^2 + q^4y^2 + p^4 z^2 \biggr]^{- 1 / 2} \, .

It is therefore clear that the properly normalized normal unit vector that should be associated with any ~\lambda_1 = constant ellipsoidal surface is,

~\hat{e}_1

~\equiv

~
\frac{ \bold\hat{n} }{ [ \bold{\hat{n}} \cdot \bold{\hat{n}} ]^{1 / 2} }
=
\hat\imath (x \ell_{3D}) + \hat\jmath (q^2y \ell_{3D}) + \hat{k} (p^2 z \ell_{3D}) \, .

From our accompanying discussion of direction cosines, it is clear, as well, that the scale factor associated with the ~\lambda_1 coordinate is,

~h_1^2

~=

~\lambda_1^2 \ell_{3D}^2 \, .

We can also fill in the top line of our direction-cosines table, namely,

Direction Cosines for T6 Coordinates
~\gamma_{ni} = h_n \biggl( \frac{\partial \lambda_n}{\partial x_i}\biggr)

~n ~i = x, y, z
~1  

~x\ell_{3D}
 

~q^2 y \ell_{3D} ~p^2 z \ell_{3D}
~2

 
---
 

 
---
 

 
---
 

~3

 
---
 

 
---
 

 
---
 

Other Coordinate Pair in the Tangent Plane

Let's focus on a particular point on the ~\lambda_1 = constant surface, ~(x_0, y_0, z_0), that necessarily satisfies the function, ~F(x_0, y_0, z_0) = 0. We have already derived the expression for the unit vector that is normal to the ellipsoidal surface at this point, namely,

~\hat{e}_1

~\equiv

~
\hat\imath (x_0 \ell_{3D}) + \hat\jmath (q^2y_0 \ell_{3D}) + \hat\jmath (p^2 z_0 \ell_{3D}) \, ,

where, for this specific point on the surface,

~\ell_{3D}

~=

~\biggl[ x_0^2 + q^4y_0^2 + p^4 z_0^2 \biggr]^{- 1 / 2} \, .


Tangent Plane

The two-dimensional plane that is tangent to the ~\lambda_1 = constant surface at this point is given by the expression,

~0

~=

~
(x - x_0) \biggl[ \frac{\partial \lambda_1}{\partial x} \biggr]_0 
+ (y - y_0) \biggl[\frac{\partial \lambda_1}{\partial y} \biggr]_0  
+ (z - z_0) \biggl[\frac{\partial \lambda_1}{\partial z} \biggr]_0

 

~=

~
(x - x_0) \biggl[ \frac{\partial F}{\partial x} \biggr]_0  
+ (y - y_0) \biggl[\frac{\partial F}{\partial y} \biggr]_0  
+ (z - z_0) \biggl[ \frac{\partial F}{\partial z} \biggr]_0

 

~=

~
(x - x_0) \biggl( \frac{x}{\lambda_1}\biggr)_0 + (y - y_0)\biggl( \frac{q^2 y  }{ \lambda_1 } \biggr)_0 + (z - z_0)\biggl( \frac{ p^2z }{ \lambda_1 } \biggr)_0

~\Rightarrow~~~
x \biggl( \frac{x}{\lambda_1}\biggr)_0 + y \biggl( \frac{q^2 y  }{ \lambda_1 } \biggr)_0 + z \biggl( \frac{ p^2z }{ \lambda_1 } \biggr)_0

~=

~
x_0 \biggl( \frac{x}{\lambda_1}\biggr)_0 + y_0 \biggl( \frac{q^2 y  }{ \lambda_1 } \biggr)_0 + z_0 \biggl( \frac{ p^2z }{ \lambda_1 } \biggr)_0

~\Rightarrow~~~
x x_0  + q^2 y y_0  + p^2 z z_0

~=

~
x_0^2  +  q^2 y_0^2 + p^2 z_0^2

~\Rightarrow~~~
x x_0  + q^2 y y_0  + p^2 z z_0

~=

~
(\lambda_1^2)_0 \, .

Fix the value of ~\lambda_1. This means that the relevant ellipsoidal surface is defined by the expression,

~\lambda_1^2

~=

~x^2 + q^2y^2 + p^2z^2 \, .

