21
$\begingroup$

When teaching the integration method of u-substitution, I like to emphasize its connection with the chain rule of integration.
Likewise, the intimate connection between the product rule of derivatives and the method of integration by parts comes up in discussion.

Is there an analogous rule of integration for the quotient rule?

Of course, if you spot an integral of the form $\int \left (\frac{f(x)}{g(x)} \right )' = \int \frac{g(x) \cdot f(x)' - f(x) \cdot g(x)'}{\left [ g(x)\right ]^2 }$,
then the antiderivative is obvious. But is there another form/manipulation/"trick"?

$\endgroup$
9
  • 17
    $\begingroup$ Note that $\frac fg = f\cdot {\frac1g}$, so there is in fact no new rule for the quotient. The only rule is for product. $\endgroup$
    – SBF
    Sep 29, 2011 at 12:28
  • 2
    $\begingroup$ Yeah, nothing really that new. FWIW, I always had derived the quotient rule from both the product rule and chain rule, so I'd say those two are more fundamental, and you've covered those already. In any event... "product rule of integration"? $\endgroup$ Sep 29, 2011 at 12:37
  • 1
    $\begingroup$ @J. M. - it was a test to see who had their coffee this morning. That reminds me... :) $\endgroup$ Sep 29, 2011 at 12:39
  • 3
    $\begingroup$ @TheChaz: I believe our location is different. On the North Sea beach it's now afternoon and very hot, so coffee doesn't help much (( $\endgroup$
    – SBF
    Sep 29, 2011 at 12:46
  • 1
    $\begingroup$ @Gortaur: Next time I'm up there, I'll buy a round of iced coffees! But seriously, would you care to (slightly) elaborate on your comment and put it as an answer? $\endgroup$ Sep 29, 2011 at 12:48

5 Answers 5

21
$\begingroup$

I guess you could arrange an analog to integration by parts, but making students learn it would be superfluous.

$$ \int \frac{du}{v} = \frac{u}{v} + \int \frac{u}{v^2} dv.$$

$\endgroup$
3
  • 13
    $\begingroup$ I kind of like it, no minus signs. Also has nice natural form when one uses integration by parts (sorry, the Ault Rule) for estimation. $\endgroup$ Sep 29, 2011 at 13:54
  • $\begingroup$ Cool, I like it as well for the same reason of the absence of minuses. $\endgroup$
    – SBF
    Sep 29, 2011 at 13:55
  • $\begingroup$ @AndréNicolas a minus sign though strikes out the duality between differentiation and integration $\endgroup$
    – SBF
    Jun 2 at 17:40
19
$\begingroup$

As for me, I cannot see an advantage in introduction of such a rule since for any two functions $f,g$ it clearly holds that $$ \frac fg = f\cdot\frac1g $$ so the 'quotient rule' for derivatives is a product rule in disguise, and the same will also hold for the integration by parts. Indeed, when you are looking for the proper function to put under the differential sign integrating by parts, in case you have a bit of experince with such a procedure, you also will think about the 'quotients'.

As an example: $$ \int\frac{\sin\frac1x}{x^2}\,dx $$ Of course you can present it as $\frac{f(x)}{x^2}$ and apply the new integration by parts based on the quotient rule, but I almost sure that a lot of the readers will rather think of the fact that $\frac1{x^2}\,dx = -d\frac1x$, by this seeing a product in the integrand rather than a quotient.

$\endgroup$
16
$\begingroup$

It deserves to be better known that a "quotient rule" does play a role in Hermite's algorithm for integrating rational functions. It works as follows. By squarefree decomposing the denominator and partial fraction expanding, we reduce to integrating $\rm\,A/D^k\in \mathbb Q(x),\,$ with $\rm\,\deg\,A < \deg\,D^k,\,$ and where $\rm\,D\,$ is squarefree, so $\rm\,\gcd(D,D') = 1.\,$ Thus by Bezout (extended Euclidean algorithm) there are $\rm\,B,C\in \mathbb Q[x]\,$ such that $\rm\ B\,D' + C\, D\ =\ A/(1-k).\,$ Then a little algebra shows that

$$\rm\int \frac{A}{D^k}\ =\ \frac{B}{D^{k-1}}\ +\ \int \frac{(1-k)\ C - B'}{D^{k-1}} $$

