개념 해석 · Concepts

NOTE 편도함수의 동기 — 체감온도 Motivating Partial DerivativesTable 1 · p.999
한국어

체감온도 $I$는 실제 기온 $T$와 상대습도 $H$에 따라 사람이 느끼는 온도로, 두 변수의 함수 $I=f(T,H)$이다. 아래 표는 미국 기상청(NWS)이 작성한 체감온도 표의 일부를 발췌한 것이다.

표 1 · 체감온도 $I$ (°F) — 실제 기온 $T$ (°F)와 상대습도 $H$ (%)의 함수
$T$$H$ (%)
505560657075808590
909698100103106109112115119
92100103105108112115119123128
94104107111114118122127132137
96109113116121125130135141146
98114118123127133138144150157
100119124129135141147154161168

습도를 $H=70$으로 고정하고 그 열만 따라 읽으면 남는 것은 1변수 함수 $g(T)=f(T,70)$이다. $T=96$ 양쪽에서 차분을 잡으면

$$\frac{f(98,70)-f(96,70)}{2}=\frac{133-125}{2}=4,$$
$$\frac{f(94,70)-f(96,70)}{-2}=\frac{118-125}{-2}=3.5,$$

이고, 두 값을 평균해 $g'(96)\approx 3.75$로 추정한다. 반대로 기온을 $T=96$으로 고정하고 그 행을 따라 읽으면 1변수 함수 $G(H)=f(96,H)$를 얻으며,

$$\frac{f(96,75)-f(96,70)}{5}=\frac{130-125}{5}=1,$$
$$\frac{f(96,65)-f(96,70)}{-5}=\frac{121-125}{-5}=0.8,$$

이므로 $G'(70)\approx 0.9$이다. 이 두 추정값은 아래에서 도입할 표기로 각각 $f_T(96,70)\approx 3.75$, $f_H(96,70)\approx 0.9$로 쓰이며, 기온 96°F·습도 70%에서 습도를 고정한 채 기온이 1°F 오르면 체감온도는 약 3.75°F, 기온을 고정한 채 습도가 1%p 오르면 약 0.9°F 상승함을 뜻한다.

English

The heat index $I$ is the temperature a person perceives when the actual air temperature is $T$ and the relative humidity is $H$; it therefore depends on two variables, $I=f(T,H)$. The values below are excerpted from a table compiled by the National Weather Service.

Table 1 · Heat index $I$ (°F) as a function of actual temperature $T$ (°F) and relative humidity $H$ (%)
$T$$H$ (%)
505560657075808590
909698100103106109112115119
92100103105108112115119123128
94104107111114118122127132137
96109113116121125130135141146
98114118123127133138144150157
100119124129135141147154161168

Freeze the humidity at $H=70$ and read down that column: what is left is the one-variable function $g(T)=f(T,70)$. Two difference quotients on either side of $T=96$ give

$$\frac{f(98,70)-f(96,70)}{2}=\frac{133-125}{2}=4,$$
$$\frac{f(94,70)-f(96,70)}{-2}=\frac{118-125}{-2}=3.5,$$

and averaging them estimates $g'(96)\approx 3.75$. Freeze the temperature at $T=96$ instead and read across that row: the one-variable function is $G(H)=f(96,H)$, and

$$\frac{f(96,75)-f(96,70)}{5}=\frac{130-125}{5}=1,$$
$$\frac{f(96,65)-f(96,70)}{-5}=\frac{121-125}{-5}=0.8,$$

so $G'(70)\approx 0.9$. In the notation introduced below these two estimates read $f_T(96,70)\approx 3.75$ and $f_H(96,70)\approx 0.9$: at 96 °F and 70 % humidity the perceived temperature climbs by about 3.75 °F per degree of actual temperature, and by about 0.9 °F per percentage point of humidity.

DEF 2 한 점에서의 편도함수 Partial Derivatives at a PointEq. (1)–(3) · p.1000
한국어

$f$가 이변수 $x,y$의 함수라 하자. $y$를 상수 $b$로 고정하고 $x$만 움직이면 남는 것은 1변수 함수 $g(x)=f(x,b)$이다. $g$가 $a$에서 미분가능하면 그 값 $g'(a)$를 점 $(a,b)$에서 $x$에 대한 $f$의 편도함수라 하고 $f_x(a,b)$로 쓴다.

$$f_x(a,b)=g'(a)\quad\text{단, }g(x)=f(x,b).\tag{1}$$

$g'(a)$를 차분의 극한으로 풀어 쓰면 (1)은 $f$만으로 표현된 극한이 된다.

$$f_x(a,b)=\lim_{h\to 0}\frac{f(a+h,b)-f(a,b)}{h}.\tag{2}$$

두 변수의 역할을 바꾸어, 이번에는 $x=a$를 고정하고 $G(y)=f(a,y)$를 $b$에서 미분하면 점 $(a,b)$에서 $y$에 대한 $f$의 편도함수를 얻는다.

$$f_y(a,b)=\lim_{h\to 0}\frac{f(a,b+h)-f(a,b)}{h}.\tag{3}$$

이 표기로 앞서 구한 두 체감온도 추정값은 정확히 $f_T(96,70)\approx 3.75$, $f_H(96,70)\approx 0.9$이다.

English

Let $f$ be a function of the two variables $x$ and $y$. Hold $y$ at the constant value $b$ and let only $x$ vary: what remains is the single-variable function $g(x)=f(x,b)$. If $g$ is differentiable at $a$, the number $g'(a)$ is called the partial derivative of $f$ with respect to $x$ at $(a,b)$ and is written $f_x(a,b)$:

$$f_x(a,b)=g'(a)\quad\text{where }g(x)=f(x,b).\tag{1}$$

Writing $g'(a)$ out as a limit of difference quotients turns (1) into a limit in $f$ alone:

$$f_x(a,b)=\lim_{h\to 0}\frac{f(a+h,b)-f(a,b)}{h}.\tag{2}$$

Interchanging the roles of the two variables — holding $x=a$ fixed and differentiating $G(y)=f(a,y)$ at $b$ — gives the partial derivative of $f$ with respect to $y$ at $(a,b)$:

$$f_y(a,b)=\lim_{h\to 0}\frac{f(a,b+h)-f(a,b)}{h}.\tag{3}$$

With this notation the two heat-index estimates above are exactly $f_T(96,70)\approx 3.75$ and $f_H(96,70)\approx 0.9$.

