Cosine Double Angle Formula

Working with Double Angle Identities in Practice

I ran into this problem last year when I was doing harmonic analysis on a signal processing pipeline. The fundamental frequency kept producing messy integrals, and I needed a cleaner way to express everything in terms of the base angle. That's where the standard double angle approach came in handy. There are actually three equivalent forms, and knowing which one to reach for saves a lot of unnecessary algebra: cos(2) = cos²() sin²()

This is the most direct form. It comes straight from the angle addition formula for cosine: cos(A+B) = cos(A)cos(B) sin(A)sin(B). Plug in A = B = and you get it immediately. Use this when you have both sine and cosine terms in your expression and want to merge them. cos(2) = 2cos²() 1 Substitute sin²() = 1 cos²() into the first form. You get this. Use it when your problem involves only cosine — it eliminates the sine term entirely and reduces everything to one function. This was the form I used in my signal processing work because the Fourier coefficients were already expressed as cosines.

cos(2) = 1 2sin²() Same substitution trick, just the other direction. Use it when you're working with sine-heavy expressions. Here's a concrete example. Say you need to evaluate cos(/4) without a calculator. You know cos(/8) is a known value from half-angle tables, but what if it's the reverse? If you know cos(/4) = 2/2, you can solve for cos(/8):

cos(/4) = 2cos²(/8) 1
2/2 = 2cos²(/8) 1
cos²(/8) = (1 + 2/2)/2
cos(/8) = (2+2)/2 That last line gives you the exact value. No approximation needed. This is the kind of thing that shows up in physics homework, but also in real numerical libraries where you need exact representations for test cases. A common mistake people make is confusing cos(2) with 2cos(). They're completely different. cos(2) is a function of the doubled angle; 2cos() just scales the output. Beginners sometimes write cos(2x) = 2cos(x) and then wonder why their answers don't match the textbook. Check your notation first.

Another subtlety: when you're solving equations like cos(2) = 1/2, you get = ±/6 + n. But if you rewrite it as 2cos²() 1 = 1/2 and solve for cos²(), you might miss the negative cosine solutions unless you remember to take both roots. I learned this the hard way during an exam when I only found half the solutions and lost points. The main limitation of these formulas is that they don't simplify everything. If you have something like cos²() + sin³(), the double angle only helps with the squared part. You're still stuck with the cubic sine term. In those cases, you either use numerical methods or look for a different identity. Sometimes power-reduction formulas combined with double angle identities can help, but not always. If your goal is purely computational — say you're writing a library function and need to evaluate cos(2) for many values — there's an interesting tradeoff. Using 2cos²() 1 involves a squaring operation, which is slightly more expensive than a single cosine evaluation on some architectures. On others, the multiplication is faster than a transcendental function call. It depends on the hardware. For most applications, it doesn't matter, but in high-performance signal processing code, people sometimes precompute tables or use CORDIC algorithms instead of calling the identity at all.

Bottom line: learn all three forms, know when to use each one, and check whether you're accidentally treating cos(2) like 2cos(). Those two habits alone will save you more time than memorizing fifteen other identities.