If ~z = 0, the semi-major axis of the relevant x-y ellipse is ~\lambda_1, and the square of the semi-minor axis is ~\lambda_1^2/q^2. At any other value, ~z = z_0 < c, the square of the semi-major axis of the relevant x-y ellipse is, ~(\lambda_1^2 - p^2z_0^2) and the square of the corresponding semi-minor axis is, ~(\lambda_1^2 - p^2z_0^2)/q^2. Now, for any chosen ~x_0^2 \le (\lambda_1^2 - p^2z_0^2), the y-coordinate of the point on the ~\lambda_1 surface is given by the expression,

~y_0^2

~=

~\frac{1}{q^2}\biggl[ \lambda_1^2 - p^2 z_0 -x_0^2 \biggr] \, .

The slope of the line that lies in the z = z0 plane and that is tangent to the ellipsoidal surface at ~(x_0, y_0) is,

~m \equiv \frac{dy}{dx}\biggr|_{z_0}

~=

~- \frac{x_0}{q^2y_0}

Speculation1

Building on our experience developing T3 Coordinates and, more recently, T5 Coordinates, let's define the two "angles,"

~\Zeta

~\equiv

~\sinh^{-1}\biggl(\frac{qy}{x} \biggr)

      and,      

~\Upsilon

~\equiv

~\sinh^{-1}\biggl(\frac{pz}{x} \biggr) \, ,

in which case we can write,

~\lambda_1^2

~=

~x^2(\cosh^2\Zeta + \sinh^2\Upsilon)\, .

We speculate that the other two orthogonal coordinates may be defined by the expressions,

~\lambda_2

~\equiv

~x \biggl[ \sinh\Zeta \biggr]^{1/(1-q^2)} 
=
x \biggl[ \frac{qy}{x}\biggr]^{1/(1-q^2)}
=
x \biggl[ \frac{x}{qy}\biggr]^{1/(q^2-1)}
=
\biggl[ \frac{x^{q^2}}{qy}\biggr]^{1/(q^2-1)}
\, ,

~\lambda_3

~\equiv

~x \biggl[ \sinh\Upsilon \biggr]^{1/(1-p^2)} 
=
x \biggl[ \frac{pz}{x}\biggr]^{1/(1-p^2)} 
=
x \biggl[ \frac{x}{pz}\biggr]^{1/(p^2-1)} 
=
\biggl[ \frac{x^{p^2}}{pz}\biggr]^{1/(p^2-1)} 
\, .

Some relevant partial derivatives are,

~\frac{\partial \lambda_2}{\partial x}

~=

~\biggl[ \frac{1}{qy}\biggr]^{1/(q^2-1)} \biggl[ \frac{q^2}{q^2-1} \biggr]x^{1/(q^2-1)}
=
\biggl[ \frac{q^2}{q^2-1} \biggr]\biggl[ \frac{x}{qy}\biggr]^{1/(q^2-1)} 
=
\biggl[ \frac{q^2}{q^2-1} \biggr]\frac{\lambda_2}{x} 
\, ;

~\frac{\partial \lambda_2}{\partial y}

~=

~\biggl[ \frac{x^{q^2}}{q}\biggr]^{1/(q^2-1)} \biggl[ \frac{1}{1-q^2} \biggr] y^{q^2/(1-q^2)}
=
- \biggl[ \frac{1}{q^2-1} \biggr] \frac{\lambda_2}{y} 
\, ;

~\frac{\partial \lambda_3}{\partial x}

~=

~
\biggl[ \frac{p^2}{p^2-1} \biggr]\frac{\lambda_3}{x} 
\, ;

~\frac{\partial \lambda_3}{\partial z}

~=

~
- \biggl[ \frac{1}{p^2-1} \biggr] \frac{\lambda_3}{z} 
\, .