DEF 3 함수로서의 편도함수 Partial Derivatives as FunctionsDefinition 4 · p.1001
한국어

이제 (2)와 (3)에서 기준점 $(a,b)$를 움직여 보자. 두 식은 점마다 하나의 수를 대응시키므로, 편도함수 $f_x,f_y$ 자체가 다시 이변수 함수가 된다.

정의 4. 이변수함수 $f$에 대해, $f$의 편도함수는 다음 두 함수 $f_x$와 $f_y$로 정의된다.

$$f_x(x,y)=\lim_{h\to 0}\frac{f(x+h,y)-f(x,y)}{h},$$
$$f_y(x,y)=\lim_{h\to 0}\frac{f(x,y+h)-f(x,y)}{h},$$

여기서 각 편도함수는 해당 극한이 존재하는 점에서만 정의된다.

English

Now let the base point $(a,b)$ in (2) and (3) move. Each formula then assigns a number to a point, so each partial derivative is itself a function of two variables.

Definition 4. For a function $f$ of two variables, the partial derivatives of $f$ are the two functions $f_x$ and $f_y$ given by

$$f_x(x,y)=\lim_{h\to 0}\frac{f(x+h,y)-f(x,y)}{h},$$
$$f_y(x,y)=\lim_{h\to 0}\frac{f(x,y+h)-f(x,y)}{h},$$

Each partial derivative is defined exactly at those points where the corresponding limit exists.

REM 4 편도함수의 표기 Notations for Partial DerivativesNotations · p.1001
한국어

같은 대상을 가리키는 기호가 여러 개 쓰인다. $z=f(x,y)$일 때,

$$f_x(x,y)=f_x=\frac{\partial f}{\partial x}=\frac{\partial}{\partial x}f(x,y)=\frac{\partial z}{\partial x}=f_1=D_1f=D_xf,$$
$$f_y(x,y)=f_y=\frac{\partial f}{\partial y}=\frac{\partial}{\partial y}f(x,y)=\frac{\partial z}{\partial y}=f_2=D_2f=D_yf.$$

아래첨자 $1,2$는 $f$의 몇 번째 자리를 미분했는지를 나타내며, 변수에 따로 이름이 없을 때 특히 편리하다.

주의. 기호 $\partial f/\partial x$는 그 자체로 하나의 기호일 뿐이다. 1변수의 $dy/dx$와 달리, 미분 $\partial f$와 $\partial x$의 비로 나누어 해석할 수 없다.

English

Several symbols are in use for the same object. For $z=f(x,y)$,

$$f_x(x,y)=f_x=\frac{\partial f}{\partial x}=\frac{\partial}{\partial x}f(x,y)=\frac{\partial z}{\partial x}=f_1=D_1f=D_xf,$$
$$f_y(x,y)=f_y=\frac{\partial f}{\partial y}=\frac{\partial}{\partial y}f(x,y)=\frac{\partial z}{\partial y}=f_2=D_2f=D_yf.$$

The subscripts $1$ and $2$ record which slot of $f$ is being differentiated, which is convenient when the variables have no names.

Warning. The symbol $\partial f/\partial x$ is a single indivisible piece of notation; unlike $dy/dx$ in one variable, it cannot be read as a quotient of two differentials $\partial f$ and $\partial x$.

REM 5 편도함수 계산 규칙 Rule for Finding Partial DerivativesRule · p.1001
한국어

새로운 미분법이 필요한 것은 아니다. 식 (1)은 편도함수가 다른 변수를 고정한 뒤의 보통의 1변수 도함수임을 말해 준다. 따라서 $z=f(x,y)$에 대해 다음이 성립한다.

1. $f_x$를 구하려면 $y$를 상수로 보고 $f(x,y)$를 $x$에 대해 미분한다.

2. $f_y$를 구하려면 $x$를 상수로 보고 $f(x,y)$를 $y$에 대해 미분한다.

따라서 합·곱·몫의 미분법, 연쇄법칙, 음함수 미분법 등 1변수의 모든 미분 규칙이 자유로운 변수에 그대로 적용된다.

English

No new differentiation technique is needed. Equation (1) says that a partial derivative is an ordinary one-variable derivative, taken after the other variable has been frozen. Hence, for $z=f(x,y)$:

1. To find $f_x$, regard $y$ as a constant and differentiate $f(x,y)$ with respect to $x$.

2. To find $f_y$, regard $x$ as a constant and differentiate $f(x,y)$ with respect to $y$.

Consequently every one-variable rule — sum, product, quotient, chain rule, implicit differentiation — applies verbatim to the variable that is left free.

NOTE 기하적 해석 — 자취의 접선 기울기 Geometric InterpretationFigure 1 · p.1002
한국어

곡면 $S$를 $z=f(x,y)$의 그래프라 하고, $c=f(a,b)$로 두면 점 $P(a,b,c)$가 $S$ 위에 있다.

$y=b$로 고정하면 수직평면 $y=b$가 $S$와 만나는 곡선, 즉 $S$의 자취 $C_1$만 남는다. $C_1$은 $g(x)=f(x,b)$의 그래프이므로, $P$에서 $C_1$의 접선 $T_1$의 기울기는 $g'(a)=f_x(a,b)$이다.

마찬가지로 수직평면 $x=a$는 $S$와 만나 자취 $C_2$를 이루며, $C_2$는 $G(y)=f(a,y)$의 그래프이므로 $P$에서 $C_2$의 접선 $T_2$의 기울기는 $G'(b)=f_y(a,b)$이다. 두 자취 모두 $P$를 지난다.

즉 점 $(a,b)$에서의 두 편도함수는 하나의 기하적 의미를 갖는다. 평면 $y=b$와 $x=a$에서의 $S$의 자취에 대해, 점 $P(a,b,c)$에서 그은 접선의 기울기가 바로 그것이다.

z x y 0 S P(a, b, c) (a, b, 0) C₁ C₂ T₁ T₂
English

Let $S$ be the surface $z=f(x,y)$, the graph of $f$, and let $c=f(a,b)$, so that the point $P(a,b,c)$ lies on $S$.

