Unpredictable!

Calculus Level 5

Evaluate 0 0 0 sin x sin y sin z sin ( x + y + z ) x y z ( x + y + z ) d x d y d z \displaystyle \int\limits_0^\infty \int\limits_0^\infty \int\limits_0^\infty \dfrac{\sin x\sin y\sin z\sin(x+y+z)}{xyz(x+y+z)}\; dx\;dy\; dz

If you think it diverges, Enter 1.11 .

If you think it converges, enter your answer up to two decimal places.


The answer is 0.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

2 solutions

Some results

I 1 0 1 ln 2 ( 1 x 1 + x ) d x = π 2 3 I 2 = 0 sin x cos a x x d x = π 2 I 3 = 0 sin x sin a x x d x = 1 2 ln ( 1 a 1 + a ) \displaystyle \begin{array}{c}\\ I_1 & \int\limits_0^1 \ln^2 \left(\frac{1-x}{1+x}\right) \; dx = \frac{\pi^2}{3} \\ I_2= & \int\limits_0^\infty \frac{\sin x \cos ax}{x}\; dx =\frac{\pi}{2} \\ I_3 = & \int\limits_0^\infty \frac{\sin x \sin ax}{x}\; dx = \frac{1}{2}\ln \left(\frac{1-a}{1+a}\right)\end{array}\\

Now consider , F ( a ) = 0 0 0 sin x sin y sin z sin ( a ( x + y + z ) ) x y z ( x + y + z ) d x d y d z \displaystyle F(a)=\int\limits_0^\infty \int\limits_0^\infty \int\limits_0^\infty \dfrac{\sin x\sin y\sin z\sin(a(x+y+z))}{xyz(x+y+z)}\; dx\;dy\; dz

By differentiating once we have,

F ( a ) = 0 0 0 sin x sin y sin z cos ( a x + a y + a z ) x y z ( x + y + z ) d x d y d z \displaystyle F'(a) = \int\limits_0^\infty \int\limits_0^\infty \int\limits_0^\infty \dfrac{\sin x\sin y\sin z\cos(ax+ay+az)}{xyz(x+y+z)}\; dx\;dy\; dz

Now expanding cos ( a x + a y + a z ) \cos(ax+ay+az) and then expressing the integral as a sum integral we find that that there are no homogenous terms and the integrals are in variable separable form like,

F ( a ) = ( 0 sin x cos a x x d x ) ( 0 sin y cos a y y d y ) ( 0 sin z cos a z z d z ) ( 0 sin x cos a x x d x ) ( 0 sin y sin a y y d y ) ( 0 sin z cos a z z d z ) ( 0 sin x sin a x x d x ) ( 0 sin y cos a y y d y ) ( 0 sin z sin a z z d z ) ( 0 sin x sin a x x d x ) ( 0 sin y sin a y y d y ) ( 0 sin z cos a z z d z ) \displaystyle F'(a) = \left( \int_0^\infty \frac{\sin x \cos ax}{x}\; dx\right)\left( \int_0^\infty \frac{\sin y \cos ay}{y}\; dy\right)\left( \int_0^\infty \frac{\sin z \cos az}{z}\; dz\right) - \left( \int_0^\infty \frac{\sin x \cos ax}{x}\; dx\right)\left( \int_0^\infty \frac{\sin y \sin ay}{y}\; dy\right)\left( \int_0^\infty \frac{\sin z \cos az}{z}\; dz\right)- \left( \int_0^\infty \frac{\sin x \sin ax}{x}\; dx\right)\left( \int_0^\infty \frac{\sin y \cos ay}{y}\; dy\right)\left( \int_0^\infty \frac{\sin z \sin az}{z}\; dz\right)- \left( \int_0^\infty \frac{\sin x \sin ax}{x}\; dx\right)\left( \int_0^\infty \frac{\sin y \sin ay}{y}\; dy\right)\left( \int_0^\infty \frac{\sin z \cos az}{z}\; dz\right)

Now the integrals in brackets are actually independent integrals so this thing can be rewritten as,

F ( a ) = ( I 2 ) 3 3 ( I 2 ) ( I 3 ) 2 = π 3 8 3 π 8 ( ln 2 ( 1 a 1 + a ) ) \displaystyle F'(a)= (I_2)^3-3(I_2)(I_3)^2 = \frac{\pi^3}{8}-3\frac{\pi}{8}\left(\ln^2 \left(\frac{1-a}{1+a}\right)\right)

Since F ( 0 ) = 0 F(0)=0 and we want F ( 1 ) F(1) so,

F ( 1 ) = 0 1 π 3 8 d a 3 π 8 I 1 = π 3 8 π 3 8 = 0 \displaystyle F(1)= \int_0^1 \frac{\pi^3}{8}\; da -3\frac{\pi}{8}I_1 = \frac{\pi^3}{8}-\frac{\pi^3}{8}=0

Interested users having issues with the proofs of first three results may ask me for proofs.

