MathJax/LaTeX Mathematics Formulae Rendering

[Eli]

Fully adopted from this website -- Extract From Struik:


This procedure is simply a generalization of the method used in Sects. 1-3 and 1-4 to obtain the equations of the osculating plane and the osculating circle. Let $f(u)$ near $P(u=u_0)$ have finite derivatives $f^{(i)}(u_0)$, $i = 1, 2, \ldots, n+1$. Then
if we take $u=u_1$ at $A$ and write $h = u_1 - u_0$, then there exists a Taylor development of $f(u)$ of the form (compare Eq. (1-5)):

$$ f(u_1) = f(u_0) + hf'(u_0)+{h^2\over 2!}f''(u_0) + \cdots + {h^{n+1}\over (n+1)!}f^{(n+1)}(u_0) + o(h^{n+1}). $$

Here, $f(u_0)=0$ since $P$ lies on $\Sigma_2$, and $h$ is of order $AP$ (see theorem Sec. 1-2); $f(u_1)$ is of order $AD$.
Hence necessary and sufficient conditions that the surface has a contact of order $n$ at $P$ with the curve are that at $P$ the relations hold:

$$ f(u) = f'(u) = f''(u) = \cdots = f^{(n)}(u) = 0;\quad f^{(n+1)}(u) \ne 0. $$

The converse problem is somewhat more complicated: Find the curves which admit a given curve $C$ as involute. Such curves are called
evolutes of $C$ (German: Evolute; French). Their tangents are normal to $C({\bf x})$ and we can therefore write the equation of the evolute ${\bf y}$ (Fig. 1-34):
$$ {\bf y} = {\bf x} + a_1{\bf n} + a_2{\bf b}. $$

Hence

$$ {d{\bf y}\over ds} = {\bf t}(1-a_1\kappa) + {\bf n}\left({da_1\over ds}-\tau a_2\right) + {\bf b}\left({da_2\over ds}+\tau a_1\right) $$

must have the direction of $a_1{\bf n} + a_2{\bf b}$, this tangent to the evolute:

$$ \kappa = 1/a, \qquad R= a_1, $$

$$ {{da_1\over ds} - \tau a_2\over a_1} = {{da_2\over ds}+\tau a_1\over a_2}, $$

which can be written in the form:

$$ {a_2{dR\over ds} - R{da_2\over ds} \over a_2^2 + R^2} = \tau. $$

This expression can be integrated:

$$ \tan^{-1}{R\over a_2} = \int \tau\,ds + {\rm const}, $$ or

$$ a_2 = R\left[{\rm cot}\left(\int \tau\,ds + {\rm const}\right)\right]. $$

The equation of the evolute is:

$$ {\bf y} = {\bf x} + R\left[{\bf n} + {\rm cot}\left(\int \tau\,ds + {\rm const}\right){\bf b}\right]. $$

If $P(u,v)$ and $Q(u,v)$ are two functions of $u$ and $v$ on a surface, then according to Green's theorem and the expression in Chapter 2, Eq. (3-4) for the element area:

$$ \int_C P\,du + Q\, dv = \int\!\!\!\int_A \left({\partial Q\over \partial u} - {\partial P\over \partial v}\right) {1\over \sqrt{EG-F^2}}\,dA, $$

where $dA$ is the element of area of the region $R$ enclosed by the curve $C$. With the aid of this theorem we shall evaluate

$$ \int_C \kappa_g\,ds, $$

where $\kappa_g$ is the geodesic curvature of the curve $C$. If $C$ at a point $P$ makes the angle $\theta$ with the coordinate curve $v = {\rm constant}$ and if the coordinate curves are orthogonal, then, according to Liouville's formula (1-13):

$$ \kappa_g\,ds = d\theta + \kappa_1(\cos\theta)\,ds + \kappa_2(\sin\theta)\,ds. $$

Here, $\kappa_1$ and $\kappa_2$ are the geodesic curvatures of the curves $v = {\rm constant}$ and $u = {\rm constant}$ respectively. Since

$$ \cos\theta\,ds = \sqrt{E}\,du, \qquad \sin\theta\,ds = \sqrt{G}\,dv, $$

we find by application of Green's theorem:

$$ \int_C\kappa_g\,ds = \int_C d\theta + \int\!\!\!\int_A\left({\partial\over\partial u} \left(\kappa_2\sqrt{G}\,\right) - {\partial\over \partial v}\left(\kappa_1\sqrt{E}\,\right)\right)\,du\,dv. $$

The Gaussian curvature can be written, according to Chapter 3, Eq. (3-7),

$$ K = -{1\over 2\sqrt{EG}} \left[{\partial\over\partial u}{G_u\over \sqrt{EG}} + {\partial\over\partial v}{E_v\over\sqrt{EG}}\right] ={1\over\sqrt{EG}}\left[ -{\partial\over\partial u} \left(\kappa_2\sqrt{G}\,\right) + {\partial\over\partial v} \left(\kappa_1\sqrt{E}\,\right)\right],
$$

so we obtain the formula

$$ \int_C\kappa_g\,ds = \int_C d\theta - \int\!\!\!\int_A K\,dA. $$

The integral $\int\!\!\int_A K\,dA$ is known as the total or integral curvature, or curvature integral, of the region $R$, the name by which Gauss introduced it.

[Eli]

We can use bmatrix from amsmath with \cdots and \vdots for horizontal and vertical dots in vectors and matrices:

\begin{align}
v &= \begin{bmatrix}
x_{1} \\
x_{2} \\
\vdots \\
x_{m}
\end{bmatrix},
\end{align}


\begin{align}
X &= (x_{1},x_{2},\cdots, x_{N})
\begin{pmatrix}
\begin{bmatrix}
ax_{0} + bx_{1} \\
\vdots \\
ax_{n-1}+bx_{n}
\end{bmatrix} -
\begin{bmatrix}
z_{1} \\
\vdots \\
z_{n}
\end{bmatrix}
\end{pmatrix}.
\end{align}