Fixing $y=b$ restricts attention to the curve $C_1$ in which the vertical plane $y=b$ meets $S$ — that is, $C_1$ is the trace of $S$ in that plane. Since $C_1$ is the graph of $g(x)=f(x,b)$, the tangent line $T_1$ to $C_1$ at $P$ has slope $g'(a)=f_x(a,b)$.

Likewise the vertical plane $x=a$ meets $S$ in the trace $C_2$, the graph of $G(y)=f(a,y)$, so the tangent $T_2$ to $C_2$ at $P$ has slope $G'(b)=f_y(a,b)$. Both traces pass through $P$.

So the two partial derivatives at $(a,b)$ have one geometric meaning: they are the slopes at $P(a,b,c)$ of the tangent lines to the traces of $S$ in the planes $y=b$ and $x=a$.

z x y 0 S P(a, b, c) (a, b, 0) C₁ C₂ T₁ T₂
NOTE 변화율로서의 해석 Interpretation as Rates of Changep.1003
한국어

편도함수에는 앞서 체감온도 논의에서 이미 사용한 두 번째 해석이 있다. $z=f(x,y)$에서,

$\partial z/\partial x$는 $y$를 고정했을 때 $z$가 $x$에 대해 변화하는 비율이고, $\partial z/\partial y$는 $x$를 고정했을 때 $z$가 $y$에 대해 변화하는 비율이다.

뒤따르는 응용에서 편도함수는 대개 이 두 해석 가운데 하나로 읽힌다. 자취의 접선 기울기이거나, 나머지 변수를 붙잡아 둔 채의 변화율이다.

English

Partial derivatives carry a second reading, already used in the heat-index discussion. For $z=f(x,y)$:

$\partial z/\partial x$ is the rate of change of $z$ with respect to $x$ while $y$ is held fixed, and $\partial z/\partial y$ is the rate of change of $z$ with respect to $y$ while $x$ is held fixed.

In the applications that follow, a partial derivative is usually read in one of these two ways: as the slope of a trace, or as a rate of change with the remaining variables clamped.

DEF 8 3변수 이상 함수의 편도함수 Partial Derivatives of Functions of Three or More VariablesFunctions of Three or More Variables · p.1004
한국어

정의는 변수가 정확히 두 개일 것을 요구하지 않는다. 세 변수 $x,y,z$의 함수 $f$에 대해 $x$에 대한 편도함수는

$$f_x(x,y,z)=\lim_{h\to 0}\frac{f(x+h,y,z)-f(x,y,z)}{h},$$

이며, 실제로는 $y,z$를 상수로 놓고 $x$에 대해 미분하여 구한다.

일반적으로 $u=f(x_1,x_2,\dots,x_n)$이면 $i$번째 변수에 대한 편도함수는

$$\frac{\partial u}{\partial x_i}=\lim_{h\to 0}\frac{f(x_1,\dots,x_{i-1},x_i+h,x_{i+1},\dots,x_n)-f(x_1,\dots,x_i,\dots,x_n)}{h},$$

이고, 다음과 같이도 쓴다.

$$\frac{\partial u}{\partial x_i}=\frac{\partial f}{\partial x_i}=f_{x_i}=f_i=D_if.$$

$w=f(x,y,z)$에서도 $f_x=\partial w/\partial x$는 여전히 $y,z$를 고정했을 때 $w$가 $x$에 대해 변화하는 비율을 나타낸다. 다만 이런 $f$의 그래프는 4차원 공간에 놓이므로 자취를 취할 곡면이 없어, 앞 카드의 기하적 해석은 더 이상 성립하지 않는다.

English

Nothing in the definition depends on there being exactly two variables. For a function $f$ of the three variables $x$, $y$, $z$, the partial derivative with respect to $x$ is

$$f_x(x,y,z)=\lim_{h\to 0}\frac{f(x+h,y,z)-f(x,y,z)}{h},$$

computed in practice by treating $y$ and $z$ as constants and differentiating with respect to $x$.

In general, if $u=f(x_1,x_2,\dots,x_n)$, the partial derivative with respect to the $i$-th variable is

$$\frac{\partial u}{\partial x_i}=\lim_{h\to 0}\frac{f(x_1,\dots,x_{i-1},x_i+h,x_{i+1},\dots,x_n)-f(x_1,\dots,x_i,\dots,x_n)}{h},$$

also written

$$\frac{\partial u}{\partial x_i}=\frac{\partial f}{\partial x_i}=f_{x_i}=f_i=D_if.$$

For $w=f(x,y,z)$ the quantity $f_x=\partial w/\partial x$ still measures the rate of change of $w$ with respect to $x$ while $y$ and $z$ are held fixed. The geometric reading of the previous card is lost, however: the graph of such an $f$ lives in four-dimensional space, so there is no surface whose trace one could take.

DEF 9 이계·고계 편도함수와 표기 Second and Higher Partial DerivativesHigher Derivatives · p.1005
한국어

$f$가 이변수함수이면 $f_x,f_y$도 다시 이변수함수이므로 한 번 더 미분할 수 있다. 이렇게 얻는 $(f_x)_x,(f_x)_y,(f_y)_x,(f_y)_y$ 네 함수를 $f$의 이계 편도함수라 한다. $z=f(x,y)$일 때 표준 표기는 다음과 같다.

$$(f_x)_x=f_{xx}=f_{11}=\frac{\partial}{\partial x}\left(\frac{\partial f}{\partial x}\right)=\frac{\partial^2 f}{\partial x^2}=\frac{\partial^2 z}{\partial x^2},$$
$$(f_x)_y=f_{xy}=f_{12}=\frac{\partial}{\partial y}\left(\frac{\partial f}{\partial x}\right)=\frac{\partial^2 f}{\partial y\,\partial x}=\frac{\partial^2 z}{\partial y\,\partial x},$$
$$(f_y)_x=f_{yx}=f_{21}=\frac{\partial}{\partial x}\left(\frac{\partial f}{\partial y}\right)=\frac{\partial^2 f}{\partial x\,\partial y}=\frac{\partial^2 z}{\partial x\,\partial y},$$
$$(f_y)_y=f_{yy}=f_{22}=\frac{\partial}{\partial y}\left(\frac{\partial f}{\partial y}\right)=\frac{\partial^2 f}{\partial y^2}=\frac{\partial^2 z}{\partial y^2}.$$

아래첨자의 순서. $f_{xy}$는 $x$로 먼저, $y$로 나중에 미분함을 뜻하고, $f_{yx}$는 그 반대이다. 표기 $\partial^2 f/\partial y\,\partial x$도 같은 순서를 담고 있다. 분모를 오른쪽에서 왼쪽으로 읽어, 오른쪽 끝의 $\partial x$를 먼저 적용하고 $\partial y$를 나중에 적용한다.