There is a small typo in your formula for F ( a ) F'(a) . The term x + y + z x+y+z should not be in the denominator.

It is always a little dangerous differentiating under the integral sign when your resulting integrand only exists as an improper Riemann integral, and that is what you are doing when identifying F ( a ) F'(a) as a product of the I 2 , I 3 I_2,I_3 . The theorem that allows for differentiation under the integral sign is either expressed in terms of proper Riemann integrals over bounded intervals, or else Lebesgue integrable functions over more general intervals. The integrands in I 2 , I 3 I_2,I_3 are not Lebesgue integrable.

Mark Hennings - 4 years, 3 months ago

Can you tell me where did you get this problem?

Ishan Singh - 4 years ago

Log in to reply

The case of two variables was proposed by Cornel Loan Valen in the AMM monthly. I solved out the general one and posted it for the case of 3 variables. So it's a more complicated version of an AMM problem.

Mark Hennings
Feb 15, 2017

If we define F ( α , β , γ , δ ) = 0 0 0 sin x sin y sin z e i ( x + y + z ) e α x e β y e γ z e δ ( x + y + z ) x y z ( x + y + z ) d x d y d z = 0 0 0 sin x sin y sin z e ( α + δ i ) x e ( β + δ i ) y e ( γ + δ i ) z x y z ( x + y + z ) d x d y d z \begin{aligned} F(\alpha,\beta,\gamma,\delta) & = \; \int_0^\infty \int_0^\infty \int_0^\infty \frac{\sin x \sin y \sin z \,e^{i(x+y+z)}e^{-\alpha x}e^{-\beta y}e^{-\gamma z}e^{-\delta(x+y+z)}}{x y z (x+y+z)}\,dx\,dy\,dz \\ & = \int_0^\infty \int_0^\infty \int_0^\infty \frac{\sin x \sin y \sin z \,e^{-(\alpha+\delta-i)x}e^{-(\beta+\delta-i)y}e^{-(\gamma+\delta-i)z}}{x y z (x+y+z)}\,dx\,dy\,dz \end{aligned} then 4 F α β γ δ = 0 0 0 sin x sin y sin z e ( α + δ i ) x e ( β + δ i ) y e ( γ + δ i ) z d x d y d z = 1 [ ( α + δ i ) 2 + 1 ] [ ( β + δ i ) 2 + 1 ] [ ( γ + δ i ) 2 + 1 ] = 1 ( α + δ ) ( α + δ 2 i ) ( β + δ ) ( β + δ 2 i ) ( γ + δ ) ( γ + δ 2 i ) \begin{aligned} \frac{\partial^4F}{\partial\alpha\partial\beta\partial\gamma\partial\delta} & = \int_0^\infty \int_0^\infty \int_0^\infty \sin x \sin y \sin z \,e^{-(\alpha+\delta-i)x}e^{-(\beta+\delta-i)y}e^{-(\gamma+\delta-i)z}\,dx\,dy\,dz \\ & = \frac{1}{\big[(\alpha+\delta- i)^2+1\big]\big[(\beta+\delta-i)^2 + 1\big]\big[(\gamma+\delta-i)^2+1\big]} \\ & = \; \frac{1}{(\alpha+\delta)(\alpha+\delta-2i)(\beta+\delta)(\beta+\delta-2i)(\gamma+\delta)(\gamma+\delta-2i)} \end{aligned} and so F δ = g ( α + δ ) g ( β + δ ) g ( γ + δ ) \frac{\partial F}{\partial \delta} \; = \; -g(\alpha+\delta)g(\beta+\delta)g(\gamma+\delta) where g ( u ) = u d v v ( v 2 i ) = 1 2 i ln ( 1 2 i u ) u > 0 g(u) \; = \; \int_u^\infty \frac{dv}{v(v-2i)} \; = \; \tfrac12i\ln\big(1 - \tfrac{2i}{u}\big) \hspace{1cm} u > 0 Thus F ( α , β , γ , δ ) = 1 8 i δ ln ( 1 2 i α + u ) ln ( 1 2 i β + u ) ln ( 1 2 i γ + u ) d u F(\alpha,\beta,\gamma,\delta) \; = \; -\tfrac18i\int_\delta^\infty \ln\big(1 - \tfrac{2i}{\alpha+u}\big)\,\ln\big(1 - \tfrac{2i}{\beta+u}\big)\,\ln\big(1 - \tfrac{2i}{\gamma+u}\big)\,du and hence F ( 0 , 0 , 0 , δ ) = 1 8 i δ [ ln ( 1 2 i u ) ] 3 d u = 1 4 i 0 2 δ ln 3 ( 1 v i ) v 2 d v F(0,0,0,\delta) \; = -\tfrac18i\int_\delta^\infty \left[\ln\big(1 - \tfrac{2i}{u}\big)\right]^3\,du \; = \; -\tfrac14i\int_0^{\frac{2}{\delta}} \frac{\ln^3(1 - vi)}{v^2}\,dv and so F ( 0 , 0 , 0 , 0 ) = 1 4 i 0 ln 3 ( 1 v i ) v 2 d v F(0,0,0,0) \; = \; -\tfrac14i \int_0^\infty \frac{\ln^3(1 - vi)}{v^2}\,dv A simple piece of contour integration shows that F ( 0 , 0 , 0 , 0 ) = 1 4 i 0 ln 3 ( 1 v i ) v 2 d v = 1 4 0 ln 3 ( 1 + x ) x 2 d x F(0,0,0,0) \; = \; -\tfrac14i\int_0^\infty \frac{\ln^3(1 - vi)}{v^2}\,dv \; = \; -\tfrac14\int_0^\infty \frac{\ln^3(1+x)}{x^2}\,dx and hence we deduce that 0 0 0 sin x sin y sin z sin ( x + y + z ) x y z ( x + y + z ) d x d y d z = 0 \int_0^\infty \int_0^\infty \int_0^\infty \frac{\sin x \sin y \sin z \sin(x+y+z)}{x y z (x+y+z)}\,dx\,dy\,dz \; = \; \boxed{0} We have also shown that 0 0 0 sin x sin y sin z cos ( x + y + z ) x y z ( x + y + z ) d x d y d z = 1 4 0 ln 3 ( 1 + x ) x 2 d x = 3 2 ζ ( 3 ) \int_0^\infty \int_0^\infty \int_0^\infty \frac{\sin x \sin y \sin z \cos(x+y+z)}{x y z (x+y+z)}\,dx\,dy\,dz \; = \; -\tfrac14\int_0^\infty \frac{\ln^3(1+x)}{x^2}\,dx \; = \; -\tfrac32\zeta(3) for free.

