Consider this function ( I got this when I was solving an another problem)
f(n) = $$\left(\frac{1}{2}\right)\left[\left(\frac{n^2+\sqrt{n^4+16008n}}{2n}\right)+\sqrt{\left(\left(\frac{n^2+\sqrt{n^4+16008n}}{2n}\right)^2\right)-\frac{16008}{\frac{n^2+\sqrt{n^4+16008n}}{2n}}}\right]-n$$
When I searched "f(n)=0" on Wolfram alpha, it shows result: $$ n \geq \sqrt[3]{2001} $$ (That means f(n) has infinitely many zeroes ?)
Since,$$13\geq \sqrt[3]{2001}$$ f(13) should be zero , same for f(14), f(15) and so on... But when I evaluated f(13) on Wolfram alpha the result was $$5.32907\times10^{-15}$$
I also found that f(n)=0 whenever I plugged $$n \geq 60 $$ Can you help me with these questions.
If f(n) really has infinitely many zeroes ?
Is f(n) an algebraic function ?
Why Wolfram alpha behaves in this way ?
Bizarrely, this function is indeed identically equal to $0$ for all real numbers $n\ge\sqrt[3]{2001}$. It can be simplified to $$ \frac14\biggl( -3n + \sqrt{n^2+\frac{16008}n} + \sqrt{10n^2+\frac{16008}n-6n\sqrt{n^2+\frac{16008}n}} \biggr), $$ which further simplifies to $$ \frac14\biggl( -3n + \sqrt{n^2+\frac{16008}n} + \sqrt{\Bigl(3n-\sqrt{n^2+\frac{16008}n} \Bigr)^2} \biggr) $$ and therefore to $$ \frac14\biggl( -3n + \sqrt{n^2+\frac{16008}n} + \biggl| 3n-\sqrt{n^2+\frac{16008}n} \, \biggr| \biggr), $$ which finally equals $$ \begin{cases} \frac12\biggl( \sqrt{n^2+\frac{16008}n} -3n \biggr), & \text{if } \sqrt{n^2+\frac{16008}n} \ge 3n, \\ 0, & \text{if } \sqrt{n^2+\frac{16008}n} \le 3n. \end{cases} $$ This last inequality is the source of the changed behavior at $n=\sqrt[3]{2001}$.
Getting an answer of $5\times10^{-15}$ when plugging in $n=13$ is just a matter of machine-precision-level errors adding up to something that the computer thinks is nonzero.