Iterating the above rule we eventually reduce to the case $\rm\,k=1,\,$ i.e. squarefree denominator $\rm\,D.\,$ Thus using the above "quotient rule" and nothing deeper than Euclid's algorithm for polynomials (without requiring any factorization) we can mechanically compute the "rational part" of the integral of a rational function, i.e. the part of the integral not involving logarithms. This Hermite reduction rule is the basis of an algorithm due to Hermite (1872). It plays a fundamental role in the transcendental case of some (modern) integration algorithms.

$\endgroup$
3
  • $\begingroup$ In standard calculus courses,special cases of the Hermite procedure is usually called "the rule for integration by rational functions". I don't think it's really a "rule" in the usual sense we get in these courses,in that it's not an algorithm derived from using the Fundamental Theorum of Calculus to "invert" a derivative rule. It's just a handy application of Galois theory. $\endgroup$ Sep 29, 2011 at 19:24
  • 1
    $\begingroup$ @Mat There is no Galois theory involved. It's not clear to me what restricted denotation of "rule" that you intend, nor why you think it plays a role here. I dont recall ever hearing of Hermite reduction being applied in any "standard calculus course". That would be quite some course. Do you have a link? $\endgroup$ Sep 29, 2011 at 19:44
  • $\begingroup$ I'll look,Bill. $\endgroup$ Sep 29, 2011 at 22:22
8
$\begingroup$

Is there an analogous rule of integration for the quotient rule?

For a rational function $P(x)/Q(x)$, the answer is affirmative, and it's called the Ostrogradsky-Hermite method. I adapt this answer of mine.

To integrate a rational function $P(x)/Q(x)$ without decomposing it into partial fractions and without finding the roots of the denominator, we can use the Ostrogradski-Hermite method. You can find a description of this method in section 2.1 of Gradshteyn and Ryzhik's Table of Integrals, Series, and Products, where the identity $(2)$ bellow is given. The formula $(1)$ appears also on the Ostrogradsky's Wikipedia page.

Assume that $\deg P(x)<\deg $ $Q(x)$. There exist polynomials $P_{1}(x)$, $P_{2}(x)$, $Q_{1}(x)$ and $Q_{2}(x)$, with $Q_{1}(x)=\gcd \left\{ Q(x), Q^{\prime }(x)\right\}$ and $Q_{2}(x)=Q(x)/Q_{1}(x)$, $\deg P_{1}(x)<\deg Q_{1}(x)$, $\deg P_{2}(x)<\deg Q_{2}(x)$, such that

\begin{equation} \int \frac{P(x)}{Q(x)}dx=\frac{P_{1}(x)}{Q_{1}(x)}+\int \frac{P_{2}(x)}{ Q_{2}(x)}dx.\tag{1} \end{equation}

Indeed, by differentiation and multiplication by $P(x)$, we have

\begin{eqnarray*} P(x) &=&\frac{P_{1}^{\prime }(x)Q_{1}(x)-P_{1}(x)Q_{1}^{\prime }(x)}{\left\{ Q_{1}(x)\right\} ^{2}}Q(x)+\frac{P_{2}(x)}{Q_{2}(x)}Q(x) \\ &=&P_{1}^{\prime }(x)\frac{Q(x)}{Q_{1}(x)}-P_{1}(x)\frac{Q_{1}^{\prime }(x)}{Q_{1}(x)}\frac{Q(x)}{Q_{1}(x)}+P_{2}(x)\frac{Q(x)}{Q_{2}(x)} \\ &=&P_{1}^{\prime }(x)Q_{2}(x)-P_{1}(x)\left\{ \frac{Q_{1}^{\prime }(x)}{Q_{1}(x)}Q_{2}(x)\right\} +P_{2}(x)Q_{1}(x) \end{eqnarray*}