Thanks for an alternate solution, some of the integrals are a bit more complicated than they look such as the last ln (1+x) integral. :)

Aditya Narayan Sharma - 4 years, 3 months ago

Log in to reply

Not really. Firstly, the last integral is inessential to the question. Moreover the substitution 1 + x = e y 1 + x = e^y gives 0 ln 3 ( 1 + x ) x 2 d x = 0 y 3 ( e y 1 ) 2 e y d y = 0 y 3 e y ( 1 e y ) 2 d y = n = 1 n 0 y 3 e n y d y = n = 1 n × 3 ! n 4 = 6 ζ ( 3 ) \begin{aligned} \int_0^\infty \frac{\ln^3(1+x)}{x^2}\,dx & = \int_0^\infty \frac{y^3}{(e^y-1)^2}e^y\,dy \; = \; \int_0^\infty y^3 e^{-y}(1 - e^{-y})^{-2}\,dy \\ & = \sum_{n=1}^\infty n\int_0^\infty y^3 e^{-ny}\,dy \; = \; \sum_{n=1}^\infty n \times \frac{3!}{n^4} \\ & = 6\zeta(3) \end{aligned} The formula 0 y μ 1 sinh 2 a y d y = 4 ( 2 a ) μ Γ ( μ ) ζ ( μ 1 ) R e a > 0 , R e μ > 2 \int_0^\infty \frac{y^{\mu-1}}{\sinh^2ay}\,dy \; = \; \frac{4}{(2a)^\mu}\Gamma(\mu)\zeta(\mu-1) \hspace{1cm} \mathrm{Re}\,a > 0\,,\, \mathrm{Re}\mu > 2 is the general version of this last integral.

Mark Hennings - 4 years, 3 months ago

Log in to reply

Wow! The general one is really a wonderful result

Aditya Narayan Sharma - 4 years, 3 months ago

@Mark Hennings - Your solutions are absolutely amazing! How did you learn to master LaTeX like this? You must spend more time on the writing of solutions than actually doing the problems. Thank you for teaching our Brilliant Community so much !!

Bob Kadylo - 4 years, 1 month ago

Typo in the first line. It should be e δ ( x + y + z ) e^{-\delta(x+y+z)} instead of e δ ( a + y + z ) e^{-\delta(a+y+z)} .

Ishan Singh - 4 years ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...