Showing the maximum value using alternate approach If $x_1,x_2...x_n$ are postive numbers satisfying $x_1\cdot x_2 \cdots x_n = 1$ , then find the maximum value of $\frac{1}{\sqrt{x_1 ^2 + 1}\,\cdot\, \sqrt{x_2^2 +1}\,\cdots\,\sqrt{x_n^2+1} }$   .

What i considered was applying the RMS -GM inequality which is $\sqrt{a^2 +b^2} \geq 2\sqrt{ab}$ which gives the maximum of the expression but is there an another way which can be used to systematically show and get the maximum ?


*

*Note : originally problem was given in terms of $\cot z_1 \cdots \cot z_n = 1$ and all angles lie in $(0,π/2)$
and we need to find maximum of $\cos z_1\cdots\cos z_n$ but for better understanding i converted into this form

 A: For the rigorous proof it is convenient to use the method of Lagrange multipliers https://en.wikipedia.org/wiki/Lagrange_multiplier#:~:text=In%20mathematical%20optimization%2C%20the%20method,chosen%20values%20of%20the%20variables).
We can define $\displaystyle f(x_1, ..., x_n)=\ln \frac{1}{\sqrt{x_1 ^2 + 1}\,\cdot\, \sqrt{x_2^2 +1}\,\cdots\,\sqrt{x_n^2+1} }=-\frac{1}{2}\Big(\ln(x_1^2+1)+...+\ln(x_n^2+1)\Big)$.
As logarithm is a growing function, it is enough to find the maximum of $f(x)$. Given that $\ln(x_1...x_n)=0$, we also introduce the additional condition via $g(x_1, ..., x_n)=\ln(x_1...x_n)=\ln x_1+... \ln x_n$, and consider the Lagrangian function
$$L(x_1, ... x_n)=f(x_1, ..., x_n)+\lambda g(x_1, ..., x_n)$$
We we are looking for an unconditional extremum of $L(x_1, ... x_n)$:
$$\frac{\partial}{\partial x_i}L(x_1, ... x_n)=0\,\,\Rightarrow\,\,\frac{\lambda}{x_i}-\frac{x_i}{x_i^2+1}=0\,\,\Rightarrow\,\,x_i^2=\frac{1-\lambda}{\lambda}$$
From the condition $\displaystyle g(x_1, ..., x_n)=\ln x_1+... \ln x_n=0$ we get
$$2n\ln\frac{1-\lambda}{\lambda}=0\,\,\Rightarrow\,\,\lambda=\frac{1}{2}\,\,\Rightarrow\,\, \,x_1=x_2=...=x_n=1$$
We have a single conditional extremum; taking, for example, $x_1=...x_{n-2}=1,\,x_{n-1}=\frac{1}{x_n}$ and leading $x_n\to 0$ (and $x_{n-1}\to\infty$ correspondingly), we can make $f(x)$ as small as we wish; therefore, this extremum is a maximum.