English

If $f$ is a function of two variables, then $f_x$ and $f_y$ are again functions of two variables, so each may be differentiated once more. The four resulting functions $(f_x)_x$, $(f_x)_y$, $(f_y)_x$, $(f_y)_y$ are the second partial derivatives of $f$. For $z=f(x,y)$ the standard notations are

$$(f_x)_x=f_{xx}=f_{11}=\frac{\partial}{\partial x}\left(\frac{\partial f}{\partial x}\right)=\frac{\partial^2 f}{\partial x^2}=\frac{\partial^2 z}{\partial x^2},$$
$$(f_x)_y=f_{xy}=f_{12}=\frac{\partial}{\partial y}\left(\frac{\partial f}{\partial x}\right)=\frac{\partial^2 f}{\partial y\,\partial x}=\frac{\partial^2 z}{\partial y\,\partial x},$$
$$(f_y)_x=f_{yx}=f_{21}=\frac{\partial}{\partial x}\left(\frac{\partial f}{\partial y}\right)=\frac{\partial^2 f}{\partial x\,\partial y}=\frac{\partial^2 z}{\partial x\,\partial y},$$
$$(f_y)_y=f_{yy}=f_{22}=\frac{\partial}{\partial y}\left(\frac{\partial f}{\partial y}\right)=\frac{\partial^2 f}{\partial y^2}=\frac{\partial^2 z}{\partial y^2}.$$

Order of the subscripts. In $f_{xy}$ the differentiation is with respect to $x$ first and $y$ second; in $f_{yx}$ the order is reversed. The notation $\partial^2 f/\partial y\,\partial x$ records the same order: read its denominator from right to left, so the rightmost symbol $\partial x$ is applied first and $\partial y$ second.

THM 10 클레로 정리 Clairaut's TheoremClairaut's Theorem · p.1005
한국어

예제 7에서 확인한 $f_{xy}=f_{yx}$는 우연이 아니다. 실제로 다루는 대부분의 함수에서 두 혼합 편도함수는 일치하며, 다음 정리가 그 충분조건을 제시한다. 알렉시 클레로(Alexis Clairaut, 1713–1765)의 이름을 딴 정리이다.

클레로 정리. $f$가 점 $(a,b)$를 포함하는 원판 $D$에서 정의되어 있다고 하자. $f_{xy}$와 $f_{yx}$가 모두 $D$에서 연속이면

$$f_{xy}(a,b)=f_{yx}(a,b).$$

증명은 교재 Appendix F 참조.

English

The equality $f_{xy}=f_{yx}$ observed in Example 7 is no accident: for the functions met in practice the two mixed partial derivatives agree. The following theorem, due to Alexis Clairaut (1713–1765), states a sufficient condition.

Clairaut's Theorem. Suppose $f$ is defined on a disk $D$ containing the point $(a,b)$. If the functions $f_{xy}$ and $f_{yx}$ are both continuous on $D$, then

$$f_{xy}(a,b)=f_{yx}(a,b).$$

Proof: see Appendix F.

REM 11 3계 이상 편도함수 Higher-Order Mixed Partialsp.1005
한국어

한 번 더 미분하면 3계 이상의 편도함수를 얻는다. 예를 들어

$$f_{xyy}=(f_{xy})_y=\frac{\partial}{\partial y}\left(\frac{\partial^2 f}{\partial y\,\partial x}\right)=\frac{\partial^3 f}{\partial y^2\,\partial x}.$$

인접한 두 미분 순서마다 클레로 정리를 되풀이해서 적용하면, 관련된 함수들이 모두 연속인 한 각 변수로 미분한 횟수만 같으면 되고 그 순서는 중요하지 않음을 알 수 있다.

$$f_{xyy}=f_{yxy}=f_{yyx}.$$
English

Differentiating again produces partial derivatives of order three and higher. For instance

$$f_{xyy}=(f_{xy})_y=\frac{\partial}{\partial y}\left(\frac{\partial^2 f}{\partial y\,\partial x}\right)=\frac{\partial^3 f}{\partial y^2\,\partial x}.$$

Applying Clairaut's Theorem repeatedly to the pairs of adjacent differentiations shows that, when the functions involved are continuous, only the number of differentiations in each variable matters and not their order:

$$f_{xyy}=f_{yxy}=f_{yyx}.$$
DEF 12 라플라스 방정식과 조화함수 Laplace's Equation and Harmonic FunctionsPartial Differential Equations · p.1006
한국어

여러 물리 법칙은 미지함수의 편도함수들 사이의 관계식인 편미분방정식으로 표현된다. 그중 가장 유명한 것이

$$\frac{\partial^2 u}{\partial x^2}+\frac{\partial^2 u}{\partial y^2}=0,$$

피에르 라플라스(Pierre Laplace, 1749–1827)의 이름을 딴 라플라스 방정식이다. 이 방정식의 해를 조화함수라 하며, 열전도·유체 흐름·전위 문제를 다스린다.

English

Many physical laws are stated as partial differential equations, that is, equations relating the partial derivatives of an unknown function. The most famous of them,

$$\frac{\partial^2 u}{\partial x^2}+\frac{\partial^2 u}{\partial y^2}=0,$$

is Laplace's equation, named after Pierre Laplace (1749–1827). Its solutions are called harmonic functions, and they govern problems of heat conduction, fluid flow and electric potential.