And the associated scale factors are,

~h_2^2

~=

~
\biggl\{ \biggl[ \biggl( \frac{q^2}{q^2-1} \biggr)\frac{\lambda_2}{x} \biggr]^2 + \biggl[ - \biggl( \frac{1}{q^2-1} \biggr) \frac{\lambda_2}{y} \biggr]^2 \biggr\}^{-1}

 

~=

~
\biggl\{ \biggl( \frac{q^2}{q^2-1} \biggr)^2 \frac{\lambda_2^2}{x^2}  + \biggl( \frac{1}{q^2-1} \biggr)^2 \frac{\lambda_2^2}{y^2} \biggr\}^{-1}

 

~=

~
\biggl\{x^2 + q^4 y^2 \biggr\}^{-1}
\biggl[ \frac{(q^2 - 1)^2x^2 y^2}{\lambda_2^2} \biggr] \, ;

~h_3^2

~=

~
\biggl\{x^2 + p^4 z^2 \biggr\}^{-1}
\biggl[ \frac{(p^2 - 1)^2x^2 z^2}{\lambda_3^2} \biggr] \, .

We can now fill in the rest of our direction-cosines table, namely,

Direction Cosines for T6 Coordinates
~\gamma_{ni} = h_n \biggl( \frac{\partial \lambda_n}{\partial x_i}\biggr)

~n ~i = x, y, z
~1  

~x\ell_{3D}
 

~q^2 y \ell_{3D} ~p^2 z \ell_{3D}
~2

~q^2 y \ell_q

~-x\ell_q

~0

~3

~p^2 z \ell_p

~0

~-x\ell_p

Hence,

~\hat{e}_2

~=

~
\hat\imath \gamma_{21}
+ \hat\jmath \gamma_{22}
+\hat{k} \gamma_{23}
=
\hat\imath (q^2y\ell_q)
- \hat\jmath (x\ell_q) \, ;

~\hat{e}_3

~=

~
\hat\imath \gamma_{31}
+ \hat\jmath \gamma_{32}
+\hat{k} \gamma_{33}
=
\hat\imath (p^2z\ell_p)
-\hat{k} (x\ell_p) \, .

Check:

~\hat{e}_2 \cdot \hat{e}_2

~=

~
(q^2y\ell_q)^2
+ (x\ell_q)^2
=
1 \, ;

~\hat{e}_3 \cdot \hat{e}_3

~=

~
(p^2z\ell_p)^2
+ (x\ell_p)^2
=
1 \, ;

~\hat{e}_2 \cdot \hat{e}_3

~=

~
(q^2y\ell_q)(p^2z\ell_p) \ne 0 \, .

Speculation2

Try,

~\lambda_2

~=

~
\frac{x}{y^{1/q^2} z^{1/p^2}} \, ,

in which case,

~\frac{\partial \lambda_2}{\partial x}

~=

~
\frac{\lambda_2}{x} \, ,

~\frac{\partial \lambda_2}{\partial y}

~=

~
\frac{x}{z^{1/p^2}} \biggl(-\frac{1}{q^2}\biggr) y^{-1/q^2 - 1} 
=
-\frac{\lambda_2}{q^2 y}
\, ,

~\frac{\partial \lambda_2}{\partial z}

~=

~
-\frac{\lambda_2}{p^2 z}
\, .

The associated scale factor is, then,

~h_2^2

~=

~\biggl[
\biggl( \frac{\partial \lambda_2}{\partial x} \biggr)^2
+ 
\biggl( \frac{\partial \lambda_2}{\partial y} \biggr)^2
+ 
\biggl( \frac{\partial \lambda_2}{\partial z} \biggr)^2
\biggr]^{-1}

 

~=

~\biggl[
\biggl( \frac{ \lambda_2}{x} \biggr)^2
+ 
\biggl( -\frac{\lambda_2}{q^2y} \biggr)^2
+ 
\biggl( - \frac{\lambda_2}{p^2z} \biggr)^2
\biggr]^{-1}

Speculation3

Try,

~\lambda_2

~=

~
\frac{(x+p^2 z)^{1 / 2}}{y^{1/q^2} } \, ,

in which case,

~\frac{\partial \lambda_2}{\partial x}

~=

~
\frac{1}{2y^{1/q^2}}\biggl(x + p^2z\biggr)^{- 1 / 2}
=
\frac{\lambda_2}{2(x + p^2z) }
\, ,

~\frac{\partial \lambda_2}{\partial y}

~=

~
-\frac{\lambda_2}{q^2y}
\, ,

~\frac{\partial \lambda_2}{\partial z}

~=

~

\, .

