Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

My intention is to express the LHS in the RHS -form

$$\int \frac1{\sqrt{x}}\ln(x)\sin(x) dx=x \sqrt{x}\,\left[F(x)\ln(x)+G(x)\right] + C$$

where $x>0$. WA shows that the solution requires something-called gamma-fuctions and generalized hypergeometric functions, here. $F(x)$ and $G(x)$ are undefined. When I look at the WA -solution, I am bit flabbergasted what to believe because WA has more terms than the supposed. Could someone outline this problem? Is the problem just chain-rule or is there some easy-way to get into the right hand side?

I have started from series expansion but I am not sure wheher the right way. I have tried to research why gamma-function appears in this integral but cannot yet understand it, wikipedia here. I am not sure from WA -notation whether the gamma function at hand is in even defined by reals or complexes (I think in complexes now). In reals, it is apparently just a generalization of factorial. And here is the hypergeometric

$$_{2}F_{1}(a, b; c; z)=\sum_{n=0}^{\infty}\frac{(a)_{n} (b)_{n}}{(c)_{n}}\frac{z^{n}}{n!},$$

fully defined here.

share|improve this question
    
What kind of class is this homework for? I suspect there may be a mistake in the question, because I'm pretty sure this integral can't be expressed as an elementary function - you need special functions (Maple uses LommelS1 and its derivatives with respect to the first and second parameters). –  Robert Israel Feb 5 '12 at 8:22
    
@RobertIsrael: It is an engineering course, page 538 here on the M-word-documented course book (sorry not in English). This question is jammed in the part serie-expanion and partial-fraction. It is very open-ended and misleading question with free functions $F(x)$ and $G(x)$ -- or it is possible that the $F(x)$ is defined so that: $\begin{cases}F'(x)=\frac{1}{\sqrt{2}}e^{\frac{-x^{2}}{2}}, x\in\mathbb R \\ F(0)=\frac{1}{2} \end{cases}$. Sorry this question is not one of the clearest and one can understand this in multiple ways. –  hhh Feb 5 '12 at 8:38

2 Answers 2

up vote 2 down vote accepted

The idea of this homework is to identify some functions $F$ and $G$ such that the RHS is a solution. Note that the derivative of the RHS is $$ \tfrac32\sqrt{x}(F(x)\log(x)+G(x))+x\sqrt{x}(x^{-1}F(x)+F'(x)\log(x)+G'(x)), $$ Introduce the function $$ S(x)=\frac{\sin(x)}x=\sum\limits_{n\geqslant0}s_nx^{2n},\qquad s_n=\frac{(-1)^n}{(2n+1)!}. $$ Then the derivative of the RHS minus the LHS is $$ \sqrt{x}\left(\tfrac32G(x)+F(x)+xG'(x)\right)+\sqrt{x}\log(x)\left(\tfrac32F(x)+xF'(x)-S(x)\right),$$ hence the RHS is a solution as soon as $$ \tfrac32G(x)+F(x)+xG'(x)=0,\qquad \tfrac32F(x)+xF'(x)=S(x).$$ For the next step, it is probably best to use series expansions. Assume that $$ F(x)=\sum\limits_{n\geqslant0}a_nx^{2n},\quad G(x)=\sum\limits_{n\geqslant0}b_nx^{2n}. $$ The two first order differential equations above yield, for every $n\geqslant0$, $$ \tfrac32b_n+a_n+2nb_n=0,\qquad\tfrac32a_n+2na_n=s_n, $$ that is, $$ a_n=\frac{s_n}{2n+\frac32},\qquad b_n=\frac{-s_n}{(2n+\frac32)^2}. $$ This determines entirely some functions $F$ and $G$, written as series of infinite radius of convergence, such that the derivative of the RHS is the function in the integral. As a consequence, this determines completely the desired primitives on the domain $x\gt0$.


One can write the functions $F$ and $G$, hence the primitives themselves, in several equivalent ways. The one I prefer is the one above. But if one is fond of generalized hypergeometric functions, one can express $F$ and $G$ with these. For example, $F(x)=u\left(-\frac{x^2}4\right)$, with $$ u(z)=\sum\limits_{n\geqslant0}u_nz^n,\qquad u_n=\frac{4^n}{(2n+1)!(2n+\frac32)}. $$ To write $u_n$ as a ratio of Pochhammer symbols, the procedure is standard. Start from $$ \frac{u_{n+1}}{u_n}=\frac{4(2n+\frac32)}{(2n+3)(2n+2)(2n+\frac52)}=\frac{n+\frac34}{(n+\frac32)(n+\frac54)}\frac1{n+1}. $$ Since $u_0=\frac23$, $(a)_0=1$ and $(a)_{n+1}=(a)_n(a+n)$ for every $a$ and every $n\geqslant0$, this yields $$ u_n=\frac{\frac23(\frac34)_n}{(\frac32)_n(\frac54)_n}\frac1{n!}. $$ Finally, $$ F(x)=\frac23\,{}_1\!F_2\left(\frac34;\frac32,\frac54;-\frac{x^2}4\right). $$ Likewise, one can express $G$ in terms of a hypergeometric function of type ${}_2\!F_3$, an exercise we will leave to the reader.

share|improve this answer

My guess is that a series expansion is wanted.
Start from the rhs. If $J(x) = x^{3/2} (F(x) \ln(x) + G(x)) + C$, then $J'(x) = \left(\frac{3}{2} x^{1/2} F(x) + x^{3/2} F'(x) \right) \ln(x) + x^{1/2} F(x) + \frac{3}{2} x^{1/2} G(x) + x^{3/2} G'(x)$. If $F(x) = f_0 + f_1 x + f_2 x^2 + \ldots$ and $G(x) = g_0 + g_1 x + g_2 x^2 + \ldots$, this is $$ \left(\frac{3}{2} f_0 x^{1/2} + \frac{5}{2} f_1 x^{3/2} + \frac{7}{2} f_2 x^{5/2} + \ldots\right) \ln(x) + \left(\frac{3}{2} g_0 + f_0\right) x^{1/2} + \left(\frac{5}{2} g_1 + f_1\right) x^{3/2} + \left(\frac{7}{2} g_2 + f_2\right) x^{5/2} + \ldots $$ You want this to be $x^{-1/2} \sin(x) \ln(x) = \left(x^{1/2} - \frac{1}{6} x^{5/2} + \ldots\right) \ln(x)$, so you get the equations $$ \eqalign{ \frac{3}{2} f_0 = 1,\ & \frac{3}{2} g_0 + f_0 = 0\cr \frac{5}{2} f_1 = 0,\ & \frac{5}{2} g_1 + f_1 = 0\cr \frac{7}{2} f_2 = \frac{-1}{6},\ & \frac{7}{2} g_2 + f_2 = 0\cr etc &}$$

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.