DEF 13 파동방정식 The Wave EquationFigure 5 · p.1006
한국어

두 번째 기본적인 편미분방정식은 파동방정식

$$\frac{\partial^2 u}{\partial t^2}=a^2\,\frac{\partial^2 u}{\partial x^2},$$

으로, 바다의 파도·음파·광파, 또는 진동하는 줄을 따라 전달되는 파형의 운동을 기술한다. $u(x,t)$가 시각 $t$에 한쪽 끝으로부터 거리 $x$만큼 떨어진 지점에서 진동하는 줄의 변위를 나타내면 $u$는 이 방정식을 만족한다. 상수 $a$는 줄의 밀도와 장력에 의해 정해진다.

x x u(x, t)
English

A second basic partial differential equation is the wave equation

$$\frac{\partial^2 u}{\partial t^2}=a^2\,\frac{\partial^2 u}{\partial x^2},$$

which governs how a waveform travels — a swell on the ocean, a pulse of sound or of light, or a disturbance running along a vibrating string. If $u(x,t)$ denotes the displacement of a vibrating string at time $t$ at the distance $x$ from one end, then $u$ satisfies this equation. The constant $a$ is determined by the density of the string and by the tension in it.

x x u(x, t)
REM 14 3차원 라플라스 방정식 Three-Dimensional Laplace EquationEq. (5) · p.1007
한국어

세 변수의 편미분방정식도 과학과 공학에서 그 못지않게 중요하다. 3차원 라플라스 방정식은 다음과 같다.

$$\frac{\partial^2 u}{\partial x^2}+\frac{\partial^2 u}{\partial y^2}+\frac{\partial^2 u}{\partial z^2}=0.\tag{5}$$

지구물리학적 응용의 한 예로, $u(x,y,z)$가 위치 $(x,y,z)$에서의 자기장 세기라면 $u$는 (5)를 만족한다. 이 세기의 분포는 철 성분 광물의 분포를 알려 주며, 이로부터 암석의 종류와 단층의 위치를 짐작할 수 있다.

English

Partial differential equations in three variables matter just as much in science and engineering. The three-dimensional Laplace equation is

$$\frac{\partial^2 u}{\partial x^2}+\frac{\partial^2 u}{\partial y^2}+\frac{\partial^2 u}{\partial z^2}=0.\tag{5}$$

One application is geophysical: if $u(x,y,z)$ is the strength of the magnetic field at the position $(x,y,z)$, then $u$ satisfies (5). How that strength is distributed indicates the distribution of iron-rich minerals, and so reflects the rock types present and the location of faults.

예제 답안 · Worked Examples

EXAM 1 다항식의 편도함수 값 Partial Derivatives of a Polynomial at a PointExample 1 · p.1002
한국어

다항함수 $f(x,y)=x^3+x^2y^3-2y^2$ 에 대하여, 점 $(2,1)$ 에서의 두 일계 편도함수 값 $f_x(2,1)$ 과 $f_y(2,1)$ 을 구하라.

풀이. $y$를 상수로 고정한 채 $x$에 대해 통상적으로 미분하고, $x$를 상수로 고정한 채 $y$에 대해 미분하면

$$f_x(x,y)=3x^2+2xy^3,$$
$$f_y(x,y)=3x^2y^2-4y$$

를 얻는다. 점 $(2,1)$을 대입하면 $f_x(2,1)=3\cdot2^2+2\cdot2\cdot1^3=12+4=16$이고 $f_y(2,1)=3\cdot2^2\cdot1^2-4\cdot1=12-4=8$이다. 따라서 $f_x(2,1)=16$, $f_y(2,1)=8$이다.

English

For the polynomial $f(x,y)=x^3+x^2y^3-2y^2$, find the values of both first-order partial derivatives at the point $(2,1)$, that is, find $f_x(2,1)$ and $f_y(2,1)$.

Solution. Holding $y$ fixed and differentiating in the usual one-variable sense with respect to $x$, and holding $x$ fixed while differentiating with respect to $y$, gives

$$f_x(x,y)=3x^2+2xy^3,$$
$$f_y(x,y)=3x^2y^2-4y.$$

Substituting $(2,1)$ yields $f_x(2,1)=3\cdot2^2+2\cdot2\cdot1^3=12+4=16$ and $f_y(2,1)=3\cdot2^2\cdot1^2-4\cdot1=12-4=8$. Hence $f_x(2,1)=16$ and $f_y(2,1)=8$.

EXAM 2 연쇄법칙을 쓰는 편도함수 Partial Derivatives via the Chain RuleExample 2 · p.1002
한국어

합성함수 $f(x,y)=\sin\!\left(\dfrac{x}{1+y}\right)$ 를 각 변수에 대하여 편미분하여 $\dfrac{\partial f}{\partial x}$ 와 $\dfrac{\partial f}{\partial y}$ 를 구하라. (1변수 연쇄법칙을 이용하라.)

풀이. $x$에 대한 편미분에서는 $y$가 상수이므로 $u=x/(1+y)$로 두면 $1+y$가 상수 계수가 되어 $\partial u/\partial x=1/(1+y)$이다. 연쇄법칙 $\partial f/\partial x=\cos u\cdot\partial u/\partial x$에 대입하면

$$\frac{\partial f}{\partial x}=\cos\!\left(\frac{x}{1+y}\right)\cdot\frac{1}{1+y}$$

를 얻는다. $y$에 대한 편미분에서는 $x$가 상수이므로 $u=x(1+y)^{-1}$의 $y$-미분은 $\partial u/\partial y=-x(1+y)^{-2}$이고, 같은 연쇄법칙으로

$$\frac{\partial f}{\partial y}=-\cos\!\left(\frac{x}{1+y}\right)\cdot\frac{x}{(1+y)^2}$$

를 얻는다.

English

Differentiate the composite $f(x,y)=\sin\!\left(\dfrac{x}{1+y}\right)$ with respect to each variable, using the Chain Rule for functions of one variable, and so calculate $\dfrac{\partial f}{\partial x}$ and $\dfrac{\partial f}{\partial y}$.

Solution. Treating $y$ as constant and writing $u=x/(1+y)$, the $x$-derivative of $u$ is the constant factor $1/(1+y)$, so the one-variable Chain Rule $\partial f/\partial x=\cos u\cdot\partial u/\partial x$ gives

$$\frac{\partial f}{\partial x}=\cos\!\left(\frac{x}{1+y}\right)\cdot\frac{1}{1+y}.$$

Treating $x$ as constant and writing $u=x(1+y)^{-1}$, the $y$-derivative of $u$ is $-x(1+y)^{-2}$, so the same Chain Rule gives

$$\frac{\partial f}{\partial y}=-\cos\!\left(\frac{x}{1+y}\right)\cdot\frac{x}{(1+y)^2}.$$
EXAM 3 접선 기울기로 읽는 편도함수 Partial Derivatives as Slopes of TangentsExample 3 · p.1002
한국어

포물면 $f(x,y)=4-x^2-2y^2$ 에 대하여 $f_x(1,1)$ 과 $f_y(1,1)$ 을 구하라. 또 두 값을 점 $(1,1,1)$ 을 지나는 자취 곡선들의 접선 기울기로 해석하라.

풀이. $y$를 고정하고 $x$로 미분하면 $f_x(x,y)=-2x$, $x$를 고정하고 $y$로 미분하면 $f_y(x,y)=-4y$이므로 $f_x(1,1)=-2$, $f_y(1,1)=-4$이다.

이 값을 자취 곡선의 접선 기울기로 읽으면 다음과 같다. 평면 $y=1$이 곡면 $z=4-x^2-2y^2$을 자르는 자취는 포물선 $z=2-x^2$이고, 점 $(1,1,1)$에서 이 곡선의 $x$-기울기는 $\dfrac{d}{dx}(2-x^2)\Big|_{x=1}=-2x\Big|_{x=1}=-2$로 $f_x(1,1)$과 일치한다. 마찬가지로 평면 $x=1$의 자취는 포물선 $z=3-2y^2$이고, 점 $(1,1,1)$에서 그 $y$-기울기는 $\dfrac{d}{dy}(3-2y^2)\Big|_{y=1}=-4y\Big|_{y=1}=-4$로 $f_y(1,1)$과 일치한다. 두 기울기가 모두 음수이므로, 이 점에서 곡면은 양의 $x$ 방향으로도 양의 $y$ 방향으로도 내려간다.

English

For the paraboloid $f(x,y)=4-x^2-2y^2$, compute $f_x(1,1)$ and $f_y(1,1)$, and interpret the two numbers as the slopes of the tangent lines at $(1,1,1)$ to the traces of the graph in the planes $y=1$ and $x=1$.

Solution. Holding $y$ fixed and differentiating in $x$ gives $f_x(x,y)=-2x$, and holding $x$ fixed and differentiating in $y$ gives $f_y(x,y)=-4y$; thus $f_x(1,1)=-2$ and $f_y(1,1)=-4$.

To read these as slopes: the plane $y=1$ cuts the surface $z=4-x^2-2y^2$ in the parabola $z=2-x^2$, whose slope at $x=1$ is $\dfrac{d}{dx}(2-x^2)\big|_{x=1}=-2x\big|_{x=1}=-2$, matching $f_x(1,1)$. Likewise the plane $x=1$ cuts the surface in the parabola $z=3-2y^2$, whose slope at $y=1$ is $\dfrac{d}{dy}(3-2y^2)\big|_{y=1}=-4y\big|_{y=1}=-4$, matching $f_y(1,1)$. Both slopes are negative, so at that point the surface descends in the positive $x$-direction and in the positive $y$-direction alike.

EXAM 4 체질량지수의 편도함수 Partial Derivatives of the Body Mass IndexExample 4 · p.1003
한국어

체질량지수(BMI)를 $B(m,h)=\dfrac{m}{h^2}$ 로 모형화하자. 여기서 $m$ 은 kg 단위의 몸무게, $h$ 는 m 단위의 키이다. 두 편도함수를 구하고, 몸무게 $m=64$, 키 $h=1.68$ 인 사람에 대한 값을 계산한 뒤 그 값이 뜻하는 바를 해석하라.

풀이. $h$를 상수로 보고 $m$으로 미분하면 $\dfrac{\partial B}{\partial m}=\dfrac1{h^2}$이고, $m$을 상수로 보고 $h$로 미분하면 $\dfrac{\partial B}{\partial h}=-\dfrac{2m}{h^3}$이다.

$m=64$, $h=1.68$을 대입하면 $\dfrac{\partial B}{\partial m}(64,1.68)=\dfrac1{1.68^2}\approx0.35$, $\dfrac{\partial B}{\partial h}(64,1.68)=-\dfrac{2\cdot64}{1.68^3}\approx-27$이며, 이때 $B(64,1.68)=\dfrac{64}{1.68^2}\approx22.68$이다.

해석하면, 키를 $1.68\,\mathrm m$으로 고정한 채 몸무게가 $64\,\mathrm{kg}$에서 $1\,\mathrm{kg}$ 늘면 BMI는 약 $0.35$ 증가하고, 몸무게를 $64\,\mathrm{kg}$으로 고정한 채 키가 $1.68\,\mathrm m$에서 $1\,\mathrm{cm}(=0.01\,\mathrm m)$ 늘면 BMI는 약 $27\times0.01=0.27$ 감소한다.

English

Model the body mass index (BMI) by $B(m,h)=\dfrac{m}{h^{2}}$, where the mass $m$ is measured in kilograms and the height $h$ in meters. Find both partial derivatives of $B$, evaluate them for a young man with $m=64$ and $h=1.68$, and interpret the two numbers.

Solution. Treating $h$ as constant, $\dfrac{\partial B}{\partial m}=\dfrac1{h^2}$; treating $m$ as constant, $\dfrac{\partial B}{\partial h}=-\dfrac{2m}{h^3}$.

At $m=64$, $h=1.68$: $\dfrac{\partial B}{\partial m}(64,1.68)=\dfrac1{1.68^2}\approx0.35$ and $\dfrac{\partial B}{\partial h}(64,1.68)=-\dfrac{2\cdot64}{1.68^3}\approx-27$, while $B(64,1.68)=\dfrac{64}{1.68^2}\approx22.68$.

Interpretation: with height fixed at $1.68\,\mathrm m$, each extra kilogram of mass near $64\,\mathrm{kg}$ raises the BMI by about $0.35$; with mass fixed at $64\,\mathrm{kg}$, each extra centimeter of height near $1.68\,\mathrm m$ (i.e. $0.01\,\mathrm m$) lowers the BMI by about $27\times0.01=0.27$.

EXAM 5 음함수로 정의된 곡면의 편도함수 Partial Derivatives of an Implicitly Defined FunctionExample 5 · p.1004
한국어

방정식 $x^3+y^3+z^3+6xyz+4=0$ 이 $z$ 를 $x,y$ 의 함수로 음함수적으로 정의한다고 하자. $\dfrac{\partial z}{\partial x}$ 와 $\dfrac{\partial z}{\partial y}$ 를 구하고, 점 $(-1,1,2)$ 에서의 값을 계산하라.

풀이. $z$를 $x,y$의 함수로 보고 등식의 양변을 $x$에 대해 편미분한다($y$는 상수, $z$는 $x$에 종속). $3x^2+3z^2\dfrac{\partial z}{\partial x}+6yz+6xy\dfrac{\partial z}{\partial x}=0$에서 $\partial z/\partial x$ 항을 모으면 $(3z^2+6xy)\dfrac{\partial z}{\partial x}=-3x^2-6yz$이므로

$$\frac{\partial z}{\partial x}=-\frac{x^2+2yz}{z^2+2xy}$$

이다. 같은 등식을 $y$에 대해 편미분하면($x$가 상수) $3y^2+3z^2\dfrac{\partial z}{\partial y}+6xz+6xy\dfrac{\partial z}{\partial y}=0$이 되고, 정리하면

$$\frac{\partial z}{\partial y}=-\frac{y^2+2xz}{z^2+2xy}$$

이다. (같은 결과를 $F(x,y,z)=x^3+y^3+z^3+6xyz+4$로 두고 $\partial z/\partial x=-F_x/F_z$, $\partial z/\partial y=-F_y/F_z$ 공식으로 바로 얻을 수도 있다: $F_x=3x^2+6yz$, $F_y=3y^2+6xz$, $F_z=3z^2+6xy$이므로 분자·분모를 $3$으로 약분하면 위와 같다.) 점 $(-1,1,2)$에서 $x^2+2yz=1+4=5$, $y^2+2xz=1-4=-3$, $z^2+2xy=4-2=2$이므로 $\dfrac{\partial z}{\partial x}(-1,1,2)=-\dfrac52$, $\dfrac{\partial z}{\partial y}(-1,1,2)=\dfrac32$이다.

English

Suppose the equation $x^3+y^3+z^3+6xyz+4=0$ defines $z$ implicitly as a function of $x$ and $y$. Find $\dfrac{\partial z}{\partial x}$ and $\dfrac{\partial z}{\partial y}$, and then evaluate both at the point $(-1,1,2)$.

Solution. Regard $z$ as a function of $x$ and $y$ and differentiate the equation with respect to $x$ (with $y$ constant and $z$ depending on $x$): $3x^2+3z^2\dfrac{\partial z}{\partial x}+6yz+6xy\dfrac{\partial z}{\partial x}=0$. Collecting the $\partial z/\partial x$ terms, $(3z^2+6xy)\dfrac{\partial z}{\partial x}=-3x^2-6yz$, so

$$\frac{\partial z}{\partial x}=-\frac{x^2+2yz}{z^2+2xy}.$$

Differentiating the same equation with respect to $y$ (now $x$ constant) gives $3y^2+3z^2\dfrac{\partial z}{\partial y}+6xz+6xy\dfrac{\partial z}{\partial y}=0$, which rearranges to

$$\frac{\partial z}{\partial y}=-\frac{y^2+2xz}{z^2+2xy}.$$

(Equivalently, with $F(x,y,z)=x^3+y^3+z^3+6xyz+4$, the formulas $\partial z/\partial x=-F_x/F_z$ and $\partial z/\partial y=-F_y/F_z$ give the same result directly, since $F_x=3x^2+6yz$, $F_y=3y^2+6xz$, $F_z=3z^2+6xy$.) At $(-1,1,2)$: $x^2+2yz=1+4=5$, $y^2+2xz=1-4=-3$, $z^2+2xy=4-2=2$, so $\dfrac{\partial z}{\partial x}(-1,1,2)=-\dfrac52$ and $\dfrac{\partial z}{\partial y}(-1,1,2)=\dfrac32$.

EXAM 6 3변수 함수의 편도함수 Partial Derivatives of a Function of Three VariablesExample 6 · p.1004
한국어

$z>0$ 인 영역에서 정의된 함수 $f(x,y,z)=e^{xy}\ln z$ 의 세 일계 편도함수 $f_x$, $f_y$, $f_z$ 를 모두 구하라.

풀이. $y,z$를 고정하고 $x$로 미분한다. $e^{xy}$의 $x$-미분이 $ye^{xy}$이므로 $f_x=ye^{xy}\ln z$이다. $x,z$를 고정하고 $y$로 미분하면 같은 방식으로 $f_y=xe^{xy}\ln z$이다. $x,y$를 고정하고 $z$로 미분하면 $e^{xy}$는 상수이고 $\ln z$의 미분이 $1/z$이므로 $f_z=\dfrac{e^{xy}}{z}$이다.

English

Compute all three first-order partial derivatives $f_x$, $f_y$ and $f_z$ of the function $f(x,y,z)=e^{xy}\ln z$, defined for $z>0$.

Solution. Fix $y,z$ and differentiate in $x$: since the $x$-derivative of $e^{xy}$ is $ye^{xy}$, we get $f_x=ye^{xy}\ln z$. Fixing $x,z$ and differentiating in $y$ gives, by the same reasoning, $f_y=xe^{xy}\ln z$. Fixing $x,y$ and differentiating in $z$: $e^{xy}$ is constant and the derivative of $\ln z$ is $1/z$, so $f_z=\dfrac{e^{xy}}{z}$.

EXAM 7 이계 편도함수 네 개 The Four Second Partial DerivativesExample 7 · p.1005
한국어

예제 1의 함수 $f(x,y)=x^3+x^2y^3-2y^2$ 에 대하여 이계 편도함수 $f_{xx}$, $f_{xy}$, $f_{yx}$, $f_{yy}$ 를 모두 구하라.

풀이. 예제 1에서 $f_x(x,y)=3x^2+2xy^3$, $f_y(x,y)=3x^2y^2-4y$를 얻었다. $f_x$를 다시 $x$로 미분하면($y$ 고정) $f_{xx}=6x+2y^3$이고, $f_x$를 $y$로 미분하면($x$ 고정) $f_{xy}=6xy^2$이다. $f_y$를 $x$로 미분하면 $f_{yx}=6xy^2$이고, $f_y$를 $y$로 미분하면 $f_{yy}=6x^2y-4$이다. $f_{xy}=f_{yx}=6xy^2$로 두 혼합 편도함수가 일치하는데, 이는 앞서 살펴본 클레로 정리가 성립하는 한 예이다.

