7
$\begingroup$

I'm trying to find the length of $x$ in the following figure, without needing to solve a quartic.

enter image description here

With Pythagoras it's easy to show, that $x^4+2x^3-x^2-6x-3=0$ .

Using any calculator gives the solution of $x=\phi$. Also the segment of length $\sqrt{3}$ bisects the side of the square in the golden ratio. Is there a way to calculate $x$ by using properties of $\phi$ or the golden ratio to circumvent the quartic?

I've seen the very similar thread, where the "solution" is to type the problem into wolfram alpha.

$\endgroup$
4
  • 3
    $\begingroup$ It does not answer your question, but $\phi$ is a root since $x^4+2x^3-x^2-6x-3=(x^2 - x - 1) (x^2 + 3 x + 3)$ $\endgroup$ Commented Jun 30 at 13:51
  • $\begingroup$ what do you mean by circumvent. also they did need to solve a quartic, but now that you know that $\phi$ is a root of you polynomial, try factoring by $x^2-x-1$ the solution will be very clear $\endgroup$ Commented Jun 30 at 14:02
  • $\begingroup$ With "circumvent" i mean to find a different solution method. Initially i don't know, that $\phi$ is a solution. But maybe i can show, that a certain ratio holds in the case of $\sqrt{3}$ to get the solution $\phi$ for free. I'm just not sure, if solving a quartic is the only way. $\endgroup$ Commented Jun 30 at 14:36
  • $\begingroup$ Generalizing my comment to the similar question, replacing $\sqrt3$ with $p$, the corresponding quartic factors: $$x^4+2x^3+(2-p^2)x^2-2p^2x-p^2=(x^2+xu+u)(x^2+xu'+u')$$ where $u=1-\sqrt{p^2+1}$ and $u':=1+\sqrt{p^2+1}$. The $u'$-factor is clearly extraneous, so we only care about $u$. Note that $\sqrt{p^2+1}$ is geometrically meaningful, as the hypotenuse of a right triangle with legs $1$ and $p$ (the only lengths given in the problem); thus also $u$ is meaningful. Perhaps there's a direct route to the quadratic that avoids the quartic. $\endgroup$ Commented Jun 30 at 21:09

6 Answers 6

8
$\begingroup$

Comment: You can use this property of pentagon which is:

$\phi=\frac{diagonl}{side}$

For prove see https://www.youtube.com/watch?v=lL9VN8ELH1M&pp=ygUMI3dhemlyeGdyYXBo

The construction below is based on this property.

enter image description here

Where $BF=m= \sqrt 3$

Since $AB=1$ then $BE=\phi$

$\endgroup$
1
  • 1
    $\begingroup$ That's a nice idea. But how do we know, that $AB=1$ or rather know that the angles match a regular pentagon? $\endgroup$ Commented Jul 2 at 23:11
1
$\begingroup$

The easy solution. If the vertical side is cut in the golden ratio proportion, then the two pieces are of length $1/\varphi+1/\varphi^2=1$. Then by the Pythagorean theorem, $x=\sqrt{3-1/\varphi^2}=\varphi$. (Hint: we've already seen that $1/\varphi^2=1-1/\varphi$, so substitute this under the square root).

$\endgroup$
2
  • $\begingroup$ The fact, that the vertical is cut in that proportion is just something i noticed, if one already knows that $x=\phi$. $\endgroup$ Commented Jul 2 at 23:16
  • $\begingroup$ Of course that is so from the Pythagorean theorem. But I found the division from the proportions of the side, not from any triangle. This is the only direct answer that has been given. $\endgroup$ Commented Jul 4 at 0:20
0
$\begingroup$

We play with Fibonacci numbers.

$F_{n+4}+2F_{n+3}-F_{n+2}-6F_{n+1}-3F_n$

$=F_{n+3}+F_{n+2}+2F_{n+3}-F_{n+2}-6F_{n+1}-3F_n$

$=3(F_{n+3}-2F_{n+1}-F_n)$

$=3(F_{n+2}+F_{n+1}-2F_{n+1}-F_n)$

$=3(F_{n+2}-F_{n+1}-F_n)=0.$

As $n\to\infty,F_{n+k}/F_n\to\phi^k$, so the above Fibonacci number result implies

$\phi^4+2\phi^3-\phi^2-6\phi-3=0.$

$\endgroup$
0
$\begingroup$

Instead of showing that $\sqrt 3$ entails $\phi$ in the posted figure, the following proceeds conversely, that $\phi$ entails $\sqrt 3$.

