4

Eliminate $\theta$ from the equations. $$\cos^3\theta +a\cos\theta =b$$ $$\sin^3\theta +a\sin\theta =c$$

Can anyone solve this question?

Blue
  • 75,673
  • 5
    If you show us what you tried, someone could probably either find an error you have help you finish. By the way "Can anyone solve this question?" has only 2 answers: Yes or No. – MasB Oct 02 '21 at 13:00
  • 1
    What does it even mean to eliminate a variable? Do you mean solve for $\theta$? – William Oct 02 '21 at 13:18
  • @William The two equations are not independent, so there must be some relation between $a,b,c$ so one must find this relation. – Rene Schipperus Oct 02 '21 at 13:20
  • @William Solving for $\theta$ is not enough, I can do this. You need the condition on $a,b,c$. – Rene Schipperus Oct 02 '21 at 13:23
  • @ReneSchipperus I suppose you are not looking for half-angle tangent substitution or similar tricks. – user Oct 02 '21 at 13:53
  • 1
    @user Yes, those techniques are good to use. And probably are essential to the solution. I can solve for $\cos(4\theta)$. One can use the triple angle formula change the exponent $3$ to $\cos (3\theta)$ and $\sin(3\theta)$. In theory you can replace $\cos\theta, \sin\theta$ by $x,y$ with the relation $x^2+y^2=1$, but I have made more progress with other methods. – Rene Schipperus Oct 02 '21 at 14:01
  • @ReneSchipperus I'm really curious to see your solution later! Bye – user Oct 02 '21 at 14:02
  • @user Hardly, I am very frustrated by this question. Which is why I ask it. – Rene Schipperus Oct 02 '21 at 14:03
  • 1
    Using Mathematica and with $x=\cos\theta$, the Resultant$[x^3+ax-b, (1-x^2)^3+2a(1-x^2)^2+a^2(1-x^2)^2-c^2, x] = 16 a^8+96 a^7-16 a^6 b^2-16 a^6 c^2+248 a^6-72 a^5 b^2-72 a^5 c^2+360 a^5-136 a^4 b^2-136 a^4 c^2+321 a^4+8 a^3 b^4-48 a^3 b^2 c^2-138 a^3 b^2+8 a^3 c^4-138 a^3 c^2+180 a^3+17 a^2 b^4-110 a^2 b^2 c^2-79 a^2 b^2+17 a^2 c^4-79 a^2 c^2+62 a^2+12 a b^4-84 a b^2 c^2-24 a b^2+12 a c^4-24 a c^2+12 a-b^6-3 b^4 c^2+3 b^4-3 b^2 c^4-21 b^2 c^2-3 b^2-c^6+3 c^4-3 c^2+1$, so doesn't seem its going to really simplify much. – Macavity Oct 02 '21 at 18:08
  • @Macavity Thanks for that. I have solved the problem and get the result, $[(1+a)^2-(b^2+c^2)]\left((1+a)(2a^2+2a+1)-(b^2+c^2)\right)^2=b^2c^2(2a+1)^3$ Its not exactly the same as your result but very similar, so maybe there is a constant missing in my answer. Maybe I will give my calculation as an answer to the question. – Rene Schipperus Oct 02 '21 at 19:19
  • @ReneSchipperus Not sure it simplifies much in the end and didn't follow it all the way through, but another way could be to let $,z= \cos \theta + i \sin \theta,$ then the condition is equivalent to $,p(z)=z^4 + u z + v,$ having a root on the unit circle, where $,u=-4(b-ic),$ and $,4a+3,$ (if I got those right). – dxiv Oct 03 '21 at 00:09
  • This question actually comes from Durell & Robson's Advanced Trigonometry, Section 14g, question #21. It is a very difficult question with multiple approaches, and I give a hat tip to those who made the attempts to make the solutions easier to understand. – bjcolby15 Oct 05 '21 at 22:54
  • @bjcolby15 Thanks for the reference, I did not know that book. I got it from Hobson's Treatise on Plane Trigonometry. I have asked another question from the same book, it hasnt got any traction, so maybe ill set a bounty on it. https://math.stackexchange.com/questions/4268371/five-porismatic-equations – Rene Schipperus Oct 06 '21 at 11:18
  • @ReneSchipperus I've been trying to solve that system for months, but it's always best to show several approaches. (Your solution letting $x = \cos \theta$ and $y = \sin \theta$ was pure genius.) – bjcolby15 Oct 06 '21 at 23:22

4 Answers4

4

$\def\¿{\mathcal}$ Maybe I am complicating this, yet another method.

