Math 555: Differential Equations I |
Consider a first order differential equation of the form
$$
M(x,y)\, dx + N(x,y)\,dy = 0. \tag{1}
$$
Equation $(1)$ is said to be exact if and only if there is a function
$\varphi(x,y)$ for which $\partial_x \varphi = M$ and $\partial_y \varphi = N$.
Such a $\varphi$ will exist if and only if
$$
\frac{\partial M}{\partial y} = \frac{\partial N}{\partial x}. \tag{2}
$$
In this case, assuming $y=y(x)$ is an implicit function of $x$, then equation $(1)$
may be rewritten as
$$
\frac{\partial \varphi}{\partial x} + \frac{\partial \varphi}{\partial y}\frac{dy}{dx} = \frac{d}{dx}\Big[\varphi(x,y(x))\Big] = 0. \tag{*}
$$
Integrating, we obtain the solution
$$
\varphi(x,y) = C, \tag{3}
$$
where
$$
\varphi(x,y) = \int M(x,y)\,dx = \int N(x,y)\, dy. \tag{4}
$$
For readers with a Calculus III background, an exact differential equation is closely
related to a conservative vector field. The solution $\varphi$ of the differential
equation is a potential function of the vector field. Solution curves of the
differential equation are equipotential lines of the vector field.
If the functions $M$ and $N$ of equation $(1)$ do not satisfy the criterion of
equation $(2)$, then we must try to find an integrating factor, $\mu$.
An acceptable integrating factor is a function $\mu = \mu(x,y)$ that is never equal
to $0$ (think: $\mu > 0$) and makes the equation
$$
\mu(x,y)M(x,y)\, dx + \mu(x,y)N(x,y)\,dy = 0 \tag{5}
$$
exact. Equation $(5)$ being exact requires that
$$
\frac{\partial}{\partial y} \Big[ \mu(x,y)M(x,y)\Big] =
\frac{\partial}{\partial x} \Big[ \mu(x,y)N(x,y)\Big]. \tag{6}
$$
If $\mu$ is solely a function of $x$, and not $y$, then $\mu$ will satisfy the
differential equation
$$
\frac{1}{\mu} d\mu = \frac{\partial_y M - \partial_x N}{N} dx \tag{8}
$$
where the quotient on the righthand side must be a function of $x$ alone. Similarly,
if $\mu$ is solely a function of $y$, then it will satisfy the differential equation
$$
\frac{1}{\mu} d\mu = \frac{\partial_y M - \partial_x N}{-M} dy \tag{9}
$$
where the righthand side is a function of only $y$. Equations $(8)$ and $(9)$ give
two special cases for choosing $\mu$, but in general it may be difficult to find
an acceptable integrating factor.
We consider some special cases and look at some examples below.
Example 1.
Suppose $M(x,y) = M(x)$ and $N(x,y) = N(y)$. Then equation $(2)$ is trivially satisfied, so that equation $(1)$ is exact. But in this case equation $(1)$ may be rewritten as
$$
\frac{dy}{dx} = \frac{-M(x)}{N(y)};
$$
a separable equation. So, all separable FODE are exact.
Example 2. Suppose $M(x,y) = p(x)y - q(x)$ and $N(x,y) = 1$. Then equation $(1)$ may be rewritten as
$$
\frac{dy}{dx} + p(x)y = q(x);
$$
a linear equation. In general, linear equations are not exact. As an exercise, work through the procedure described above to determine $\mu$ as a function of $x$ alone. You should get the same equation for $\mu$ that we derived for linear DE,
$$
\mu(x) = e^{\int p(x)\,dx}.
$$
Example 3. Consider the differential equation
$$
2x + y^2 + 2xy\,y' = 0.
$$
This DE is neither separable nor linear. Comparing with equation $(1)$, $M(x,y) = 2x + y^2$ and $N(x,y) = 2xy$. The partial derivatives are
$$
\frac{\partial M}{\partial y} = 2y,\ \ \ \frac{\partial N}{\partial x} = 2y,
$$
so the equation is exact. Integrating we obtain
$$
\varphi = \int 2x + y^2\, dx = x^2 + xy^2 + C_1(y),
$$
where $C_1$ could be an unknown function of $y$. Similarly,
$$
\varphi = \int 2xy\, dy = xy^2 + C_2(x),
$$
where $C_2$ could be an unknown function of $x$. These two integrals must agree--they are both equal to $\varphi$. Therefore $C_2(x)$ must be set equal to $x^2$. The function $C_1(y)$ may be set to $0$. Therefore, we have obtained that
$$
\varphi = x^2 + xy^2
$$
is a function satisfying $\partial_x \varphi = M$ and $\partial_y \varphi = N$. Hence the solution to the DE is
$$
x^2 + xy^2 = C.
$$
We plot the slope field and some solution curves below.
Example 4. Consider the differential equation
$$
(3xy-y^2)\, dx + x(x-y)\,dy= 0.
$$
This equation is not exact since
$$
\frac{\partial M}{\partial y} = 3x - 2y,\ \ \
\frac{\partial N}{\partial x} = 2x -y.
$$
Notice that $\partial_y M - \partial_x N = (x - y)$, so that
$$
\frac{\partial_y M - \partial_x N}{N} = \frac{1}{x}
$$
is a function solely of $x$. Therefore an acceptable integrating factor for this equation is
$$
\mu(x) = e^{\int \frac{1}{x}\, dx} = x.
$$
Using this integrating factor, the solution is
$$
\varphi = x^3y - \tfrac{1}{2}x^2y^2 = C.
$$
The details are left as an exercise.
Again, we plot the slope field together with some solution curves.
Back to main page
Your use of Wichita State University content and this material is subject to our Creative Common License.