English

Find all four second-order partial derivatives $f_{xx}$, $f_{xy}$, $f_{yx}$ and $f_{yy}$ of the function $f(x,y)=x^3+x^2y^3-2y^2$ of Example 1.

Solution. From Example 1, $f_x(x,y)=3x^2+2xy^3$ and $f_y(x,y)=3x^2y^2-4y$. Differentiating $f_x$ again in $x$ (with $y$ fixed) gives $f_{xx}=6x+2y^3$, and differentiating $f_x$ in $y$ (with $x$ fixed) gives $f_{xy}=6xy^2$. Differentiating $f_y$ in $x$ gives $f_{yx}=6xy^2$, and differentiating $f_y$ in $y$ gives $f_{yy}=6x^2y-4$. Since $f_{xy}=f_{yx}=6xy^2$, the two mixed partials agree, illustrating Clairaut's Theorem stated above.

EXAM 8 4계 혼합 편도함수 A Fourth-Order Mixed Partial DerivativeExample 8 · p.1006
한국어

$f(x,y,z)=\sin(3x+yz)$ 의 4계 혼합 편도함수 $f_{xxyz}$ 를 계산하라.

풀이. 미분 순서를 $x,x,y,z$로 두고 차례로 계산한다. 먼저 $x$로 미분하면 $f_x=3\cos(3x+yz)$이고, 다시 $x$로 미분하면 $f_{xx}=-9\sin(3x+yz)$이다. 이어서 $y$로 미분하면($3x+yz$의 $y$-미분이 $z$이므로) $f_{xxy}=-9z\cos(3x+yz)$이다. 마지막으로 $z$로 미분할 때는 $-9z\cos(3x+yz)$에서 $z$와 $\cos(3x+yz)$가 모두 $z$에 의존하므로 곱미분법을 쓴다:

$$\begin{aligned}\frac{\partial}{\partial z}\big[-9z\cos(3x+yz)\big]&=-9\cos(3x+yz)-9z\cdot\big(-\sin(3x+yz)\big)\cdot y\\&=9yz\sin(3x+yz)-9\cos(3x+yz).\end{aligned}$$

따라서 $f_{xxyz}=9yz\sin(3x+yz)-9\cos(3x+yz)$이다.

English

Calculate the fourth-order mixed partial derivative $f_{xxyz}$ of the function $f(x,y,z)=\sin(3x+yz)$.

Solution. Differentiate in the order $x,x,y,z$. First, $f_x=3\cos(3x+yz)$; differentiating again in $x$ gives $f_{xx}=-9\sin(3x+yz)$. Differentiating in $y$ next (the $y$-derivative of $3x+yz$ is $z$) gives $f_{xxy}=-9z\cos(3x+yz)$. Finally, differentiating in $z$ requires the product rule, since both the factor $z$ and $\cos(3x+yz)$ depend on $z$:

$$\begin{aligned}\frac{\partial}{\partial z}\big[-9z\cos(3x+yz)\big]&=-9\cos(3x+yz)-9z\cdot\big(-\sin(3x+yz)\big)\cdot y\\&=9yz\sin(3x+yz)-9\cos(3x+yz).\end{aligned}$$

Hence $f_{xxyz}=9yz\sin(3x+yz)-9\cos(3x+yz)$.

EXAM 9 라플라스 방정식의 해 확인 Verifying a Solution of Laplace's EquationExample 9 · p.1006
한국어

함수 $u(x,y)=e^{x}\sin y$ 가 라플라스 방정식 $u_{xx}+u_{yy}=0$ 의 해임을, 즉 조화함수임을 보여라.

증명. $u_x=e^x\sin y$이므로 다시 $x$로 미분하면 $u_{xx}=e^x\sin y$이다. 한편 $u_y=e^x\cos y$이므로 다시 $y$로 미분하면 $u_{yy}=-e^x\sin y$이다. 두 값을 더하면 $u_{xx}+u_{yy}=e^x\sin y-e^x\sin y=0$이므로 $u$는 라플라스 방정식을 만족하는 조화함수이다.

$\square$
English

Show that the function $u(x,y)=e^{x}\sin y$ is a solution of Laplace's equation $u_{xx}+u_{yy}=0$; that is, show that $u$ is harmonic.

Proof. Since $u_x=e^x\sin y$, differentiating again in $x$ gives $u_{xx}=e^x\sin y$. Since $u_y=e^x\cos y$, differentiating again in $y$ gives $u_{yy}=-e^x\sin y$. Adding, $u_{xx}+u_{yy}=e^x\sin y-e^x\sin y=0$, so $u$ satisfies Laplace's equation and is harmonic.

$\square$
EXAM 10 파동방정식의 해 확인 Verifying a Solution of the Wave EquationExample 10 · p.1006
한국어

진행파 $u(x,t)=\sin(x-at)$ 가 파동방정식 $u_{tt}=a^2u_{xx}$ 를 만족함을 확인하라.

증명. $u_x=\cos(x-at)$이므로 다시 $x$로 미분하면 $u_{xx}=-\sin(x-at)$이다. 한편 $u_t=-a\cos(x-at)$이므로($x-at$의 $t$-미분이 $-a$) 다시 $t$로 미분하면 $u_{tt}=-a\cdot\big(a\sin(x-at)\big)=-a^2\sin(x-at)$이다. 따라서 $u_{tt}=-a^2\sin(x-at)=a^2\big(-\sin(x-at)\big)=a^2u_{xx}$이므로 $u$는 파동방정식 $u_{tt}=a^2u_{xx}$를 만족한다.

$\square$
English

Verify that the traveling wave $u(x,t)=\sin(x-at)$ satisfies the wave equation $u_{tt}=a^{2}u_{xx}$.

Proof. Since $u_x=\cos(x-at)$, differentiating again in $x$ gives $u_{xx}=-\sin(x-at)$. Since $u_t=-a\cos(x-at)$ (the $t$-derivative of $x-at$ is $-a$), differentiating again in $t$ gives $u_{tt}=-a\cdot\big(a\sin(x-at)\big)=-a^2\sin(x-at)$. Thus $u_{tt}=-a^2\sin(x-at)=a^2\big(-\sin(x-at)\big)=a^2u_{xx}$, so $u$ satisfies the wave equation $u_{tt}=a^2u_{xx}$.

$\square$

연결 · Links