In mathematics, the values of the trigonometric functions can be expressed approximately, as in , or exactly, as in . While trigonometric tables contain many approximate values, the exact values for certain angles can be expressed by a combination of arithmetic operations and square roots. The angles with trigonometric values that are expressible in this way are exactly those that can be constructed with a compass and straight edge, and the values are called constructible numbers.
The trigonometric functions of angles that are multiples of 15°, 18°, or 22.5° have simple algebraic values. These values are listed in the following table for angles from 0° to 90°.[1] In the table below, the label "Undefined" represents a ratio If the codomain of the trigonometric functions is taken to be the real numbers these entries are undefined, whereas if the codomain is taken to be the projectively extended real numbers, these entries take the value (see division by zero).
A trigonometric number is a number that can be expressed as the sine or cosine of a rational multiple of πradians.[2] Since the case of a sine can be omitted from this definition. Therefore any trigonometric number can be written as , where k and n are integers. This number can be thought of as the real part of the complex number. De Moivre's formula shows that numbers of this form are roots of unity:
Since the root of unity is a root of the polynomial xn − 1, it is algebraic. Since the trigonometric number is the average of the root of unity and its complex conjugate, and algebraic numbers are closed under arithmetic operations, every trigonometric number is algebraic.[2] The minimal polynomials of trigonometric numbers can be explicitly enumerated.[3] In contrast, by the Lindemann–Weierstrass theorem, the sine or cosine of any non-zero algebraic number is always transcendental.[4]
The real part of any root of unity is a trigonometric number. By Niven's theorem, the only rational trigonometric numbers are 0, 1, −1, 1/2, and −1/2.[5]
An angle can be constructed with a compass and straightedge if and only if its sine (or equivalently cosine) can be expressed by a combination of arithmetic operations and square roots applied to integers.[6] Additionally, an angle that is a rational multiple of radians is constructible if and only if, when it is expressed as radians, where a and b are relatively prime integers, the prime factorization of the denominator, b, is the product of some power of two and any number of distinct Fermat primes (a Fermat prime is a prime number one greater than a power of two).[7]
Thus, for example, is a constructible angle because 15 is the product of the Fermat primes 3 and 5. Similarly is a constructible angle because 12 is a power of two (4) times a Fermat prime (3). But is not a constructible angle, since is not the product of distinct Fermat primes as it contains 3 as a factor twice, and neither is , since 7 is not a Fermat prime.[8]
It results from the above characterisation that an angle of an integer number of degrees is constructible if and only if this number of degrees is a multiple of 3.
The values of sine and cosine of 30 and 60 degrees are derived by analysis of the equilateral triangle. In an equilateral triangle, the 3 angles are equal and sum to 180°, therefore each corner angle is 60°. Bisecting one corner, the special right triangle with angles 30-60-90 is obtained. By symmetry, the bisected side is half of the side of the equilateral triangle, so one concludes . The Pythagorean and reflection identities then give .
The sines and cosines of all other angles between 0 and 90° that are multiples of 3° can be derived from the angles described above and the sum and difference formulas. Specifically,[10]
For example, since , its cosine can be derived by the cosine difference formula:
If the denominator, b, is multiplied by additional factors of 2, the sine and cosine can be derived with the half-angle formulas. For example, 22.5° (π/8 rad) is half of 45°, so its sine and cosine are:[11]
Repeated application of the half-angle formulas leads to nested radicals, specifically nested square roots of 2 of the form . In general, the sine and cosine of most angles of the form can be expressed using nested square roots of 2 in terms of . Specifically, if one can write an angle as
Since 17 is a Fermat prime, a regular 17-gon is constructible, which means that the sines and cosines of angles such as radians can be expressed in terms of square roots. In particular, in 1796, Carl Friedrich Gauss showed that:[13][14]
The sines and cosines of other constructible angles of the form (for integers ) can be derived from this one.
As discussed in § Constructibility, only certain angles that are rational multiples of radians have trigonometric values that can be expressed with square roots. The angle 1°, being radians, has a repeated factor of 3 in the denominator and therefore cannot be expressed using only square roots. A related question is whether it can be expressed using cube roots. The following two approaches can be used, but both result in an expression that involves the cube root of a complex number.
Using the triple-angle identity, we can identify as a root of a cubic polynomial: . The three roots of this polynomial are , , and . Since is constructible, an expression for it could be plugged into Cardano's formula to yield an expression for . However, since all three roots of the cubic are real, this is an instance of casus irreducibilis, and the expression would require taking the cube root of a complex number.[15][16]
^Lehmer, D. H. (1933). "A Note on Trigonometric Algebraic Numbers". The American Mathematical Monthly. 40 (3): 165–166. doi:10.2307/2301023. JSTOR2301023.
^Durbha, Subramanyam (2012). "A Geometric Method of Finding the Trigonometric Ratios of 22 ½° and 75°". Mathematics in School. 41 (3): 22–23. JSTOR23269221.
Watkins, William; Zeitlin, Joel (1993). "The minimal polynomial of cos(2*π/n)". American Mathematical Monthly. 100 (5): 471–474. doi:10.2307/2324301. JSTOR2324301.
Bracken, Paul; Cizek, Jiri (2002). "evaluation of quanum mechanical perturbative sums in terms of quadratic surds and their use in the approximation of ζ(3)/π3". International Journal of Quantum Chemistry. 90: 42–53. doi:10.1002/qua.1803.
Servi, L. D. (2003). "Nested square roots of 2". American Mathematical Monthly. 110 (4): 326–330. doi:10.2307/3647881. JSTOR3647881.