or

\begin{equation} P(x)=P_{1}^{\prime }(x)Q_{2}(x)-P_{1}(x)\left\{ T(x)-Q_{2}^{\prime }(x)\right\}+P_{2}(x)Q_{1}(x),\tag{2} \end{equation}

with $T(x)=Q^{\prime }(x)/Q_{1}(x)$, because from

\begin{equation*} Q^{\prime }(x)=\left\{ Q_{1}(x)Q_{2}(x)\right\} ^{\prime }=Q_{1}^{\prime }(x)Q_{2}(x)+Q_{1}(x)Q_{2}^{\prime }(x)=T(x)Q_{1}(x) \end{equation*}

we obtain

\begin{equation*} \frac{Q_{1}^{\prime }(x)}{Q_{1}(x)}Q_{2}(x)+Q_{2}^{\prime }(x)=T(x). \end{equation*}

To find the coefficients of the polynomials $P_{1}(x)$ and $P_{2}(x)$ we equate the coefficients of like powers of $x$ and/or make $x=x_1,x_2,\dots$ until we obtain a system of linear equations in those coefficients.

$\endgroup$
1
$\begingroup$

For integrating a quotient of two functions, usually the rule for integration by parts is recommended: \begin{equation}\int f(x)g'(x)dx=f(x)g(x)-\int f'(x)g(x)dx,\end{equation} \begin{equation}\int f'(x)g(x)dx=f(x)g(x)-\int f(x)g'(x)dx.\end{equation} You have to choose $f$ and $g$ so that the integrand at the left side of one of the both formulas is the quotient of your given functions.

But that needs some experience, and for the inexperienced or no-more skilled, this is somewhat unwieldy. In "A Quotient Rule Integration by Parts Formula" and in "Quotient-Rule-Integration-by-Parts", the authors integrate the quotient rule of differentiation and get so a quotient rule integration by parts formula, where $F(x)=\int f(x) dx+c_{1}$, $c_{1}$ a constant: \begin{equation}\int\frac{f'(x)}{g(x)}dx=\frac{f(x)}{g(x)}+\int f(x)\frac{g'(x)}{g(x)^{2}}dx,\end{equation} \begin{equation}\int\frac{f(x)}{g'(x)}dx=\frac{F(x)}{g'(x)}+\int F(x)\frac{g''(x)}{g'(x)^{2}}dx.\end{equation} Just as we get the quotient rule for differentiation from the product rule, we get this quotient rule for integration from the rule for integration by parts.

Because these formulas still do not look like true quotient rules, I brought them to the following form, where again $F(x)=\int f(x) dx+c_{1}$, $c_{1}$ a constant:

\begin{equation}\int\frac{f(x)}{g(x)}dx=\frac{F(x)}{g(x)}+\int F(x)\frac{g'(x)}{g(x)^{2}}dx,\end{equation}

\begin{equation}\int_a^b\frac{f(x)}{g(x)}dx=\frac{F(x)}{g(x)}\bigg\vert_{a}^{b}+\int_a^b F(x)\frac{g'(x)}{g(x)^{2}}dx.\end{equation}

This quotient rule can also be deduced from the formula for integration by parts. The new formula is simply the formula for integration by parts in another shape. Therefore it has no new information, but its form allows to see what is needed for calculating the integral of the quotient of two functions.

I derived an anlog formula for the product rule of integration in "Are the real product rule and quotient rule for integration already known?".

Recently, this quotient rule of integration was also published in
Will, J.: Produktregel, Quotientenregel, Reziprokenregel, Kettenregel und Umkehrregel für die Integration. May 2017,
Will, J.: Product rule, quotient rule, reciprocal rule, chain rule and inverse rule for integration. May 2017.

The experienced will use the rule for integration of parts, but the others could find the new formula somewhat easier.

It can be assumed that other quotient rules are possible. That means, the right side of the quotient rule can be written also in different forms.

$\endgroup$
0

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .