I found this question from past year's maths competition in my country. I've tried any possible way to find it, but it is just way too hard.

Find $x$ if \begin{align}\frac {1} {x} + \frac {1} {y+z} &= \frac {1} {2}\\ \frac {1} {y} + \frac {1} {x+z} &= \frac{1}{3}\\ \frac {1} {z} + \frac {1} {x+y} &= \frac {1} {4}\end{align}

$(A)\;\frac 32$

$(B)\;\frac {17}{10}$

$(C)\;\frac {19}{10}$

$(D)\;\frac {21}{10}$

$(E)\;\frac {23}{10}$

EDIT: I'm very sorry guys, it should be $\frac {1} {x+y}$ not $xy$, I'm sorry for the typos (idk what is wrong with me)


closed as off-topic by Did, Ivo Terek, Ali Caglayan, Tim Raczkowski, Jonas Meyer Jul 26 '15 at 0:00

This question appears to be off-topic. The users who voted to close gave this specific reason:

  • "This question is missing context or other details: Please improve the question by providing additional context, which ideally includes your thoughts on the problem and any attempts you have made to solve it. This information helps others identify where you have difficulties and helps them write answers appropriate to your experience level." – Did, Ivo Terek, Ali Caglayan, Tim Raczkowski, Jonas Meyer
If this question can be reworded to fit the rules in the help center, please edit the question.

  • $\begingroup$ Hint: start with the last formula $$\frac {1} {x} +\frac {1} {y} + \frac {1} {z} + \frac {1} {xy} + \frac {1} {yz} + \frac {1} {xz}= \frac {13}{12}$$ and multiply both sides by $xyz.$ $\endgroup$ – Khadija Mbarki Jul 25 '15 at 9:45
  • $\begingroup$ are you sure that you have made no typo? $\endgroup$ – Dr. Sonnhard Graubner Jul 25 '15 at 9:53
  • $\begingroup$ are $x,y,z$ real numbers? $\endgroup$ – Dr. Sonnhard Graubner Jul 25 '15 at 9:59
  • $\begingroup$ do you want to find special method to find x?or just find x? it is easy to find one value from first equation put another by this way find others $\endgroup$ – haqnatural Jul 25 '15 at 10:03
  • $\begingroup$ you mean eliminate the variable $x$? $\endgroup$ – Dr. Sonnhard Graubner Jul 25 '15 at 10:17

\begin{align} \frac {1} {x} + \frac {1} {y+z} &= \frac {1} {2}\\ \frac {1} {y} + \frac {1} {x+z} &= \frac{1}{3}\\ \frac {1} {z} + \frac {1} {x+y} &= \frac {1} {4} \end{align}

Is equivalent to the system \begin{align} xy+xz&=2(x+y+z) \\ xy+yz&=3(x+y+z) \\ xz+yz&=4(x+y+z), \end{align} Solving this system as a linear system of equations in terms of $xy,xz$ and $yz$ and $x+y+z$, we arrive at

\begin{align} xy&=\tfrac12 (x+y+z), \quad(1) \\ yz&=\tfrac52 (x+y+z), \\ xz&=\tfrac32 (x+y+z). \end{align}

Dividing these equations pairwise, we get \begin{align} z&=5x \\ y&=\tfrac53 x, \end{align}

combining with (1) we have

\begin{align} \tfrac{5}{3}x^2-\tfrac{23}{6}x&=0 \end{align} and since $x\ne0$, the answer is $x=\tfrac{23}{10}$.


NOte: edit #1, equations changed when typo was noticed NOTE!!! This is not a complete solution, and there is likely something wrong with it. I'm posting it for 2 reasons: 1: To show the type of logic one would use, and 2: To see if anyone can spot the mistake so it can be fixed. (and 3, I spent so much time on it I don't want to delete it!)


Multiply through each equation by $xyz$ to get rid of the factions, and turn it into 3 equations in 3 unknowns: $$yz+x=\frac 1 2 xyz $$ $$xz +y=\frac 1 3 xyz $$ $$xy+z=\frac 1 4 xyz $$

Let's go backwards and use the third equation to solve for z: Subtract z from both sides and factor out the z, and you get $$xy=z(\frac 1 4 xy -1)$$, to clear the fractions I'm going to multiply through by 4 for convenious, giving $$4xy=z(xy-4)$$, thus $$z=\frac {4xy} {xy-4}$$ Now we plug that into our first two equations for $z$, giving us $$y(\frac {4xy} {xy-4})+x=\frac 1 2 xy(\frac {4xy} {xy-4}) $$ and $$x(\frac {4xy} {xy-4})+y=\frac 1 3 xy(\frac {4xy} {xy-4}) $$

