58^circ is not a multiple of 3^circ. Multiples of 1^circ that are not multiples of 3^circ are not constructible with a straightedge and compass, and their trig functions are not the result of some composition of integers using addition, subtraction, multiplication, division and square rooting.
That doesn't mean we can't write down some expression for cos 58^circ. Let's take the degree sign to mean a factor of {2pi}/360.
e^{i 58^circ} = cos 58^circ + i \ sin 58^circ
e^{-i 58^circ} = cos 58^circ - i \ sin 58^circ
e^{i 58^circ} + e^{-i 58^circ} = 2 \ cos 58^circ
cos 58^circ = 1/2 ( e^{i 58^circ} + e^{-i 58^circ} )
Not that helpful.
We can try to write down a polynomial equation one of whose roots is cos 58^circ but it's probably going to be too big to fit.
theta=2^circ is 180th of a circle. Since cos 88^circ = -cos 92^circ that means cos 2^circ satisfies
cos (44 theta) = -cos (46 theta)
cos(180^circ -44 theta) = cos(46 theta)
Let's solve this for theta first. cos x=cos a has roots x=pm a + 360^circ k, integer k.
180^circ -46 theta = \pm 44 theta - 360^circ k
46 theta \pm 44 theta = 180 ^circ + 360^circ k
theta = 2^ circ + 4^circ k or theta = 90^circ + 180^circ k
That's a lot of roots, and we see theta=58^circ among them.
The polynomials T_n(x), called the Chebyshev Polynomials of the first kind, satisfy cos(n theta) = T_n(cos theta). They have integer coefficients. We know the first few from the double and triple angle formulas:
cos(0 theta) = 1 quad quad so quad quad T_0(x)=1
cos(1 theta) = cos theta quad quad so quad quad T_1(x)=x
cos (2 theta) = 2cos^2 theta - 1 quad quad so quad quad T_2(x)=2x^2-1
cos (3 theta) = 4cos^3 theta - 3 cos theta quad quad so quad quad T_3(x)=4x^4-3x
There's a nice recursion relation we can verify:
T_{n+1}(x) = 2x T_{n} (x) - T_{n-1}(x)
So in theory we can generate these for as big n as we care to.
If we let x= cos theta, our equation
cos (44 theta) = -cos (46 theta)
becomes
T_{44}(x) = -T_{46} (x)
Wolfram Alpha is happy to tell us what those are. I'll write the equation just to test the math rendering:
8796093022208 x^44 - 96757023244288 x^42 + 495879744126976 x^40 - 1572301627719680 x^38 + 3454150138396672 x^36 - 5579780992794624 x^34 + 6864598984556544 x^32 - 6573052309536768 x^30 + 4964023879598080 x^28 - 2978414327758848 x^26 + 1423506847825920 x^24 - 541167892561920 x^22 + 162773155184640 x^20 - 38370843033600 x^18 + 6988974981120 x^16 - 963996549120 x^14 + 97905899520 x^12 - 7038986240 x^10 + 338412800 x^8 - 9974272 x^6 + 155848 x^4 - 968 x^2 + 1 = -( 35184372088832 x^46 - 404620279021568 x^44 + 2174833999740928 x^42 - 7257876254949376 x^40 + 16848641306132480 x^38 - 28889255702953984 x^36 + 37917148110127104 x^34 - 38958828003262464 x^32 + 31782201792135168 x^30 - 20758645314682880 x^28 + 10898288790208512 x^26 - 4599927086776320 x^24 + 1555857691115520 x^22 - 418884762992640 x^20 + 88826010009600 x^18 - 14613311324160 x^16 + 1826663915520 x^14 - 168586629120 x^12 + 11038410240 x^10 - 484140800 x^8 + 13034560 x^6 - 186208 x^4 + 1058 x^2 - 1
)
Yes, this answer is getting long, thanks Socratic. Anway, one of the roots of that 46th degree polynomial with integer coefficients is \cos 58^circ.