Speculation4

Here we stick with the primary (radial-like) coordinate as defined above; for example,

~\lambda_1

~\equiv

~(x^2 + q^2 y^2 + p^2 z^2)^{1 / 2} \, ,

~h_1

~=

~\lambda_1 \ell_{3D} \, ,

~\ell_{3D}

~\equiv

~[ x^2 + q^4y^2 + p^4 z^2 ]^{- 1 / 2} \, ,

~\hat{e}_1

~=

~
\hat\imath (x \ell_{3D}) + \hat\jmath (q^2y \ell_{3D}) + \hat{k} (p^2 z \ell_{3D}) \, .

Note that, ~\hat{e}_1 \cdot \hat{e}_1 = 1, which means that this is, indeed, a properly normalized unit vector.

Then, drawing from our earliest discussions of "T1 Coordinates", we'll try defining the second coordinate as,

~\lambda_3

~\equiv

~
\tan^{-1} u \, ,
      where,

~u

~\equiv

\frac{y^{1/q^2}}{x} \, .

The relevant partial derivatives are,

~\frac{\partial \lambda_3}{\partial x}

~=

~
\frac{1}{1 + u^2} \biggl[ - \frac{y^{1/q^2}}{x^2} \biggr]
=
- \biggl[ \frac{u}{1 + u^2}\biggr]\frac{1}{x} \, ,

~\frac{\partial \lambda_3}{\partial y}

~=

~
\frac{1}{1 + u^2} \biggl[ \frac{y^{(1/q^2-1)}}{q^2x^2} \biggr]
=
\biggl[ \frac{u}{1 + u^2}\biggr]\frac{1}{q^2y} \, ,

which means that,

~h_3^2

~=

~
\biggl[ \biggl( \frac{\partial \lambda_3}{\partial x} \biggr)^2 + \biggl( \frac{\partial \lambda_3}{\partial y} \biggr)^2 \biggr]^{-1}

 

~=

~
\biggl[ \frac{u}{1 + u^2}\biggr]^{-2} \biggl[ \frac{1}{x^2} + \frac{1}{q^4y^2} \biggr]^{-1}

 

~=

~
\biggl[ \frac{1 + u^2}{u}\biggr]^{2} \biggl[ \frac{x^2 + q^4y^2}{x^2q^4y^2} \biggr]^{-1}

~\Rightarrow~~~h_3

~=

~
\biggl[ \frac{1 + u^2}{u}\biggr]xq^2 y \ell_q \, ,
      where,

~\ell_q

~\equiv

~[x^2 + q^4 y^2]^{-1 / 2} \, .

The third row of direction cosines can now be filled in to give,

Direction Cosines for T6 Coordinates
~\gamma_{ni} = h_n \biggl( \frac{\partial \lambda_n}{\partial x_i}\biggr)

~n ~i = x, y, z
~1  

~x\ell_{3D}
 

~q^2 y \ell_{3D} ~p^2 z \ell_{3D}
~2

 
---
 

 
---
 

 
---
 

~3

~-q^2 y \ell_q

~x \ell_q

~0

which means that the associated unit vector is,

~\hat{e}_3

~=

~
-\hat\imath (q^2 y \ell_{q}) + \hat\jmath (x \ell_{q})  \, .

Note that, ~\hat{e}_3 \cdot \hat{e}_3 = 1, which means that this also is a properly normalized unit vector. Note, as well, that the dot product between our "first" and "third" unit vectors should be zero if they are indeed orthogonal to each other. Let's see …

~\hat{e}_3 \cdot \hat{e}_1

~=

~
(- q^2y \ell_q)x\ell_{3D} + (x\ell_q) q^2y\ell_{3D} = 0 \, .

Q.E.D.

Now, even though we have not yet determined the proper expression for the "second" orthogonal coordinate, ~\lambda_2, we should be able to obtain an expression for its associated unit vector from the cross product of the "third" and "first" unit vectors. Specifically we find,

~\hat{e}_3 \times \hat{e}_1

~=

~

See Also

Whitworth's (1981) Isothermal Free-Energy Surface

© 2014 - 2020 by Joel E. Tohline
|   H_Book Home   |   YouTube   |
Appendices: | Equations | Variables | References | Ramblings | Images | myphys.lsu | ADS |

Personal tools