Now, we're going to solve the second equation for $y$. Again, it's much easier to get rid of the fractions first, so I'm going to multiply the whole equation by $3(xy-4)$ to get $$12x^2y+3xy^2-12y^2=4x^2y^2$$ Every term has a $y$ and $y\ne 0$, so cancelling, we get $$12x^2+3xy-12y=4x^2y$$ Move the two terms with $y$ to the right hand side and factor out the $y$, and we get $$12x^2=y(4x^2+12-3x)$$, hence $$y=\frac {12x^2} {4x^2 +12 -3x}$$

Now I need to plug that into our first equation, but before I do that, I'm going to clear out the denominators in the first equations, so I multiply through by $2(xy-4)$ to give us $$8x^2y^2+2x(xy-4)=xy$$ Again at this step I can divide through by an $x$, giving us $$8xy^2+2xy-8=y$$

Now, we plug in what we have $y$ for in terms of x, giving us $$8x(\frac {12x^2} {4x^2 +12 -3x})^2+2x(\frac {12x^2} {4x^2 +12 -3x})-8=\frac {12x^2} {4x^2 +12 -3x}$$

Continuing our theme of not dealing with fractions, we're going to multipy through by $(4x^2-12x-3)^2$ to get

$$8x(12x^2)^2+24x^3(4x^2-12x-3)-8(4x^2-12x-3)^2=12x^2(4x^2-12x-3) $$

Multiply this out and we have $$1152x^5+96x^5-288x^4-72x^3-128x^4+768x^3-960x^2-576x-72=48x^4-144x^3-36x^2$$ Moving everything to the left hand side and combining powers of $x$, we have $$1248x^5-464x^4+840x^3-924x^2-582x-72=0$$ The only simplification that can be done here is dividing by 2, which still gives us a quintic, which has no easy way to solve. Wolframalpha shows is now showing 3 real solutions, none of which match

  • $\begingroup$ I'm sure there is a mistake somewhere in your calculation. There are five possible real solutions to this system of equations (I have checked with Mathematica). However, none of the real solutions are nice. If $a:=\frac{1}{x}$, then $114a^5-72a^4-288a^3+156a^2+119a-60=0$. $\endgroup$ – Batominovski Jul 25 '15 at 10:39
  • $\begingroup$ my question had typo, should be $\frac {1} {x+y}$ etc, not $xy$.. $\endgroup$ – wuiyang Jul 25 '15 at 11:06
  • $\begingroup$ II changed the equation. $\endgroup$ – Alan Jul 25 '15 at 11:06

Let $x=1/p$, $y=1/q$, and $z=1/r$. The three equations become

$$\begin{align} {1\over2}&=p+{qr\over q+r}={s\over q+r}\\ {1\over3}&=q+{rp\over r+p}={s\over r+p}\\ {1\over4}&=r+{pq\over p+q}={s\over p+q}\\ \end{align}$$




$$\begin{align} 2s&=q+r\\ 3s&=r+p\\ 4s&=p+q\\ \end{align}$$

From this it follows that

$$(q+r)+(p+q)=2(r+p)\implies2q=r+p=3s\implies q={3\over2}s$$ and the others follow:

$$r=2s-q={1\over2}s\quad\text{and}\quad p=4s-q={5\over2}s$$

But now we have


hence $s=4/23$, so $p=10/23$, and thus $x=23/10$.


$p = \frac{1}{x} $ $q = \frac{1}{y} $ $r = \frac{1}{z} $

will certainly help writing the equations more neatly, I guess. Then your equations can be written as:

$$2p + 2qr = 1 \\ 3q + 3pr = 1 \\ 4r + 4pq = 1$$

using first two equations you get:

$$ 2p + 2qr = 3q + 3pr \\ \implies r = \frac{3q - 2p}{3p - 2q}$$

Substitute in equations and solve as solved by Alan above.


multiplying (2) by $x$ $(x \ne 0)$ we obtain $$\frac{1}{z}=\frac{1}{4}-\frac{1}{xy}$$ with (3) we obtain $$y=\frac{12(x^2-1)}{4x^2-3x}$$ (I) from (1) and (3) we get $$\frac{x^2y}{3}-x^2+1=\frac{xy}{4}$$ plugging (I) in this equation and simplifying we get $$-60 x^5+119 x^4+156 x^3-288 x^2-72 x+144=0$$ with five real solutions, no of them from (A) to (E) the system after the correction has the solution $$x=\frac{23}{10}$$

  • $\begingroup$ my question had typo, should be $\frac {1} {x+y}$ etc, not $xy$.. sorry $\endgroup$ – wuiyang Jul 25 '15 at 11:06
  • $\begingroup$ hm it was also my time when I have solved your problem with the typo $\endgroup$ – Dr. Sonnhard Graubner Jul 25 '15 at 11:08
  • $\begingroup$ and now i'm waiting for the downvoters of me $\endgroup$ – Dr. Sonnhard Graubner Jul 25 '15 at 11:10

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