I was playing around with my calculator and I stumbled about the most curious phenomenon. I noticed that $\tan(\pi/2-0.1)$, $\tan(\pi/2-0.01)$, $\tan(\pi/2-0.001)$ etc. seem to approach powers of $10$. Here are some calculations: $$\tan(\pi/2-0.1) \approx 9.96664$$ $$\tan(\pi/2-0.01) \approx 99.9966$$ $$\tan(\pi/2-0.001) \approx 999.999$$ $$\tan(\pi/2-0.0001) \approx 9999.99$$
And after that, my calculator rounds the numbers at powers of $10$. Is the conjecture $\lim_{n \to \infty}{\tan(\pi/2-10^{-n})-10^n} = 0$ true ? I've tried to prove it, but to no avail. I think that maybe expanding $\tan$ as Taylor series might be useful, but I haven't gotten anywhere. Does anyone have any ideas ?
Let's consider what happens to $\tan(\pi/2 - x)$ as $x \to 0$ in general. You can rewrite the tangent as
$$\frac{\sin(\pi/2 -x)}{\cos(\pi/2 - x)}.$$
Now we have the Taylor series approximations
$$\sin(\pi/2 - x) = 1 - \frac 1 2 x^2 + O(x^4)$$
and
$$\cos(\pi/2 - x) = x + O(x^3).$$
Thus, the quotient is well approximated by
$$\frac 1 x - \frac 1 2 x + O(x).$$
Therefore, $\tan(\pi/2 - x) - \frac 1 x \to 0$ as $x \to 0$; nothing special about powers of $10$. We can give explicit error estimates too, using the approximation above. If all you care about is the limit, then knowing $\sin(\pi/2 - x) = 1 + \text{error}$ is enough.