I. In Elements, VI, 30 (Dover, tr. Heath), Euclid shows how to cut a given segment $AB$ in extreme and mean ratio, i.e. so that$$\frac{AB}{AE}=\frac{AE}{EB}$$with $AE$ the greater segment: he constructs square $BC$ and applies on side $AC$ rectangle $DC=BC$ (VI, 29), and the proportion follows from VI, 14.

II. Extend Euclid's figure now: construct square $AN=AD$, square $JL=BC$, and join $LB$. Since $N$ also cuts $JK$ in extreme and mean ratio, making $\triangle KLN\cong\triangle EFB$, then $LB$ passes through $N$.

Since $AJ=AE$ has been added to $AB$, $A$ now cuts $JB$ in extreme and mean ratio (Elements, XIII, 5), and supposing $AB=1$, then$$JB=\phi$$Since $MJ=AB=1$, it remains to show that $NB=\sqrt 3$ as in the posted figure.

x=phi

III. Let $NP\perp NB$ meet $HC$ extended at $P$, let $PQ\perp NP$ meet $BF$ extended at $Q$, extend $KJ$ to $S$, and draw $QR\perp PH$.

Since $NP$, $NB$ are each hypotenuse of right triangles $NPS\cong BNJ$, then rectangle $NBQP$ is a square, in which$$\triangle NPS+\triangle PQR=CFDG=ABHC=1$$Further, since $\triangle ABO\cong\triangle HBF$, then quadrilateral$$CFBO=ABHC=1$$Lastly, since $\triangle QRF\cong \triangle NGO$, then trapezoid$$SNOC+\triangle NGO=ABHC=1$$

Therefore, square$$NBQP=3ABHC$$and$$NB=\sqrt 3$$

$\endgroup$
0
$\begingroup$

It is possible to avoid the complicated quartic that you came up with, by solving a quadratic equation for $x^2$ instead:

Consider the right triangle with hypotenuse $\sqrt3$ and base $x$. Let's call its height $y$. Then $$\frac{x}{y} = \frac{x+1}{1}$$ Or $$\frac{1}{y} - \frac{1}{x} = 1$$ So $$x-y = xy$$ Pythagoras': $$x^2 + y^2 = 3$$ $$x^2 + y^2 - 2xy = 3 - 2(x-y)$$ $$(x-y)^2 + 2(x-y) - 3 = 0$$ Therefore $$x-y = xy = 1$$ The other answer ($x-y = -3$) is not acceptable (why not?)

Pythagoras' again: $$x^2 + \frac{1}{x^2} = 3$$ $$x^4 - 3x^2 + 1 = 0$$ This is the quartic we have to solve, but luckily it is really a quadratic to solve for $x^2$ $$x^2 = \frac{3 \pm \sqrt5}{2}$$ Smells like golden ratio! Let's try to calculate further from $x^2$ to $x$: $$x^2 = \frac{6 \pm 2\sqrt5}{4} = \frac{1+5 \pm 2\sqrt5}{4} = (\frac{1 \pm \sqrt5}{2})^2$$ Finally, noting the constraint that $y = \frac{1}{x} < 1$, we are left with only one answer: $x = \frac{1 + \sqrt5}{2}$

$\endgroup$
-1
$\begingroup$

We can calculate the side lengths of the square exactly. For example, if the short leg is $y$, then the long leg is $y\phi$. Then also $y+y\phi=1$. The long leg is now given by $$y\phi=\frac{\phi}{1+\phi}=\frac{1}{\phi}=\phi-1.$$ Here I used $1+\phi=\phi^2$ and $\frac{1}{\phi}=\phi-1$. We can now use the ratio of opposite over adjacent: $$\frac{\phi-1}{x}=\frac{1}{1+x}.$$ I'll leave the final steps for the reader.

construction with side lengths of square added

$\endgroup$

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.