I will use these identities:

$\cos^6x+\sin^6x=1-\frac34\sin^22x$

$\cos^6x-\sin^6x=\cos 2x(1-\frac14\sin^22x)$

$\cos^4x+\sin^4x=1-\frac12\sin^22x$

$\cos^4x-\sin^4x=\cos2x$

Let
$c=\cos x$,
$s=\sin x$,
$\¿ C=\cos 2x$,
$\¿S=\sin2x$

I use $d$ instead of $c$ in the original problem to avoid confusion.

Now we have $$c^3+ac=b\tag{1}$$ $$s^3+as=d\tag{2}$$


$\small\mathit{(1)^2+(2)^2}$, $$c^6+s^6+2a(c^4+s^4)+a^2(c^2+s^2)=b^2+d^2$$ $$\left(1-\frac34\¿S^2\right)+2a\left(1-\frac12\¿S^2\right)+a^2=b^2+d^2$$ $$\¿S^2=\frac{4\left[\left(a+1\right)^2-(b^2+d^2)\right]}{(3+4a)}$$


$\small\mathit{(1)^2-(2)^2}$, $$c^6-s^6+2a(c^4-s^4)+a^2(c^2-s^2)=b^2-d^2$$ $$\¿C\left(1-\frac14\¿S^2\right)+2a\¿C+a^2\¿C=b^2-d^2$$ $$\¿C\left[(a+1)^2-\frac14\¿S^2\right]=b^2-d^2$$ $$\¿C=\frac{(3+4a)(b^2-d^2)}{2(a+1)^2(2a+1)+(b^2+d^2)}$$ Now we can use $$\¿C^2+\¿S^2=1$$

ACB
  • 3,713
4

So, if I understood properly your question, we have $$ \left\{ \begin{array}{l} \theta ,a,b,c \in R \\ \cos ^3 \theta + a\cos \theta - b = 0 \\ \sin ^3 \theta + a\sin \theta - c = 0 \\ \end{array} \right. $$ and we are looking for which domain of the four variables $ \theta ,a,b,c$ the system of the two equations has one or more (real) solutions.

Both equations are depressed cubics.

The discriminants are $$ \begin{array}{l} \Delta _1 = - 4a^3 - 27b^2 \\ \Delta _2 = - 4a^3 - 27c^2 \\ \end{array} $$ If

  • $\Delta >0$ the equation has three distinct real roots;
  • $\Delta = 0$ all roots are still real , and if $a=0$ gives three coincident roots, otherwise one simple root at $-3a/b; -3a/c$ and two coincident at $-3a/(2b) ; -3a/(2c)$;
  • $\Delta < 0$ the equation has only one real root.

Then, denoting by $c_1, c_2, c_3$ the roots of the equation in $\cos$, eventually reduced to only two or one, and by $s_1, s_2, s_3$ those of the equation in $\sin$, you shall impose to have at least a couple for which $$ \left| {c_{\,k} } \right| \le 1\quad \wedge \quad \left| {s_{\,k} } \right| \le 1\quad \wedge \quad c_{\,k} ^2 + s_{\,k} ^2 = 1 $$

That means to write each cubic as the product of the three monomials $(x-c_k)$, etc. and patiently translate the above bounds and conditions onto $a,b,c$ by means of Vieta's formulas.

We get in fact $$ \left\{ \begin{array}{l} c_1 + c_2 + c_3 = 0 \\ c_1 c_2 + c_1 c_3 + c_2 c_3 = a \\ c_1 c_2 c_3 = b \\ \end{array} \right.\quad \Rightarrow \quad \left\{ \begin{array}{l} c_1 = - \left( {c_2 + c_3 } \right) \\ c_2 ^2 + c_2 c_3 + c_3 ^2 = \frac{3}{4}\left( {c_2 + c_3 } \right)^2 + \frac{1}{4}\left( {c_2 - c_3 } \right)^2 = - a \\ \left( {c_2 + c_3 } \right)c_2 c_3 = - b \\ \end{array} \right. $$ and similarly for the sine, changing $b$ with $c$.
So that, if the three solutions are real, the ellipse in the second line demand $a$ to be negative, resulting in the two curves are sketched below. Eq_sin^3_1

Note that, the properties of the cubic ensure that there is a minimum and maximum value for $b$ (in function of $a$) for which the two curves simultaneously touch each other, and such values are easy to determine imposing the tangency along $c_2 = c_3$.
And we shall impose that that point at least be within $[-1,1] $, etc.
Finally note that the intersection points are six, since we have not ordered the roots.

