Showing $\sqrt{\frac{e}{2}}\cdot\frac{e}{\pi}\left(\frac{e}{2}-\frac{1}{e}\right)<1$ without a calculator 
Show that:
$$\sqrt{\frac{e}{2}}\cdot\frac{e}{\pi}\left(\frac{e}{2}-\frac{1}{e}\right)<1$$

I have tried power series of exponential around $0$ wich is :
$$e^x=1+x+\frac{x^2}{2}+O(x^3)$$
We can pursue it with another order .
Edit :
An inequality due to Nanjundiah states for $n\geq 1$ a natural number:
$$\left(1+\frac{1}{n}\right)^{\left(n+\frac{1}{2}\right)}>e$$
Edit 2 :
An inequality due to Bennett states for $x$ a real number $m,n$ natural numbers and $m,n>x$ then :
$$\left(1+\frac{x}{m}\right)^{m}\left(1-\frac{x}{n}\right)^{n}<1$$
There is a little mistake in the statement of the edit 2 we need $x\neq 0$
For $\pi$ I have tried the continued fraction see wikipedia
Edit 3 :
I transform the product into a sum using :
$$\ln(x)\leq x-1$$
For $x>0$ so we need to show :
$$\sqrt{\frac{e}{2}}\cdot\frac{e}{\pi}-1+\left(\frac{e}{2}-\frac{1}{e}\right)-1<0$$
Wich is true numerically but I have not a proof yet of this transformation .
A interesting function is :
$$f(x)=\left(\sqrt{\frac{e}{2}}\cdot\frac{e}{\pi}\right)-\frac{1}{x}+\left(\frac{e}{2}-\frac{x}{e^{x}}\right)-x$$
Numerically we have $f(x)<0$ for $x>0$  and $f'(1)=0$

Question: How to show it by hand without a calculator?

 A: Hint:
By excess rational approximations of $e$ can be found from the usual Taylor expansion.
By default rational approximations of $\pi$ can be found from the Machin formula, among others.
Five exact decimals should be enough.
Now the following inequality can be established by hand (though this will take several hours):
$$\frac{\overline{e}(\overline{e}^2-2)^2}{\underline{\pi}^2}<8.$$
(With $\overline e=2.71829$ and $\underline\pi=3.14159$, you get $7.99888\dots<8$.)
Not glamorous, but effective.
A: Using values rounded up to the fourth decimal,
$$e(e^2-2)<2.7183(2.7183^2-2)^2<2.7183\cdot5.3891^2<2.7183\cdot29.0424<78.9460$$
and down
$$78.9520=8\cdot9.8690<8\cdot3.1415^2<8\pi^2.$$
This takes four multiplies of $4$-decimal numbers and a little more.

(If using precomputed tables is allowed, this one supplies $\sqrt2,\pi,\pi^2,\sqrt e,e,e^2$: http://www.ebyte.it/library/educards/constants/MathConstants.html)
A: The simplification of the givens problem product is
$\frac{\sqrt{\frac{e}{2}}(e^2-2)}{2\pi}$
This is much easier to be estimated against one.
It goes even more nicely:
$\frac{\sqrt{\frac{e}{2}}(e-\sqrt{2})(e+\sqrt{2})}{2\pi}$
As one can see immediately from this representation is that the factor $e+\sqrt{2}$ is the biggest. Since $e \approx 2.7$ and $\sqrt{2} \approx 1.4 $.
So the biggest factor of the nominator is about $4.1325$ and the other two are smaller and than $1.5$ for sure. That is indeed smaller than $2\pi$ that is somewhat bigger than $6$.
So this is multiplying $1.166$ with $1.30$ and $4.13$ and that is smaller than $2\pi$.
So this is a problem connection two real numbers well known and famous with
$2\pi$. As well as famous and irrational.
We already used the calculator as well.
To make the result with the calculator more irresistible. The true values are
$6.28268 $ and $6.28319$
This is very little compared to the rounding of the three single factors.
This shows too representation matters! How to get something representing $e$ and $\sqrt{2}$ for simplification?
I hope You are able to follow so far.
A nice repository to expressing $e$ in term of $2$ or $\sqrt{2}$ are well known identities and inequality from the history of Mathematics:
List_of_representations_of_e.
It is now only a matter of how close You intend to approximate the two numbers in comparison to $2\pi$. A shown by the use of the calculator in my representation of the given inequality only necessary up to the third digit after the decimal colon. That is not little, that is not much.
So take for example the infinite product representations. Perhaps that from Pippengers. Or approximate with the limited sequences for $e$, perhaps Sterling's formulas. It is a matter of taste and accuracy and ethics to the numerical Mathematician what he prefers best.
Whatever you select there is always the last step in the accuracy of ease of the equation that will make use of the calculator. Numbers never lie but the paths to get good or even best numerics representations are up to the Mathematician himself.
$\frac{\sqrt{2 e}(\frac{e}{\sqrt{2}}-1)(\frac{e}{\sqrt{2}}+1)}{2\pi}$
and
Pippengers product:
$e=2\left({\frac  {2}{1}}\right)^{{1/2}}\left({\frac  {2}{3}}{\frac  {4}{3}}\right)^{{1/4}}\left({\frac  {4}{5}}{\frac  {6}{5}}{\frac  {6}{7}}{\frac  {8}{7}}\right)^{{1/8}}\cdots $
Put this into the given nominator and yeah calculator or round in the exactness of the formulare. These are just infinite products. Look at Infinite product.
You might make use of Wallis product to approximate $2\pi$.
Some are fine with multiplicating some prefer to sum the bridge is taking logarithms.
This inequality in numbers is
$0.99992<1$.
That is evidently true.
