eqn = 1472*s^4 - 256*s^2*u^2 + 1392*s^2 - 24*u^2 + 150 == 0
eqn = ![](https://www.mathworks.com/matlabcentral/answers/uploaded_files/1527186/image.png)
sol = solve(eqn, s, 'maxdegree', 4)
sol =
![](https://www.mathworks.com/matlabcentral/answers/uploaded_files/1527191/image.png)
We can see by examination that
and
are both solutions. If
had a non-zero imaginary component then
would have an imaginary component that was the negative of the one for
and therefore the system could not have all imaginary components equal. Likewise for
which has the same situation. Therefore for the imaginary components of the system to all be equal, the solutions must all be real-valued, and the question then becomes under what conditions
and
are both real-valued. eqn2 = [sol(1)^2 > 0, sol(2)^2 > 0]
eqn2 =
![](https://www.mathworks.com/matlabcentral/answers/uploaded_files/1527236/image.png)
solu1 = solve(eqn2(1) + eqn2(2), u, 'returnconditions', true)
solu1 =
u: [2×1 sym]
parameters: x
conditions: [2×1 sym]
solu1.u
ans =
![](https://www.mathworks.com/matlabcentral/answers/uploaded_files/1527241/image.png)
solu1.conditions
ans =
![](https://www.mathworks.com/matlabcentral/answers/uploaded_files/1527246/image.png)
vpa(solu1.conditions)
ans =
![](https://www.mathworks.com/matlabcentral/answers/uploaded_files/1527251/image.png)
solu2 = solve(eqn2(1) - eqn2(2), u, 'maxdegree', 4, 'returnconditions', true)
solu2.conditions
ans =
![](https://www.mathworks.com/matlabcentral/answers/uploaded_files/1527256/image.png)
The z < and z > bounds for solu1 and solu2 are the same -- those are simple bounds.
But is there a z (also known as u) such that 256*z^4 - 2232*z^2 + 4119 is real and negative?
rr = 256*z^4 - 2232*z^2 + 4119
rr = ![](https://www.mathworks.com/matlabcentral/answers/uploaded_files/1527261/image.png)
solz = solve(rr, 'real', true, 'maxdegree', 4)
solz =
![](https://www.mathworks.com/matlabcentral/answers/uploaded_files/1527266/image.png)
vpa(solz)
ans =
![](https://www.mathworks.com/matlabcentral/answers/uploaded_files/1527271/image.png)
If we compare the second of those results, 2.46-ish to the simple bound for u, about 2.33-ish, then we can deduce that there might be
u2 = vpasolve(children(solu2.conditions, 1), solz(2))
u2 = 2.4630284291817806857810551476394
backcheck1 = subs(solu2, solu2.parameters, u2)
simplify(backcheck1.conditions)
backcheck2 = subs(solu2, solu2.parameters, u2-1/100)
simplify(backcheck2.conditions)
backcheck3 = subs(solu2, solu2.parameters, u2+1/100)
simplify(backcheck3.conditions)
backcheck4 = subs(solu2, solu2.parameters, u3)
simplify(backcheck4.conditions)
backcheck5 = subs(solu2, solu2.parameters, u3-1/100)
simplify(backcheck5.conditions)
backcheck6 = subs(solu2, solu2.parameters, u3+1/100)
simplify(backcheck6.conditions)
So the actual bounds are +/- solz(2) -- that outside that range, the imaginary components of eqn should all be 0 and so should all be equal as required.