G Cab
  • 35,272
4

I have now my own answer developed out of conversation with David Quinn, so I thank him. I publish this answer to demonstrate the calculation. It seems to coincide with the calculation of Macavity in the comments above. Wow what a monster of a question.

We have $$x^3+ax=b$$ $$y^3+ay=c$$ Where $x^2+y^2=1$.

Adding and subtracting,

$$x^3+y^3+a(x+y)=b+c$$ $$x^3-y^3+a(x-y)=b-c$$

$$(x+y)(x^2-xy+y^2)+a(x+y)=b+c$$ $$(x-y)(x^2+xy+y^2)+a(x-y)=b-c$$

$$(x+y)(1-xy)+a(x+y)=b+c$$ $$(x-y)(1+xy)+a(x-y)=b-c$$

So $$(x+y)(1+a-xy)=b+c$$ $$(x-y)(1+a+xy)=b-c$$ squaring,

$$(1+2xy)((1+a)^2+x^2y^2-2(a+1)xy)=(b+c)^2$$ $$(1-2xy)((1+a)^2+x^2y^2+2(a+1)xy)=(b-c)^2$$

And adding,

$$2(1+a)^2+2x^2y^2-8(a+1)x^2y^2=2(b^2+c^2)$$

So

$$(8a+6)x^2y^2=2(1+a)^2-2(b^2+c^2)$$ $$(4a+3)x^2y^2=(1+a)^2-(b^2+c^2)$$

$$x^2y^2=\frac{(1+a)^2-(b^2+c^2)}{4a+3}$$

On the other hand if we multiply the two equations together we get

$$xy(x^2y^2 +a(1+a))=bc$$

and squaring,

$$x^2y^2(x^2y^2 +a(1+a))^2=b^2c^2$$

and substituting,

$$\frac{(1+a)^2-(b^2+c^2)}{4a+3}\left(\frac{(1+a)^2-(b^2+c^2)} {4a+3} +a(1+a)\right)^2=b^2c^2$$

which becomes,

$$[(1+a)^2-(b^2+c^2)]\left((1+a)(2a+1)^2-(b^2+c^2)\right)^2=b^2c^2(4a+3)^3$$

3

Hint

...one way of doing this would be firstly to subtract the equations and factorise out $(C-S)$ to get $$(C-S)(1+a+CS)=b-c,$$ where $C=\cos\theta$ and $S=\sin\theta$.

Similarly, you could add the two equations and get $$(C+S)(1+a-CS)=b+c$$

Now square both these equations and add them, writing $x=CS$, and, in doing so you will eliminate the $x^3$ term and get an expression for $x^2$ in terms of $a, b$ and $c$, namely, $$x^2=\frac{(1+a)^2-(b^2+c^2)}{3+4a}$$

Call this $\lambda$, for the moment.

You an then extract $C$ or $S$ by solving a quadratic in either, so, for example $$S^2=\frac{1\pm\sqrt{1-4\lambda}}{2}$$

Now substitute this into the original equation to get the equation satisfied by $a,b$ and $c$.

David Quinn
  • 34,121
  • Thanks for your answer. I would like to see it completed. You have essentially found $\sin^2(2\theta)$. But we need to find a relation in $a,b,c$. Now maybe, if you take the solutions to the quadratics, and put them in $\sin^2\theta+\cos^2\theta=1$, you can getvsuch a relation after squaring, but this remains to be seen. And maybe that relation reduces to an identity. – Rene Schipperus Oct 02 '21 at 15:07
  • you can get either $c$ or $s$ and substitute back into one of the original equations to get a relationship between $a, b$ and $c$ – David Quinn Oct 02 '21 at 15:09
  • Rather unfortunate choice of variables $c$. In theory, each original equation is a reduced cubic for $\cos\theta, \sin\theta$, thus solving those and using $\cos^2+\sin^2=1$ does the job too, but the devil is in the details and simplifications. – Macavity Oct 02 '21 at 16:55
  • @Macavity yes you're right, so I have changed to capitals - thanks – David Quinn Oct 02 '21 at 17:12
  • Thanks for helping me with this question, if you multiply the two equations together, one can finish without taking roots. – Rene Schipperus Oct 02 '21 at 19:08