跳到主要内容

Higher-Order Derivatives

Definition

If a function ff is differentiable everywhere in an interval II, then the function

f(1):=f:IR\begin{equation*} f^{(1)} := f' : I \to \mathbb{R} \end{equation*}

is called the first derivative of ff. We denote f(0):=ff^{(0)} := f.

By induction, if the nn-th derivative f(n)f^{(n)} of ff exists, and f(n)f^{(n)} is differentiable at x0x_0, then we define

f(n+1)(x0)=(f(n))(x0).\begin{equation*} f^{(n+1)}(x_0) = \big(f^{(n)}\big)'(x_0). \end{equation*}

We call f(n+1)(x0)f^{(n+1)}(x_0) the (n+1)(n+1)-th derivative of ff at x0x_0. The second and third derivatives are denoted by ff'' and ff''', respectively.

We also use the following notations:

fCn(I)means f(n) is continuous on I;\begin{equation*} f \in \mathcal{C}^n(I) \quad \text{means } f^{(n)} \text{ is continuous on } I; \end{equation*} fC(I)means f has derivatives of all orders on I.\begin{equation*} f \in \mathcal{C}^\infty(I) \quad \text{means } f \text{ has derivatives of all orders on } I. \end{equation*}

Four Basic Operations

Let f,gf, g have nn-th derivatives at x0x_0. Then λf+μg\lambda f + \mu g and fgfg also have nn-th derivatives at x0x_0, and we have:

(λf+μg)(n)(x0)=λf(n)(x0)+μg(n)(x0)\begin{equation*} (\lambda f + \mu g)^{(n)}(x_0) = \lambda f^{(n)}(x_0) + \mu g^{(n)}(x_0) \end{equation*} (fg)(n)(x0)=k=0nCnkf(k)(x0)g(nk)(x0)\begin{equation*} (fg)^{(n)}(x_0) = \sum_{k=0}^{n} C_n^k\, f^{(k)}(x_0)\, g^{(n-k)}(x_0) \end{equation*}

If ff is nn-times differentiable at x0x_0, and gg is nn-times differentiable at y0=f(x0)y_0 = f(x_0), then the composite function gfg \circ f is nn-times differentiable at x0x_0.

Let f,gf, g be nn-times differentiable at x0x_0, and suppose g(x0)0g(x_0) \neq 0. Then fg\dfrac{f}{g} is nn-times differentiable at x0x_0.

The functions arctanx\arctan x, arcsinx\arcsin x, and arccosx\arccos x are all C\mathcal{C}^\infty functions. Their derivatives are given by:

(arctanx)=11+x2,(arcsinx)=11x2,(arccosx)=11x2.\begin{equation*} (\arctan x)' = \frac{1}{1 + x^2}, \quad (\arcsin x)' = \frac{1}{\sqrt{1 - x^2}}, \quad (\arccos x)' = \frac{-1}{\sqrt{1 - x^2}}. \end{equation*}

Suppose ff is nn-times differentiable on an interval II, and f(x)0f'(x) \ne 0. (From continuity, ff has a continuous inverse function on II.) Then f1f^{-1} is nn-times differentiable on the interval f(I)f(I).

If fCn(I) and f(x)0, then f1Cn(f(I)).\begin{equation*} \text{If } f \in C^n(I) \text{ and } f'(x) \ne 0, \text{ then } f^{-1} \in C^n(f(I)). \end{equation*}

Corollay: All elementary functions are CC^{\infty} functions.

Curvature Center, Radius, and Curvature of a Plane Curve

Solution. Let the curve be parameterized as x=x(t),y=y(t)x = x(t), \, y = y(t).

The tangent line of the curve is

Yy(t)=y(t)x(t)[Xx(t)]\begin{equation*} Y - y(t) = \frac{y'(t)}{x'(t)} \,[X - x(t)] \end{equation*}

The normal line is

y(t)[Yy(t)]=x(t)[Xx(t)]\begin{equation*} y'(t) \,[Y - y(t)] = -x'(t) \,[X - x(t)] \end{equation*}

At a nearby point, the normal line is

y(t+h)[Yy(t+h)]=x(t+h)[Xx(t+h)]\begin{equation*} y'(t+h) \,[Y - y(t+h)] = -x'(t+h) \,[X - x(t+h)] \end{equation*}

Expanding by Taylor’s theorem:

[y(t)+hy(t)+o(h)][Yy(t)y(t)h+o(h)]=[x(t)+x(t)h+o(h)][Xx(t)x(t)ho(h)]\begin{equation*} [y'(t) + h y''(t) + o(h)] \,[Y - y(t) - y'(t)h + o(h)] = - [x'(t) + x''(t)h + o(h)] \,[X - x(t) - x'(t)h - o(h)] \end{equation*}   y(t)[Yy(t)]+x(t)[Xx(t)]+h[y(t)(Yy(t))+x(t)(Xx(t))(y(t))2(x(t))2]+o(h)=0\begin{equation*} \Rightarrow \; y'(t) \,[Y - y(t)] + x'(t) \,[X - x(t)] + h \,[\,y''(t)(Y - y(t)) + x''(t)(X - x(t)) - (y'(t))^2 - (x'(t))^2\,] + o(h) = 0 \end{equation*}

When h0h \to 0, the intersection of the two normals satisfies:

{y(t)[Yy(t)]+x(t)[Xx(t)]=0,y(t)[Yy(t)]+x(t)[Xx(t)]=(x(t))2+(y(t))2.\begin{equation*} \begin{cases} y'(t) \,[Y - y(t)] + x'(t) \,[X - x(t)] = 0, \\ y''(t) \,[Y - y(t)] + x''(t) \,[X - x(t)] = (x'(t))^2 + (y'(t))^2. \end{cases} \end{equation*}

Solving for the center of curvature (X,Y)(X, Y):

{X=x(t)y(t)[(x(t))2+(y(t))2]x(t)x(t)y(t)y(t),Y=y(t)+x(t)[(x(t))2+(y(t))2]x(t)x(t)y(t)y(t).\begin{equation*} \begin{cases} X = x(t) - \dfrac{y'(t) \,\big[(x'(t))^2 + (y'(t))^2\big]} {\begin{vmatrix} x'(t) & x''(t) \\[4pt] y'(t) & y''(t) \end{vmatrix}}, \\[16pt] Y = y(t) + \dfrac{x'(t) \,\big[(x'(t))^2 + (y'(t))^2\big]} {\begin{vmatrix} x'(t) & x''(t) \\[4pt] y'(t) & y''(t) \end{vmatrix}}. \end{cases} \end{equation*}

The radius of curvature is the distance from (X,Y)(X, Y) to (x(t),y(t))(x(t), y(t)):

R=[Xx(t)]2+[Yy(t)]2=((x(t))2+(y(t))2)3det(x(t)x(t)y(t)y(t))\begin{equation*} R = \sqrt{[X - x(t)]^2 + [Y - y(t)]^2} = \frac{\big(\sqrt{(x'(t))^2 + (y'(t))^2}\big)^3} {\Big| \det \begin{pmatrix} x'(t) & x''(t) \\ y'(t) & y''(t) \end{pmatrix} \Big|} \end{equation*}

The curvature is the reciprocal of the radius of curvature:

κ=det(x(t)x(t)y(t)y(t))[(x(t))2+(y(t))2]3/2\begin{equation*} \kappa = \frac{\Big| \det \begin{pmatrix} x'(t) & x''(t) \\ y'(t) & y''(t) \end{pmatrix} \Big|} {\big[(x'(t))^2 + (y'(t))^2\big]^{3/2}} \end{equation*}