Before asking this question, I have scoured the stackexchange for a satisfactory answer, but could not find one. Some answers mentioned using the $\epsilon/3$ trick. I have attempted a proof but could not find a use for this trick. Below is my proof:
Let $f_n(x)$ be a sequence such that $\forall\epsilon>0\exists N>0,\forall m,n>N,|f_m(x)-f_n(x)|<\epsilon$. Prove uniform convergence for this sequence.
Define $f(x)=lim_{n\rightarrow\infty}f_n(x)$. This is well defined as $f_n(x)$ is a cauchy sequence for all x.
For fixed m > N and a given $\epsilon>0$, $\forall n>N, |f_m(x)-f_n(x)|<\epsilon$.
Or $f_m(x)-\epsilon<f_n(x)<f_m(x)+\epsilon$.
$f_m(x)-\epsilon<\lim_{n\rightarrow\infty}f_n(x)<f_m(x)+\epsilon$.
Or $f_m(x)-\epsilon<f(x)<f_m(x)+\epsilon$.
i.e, $|f_m(x)-f(x)|<\epsilon$.
Since m was chosen arbitrarily, we have
$\forall m>N, |f_m(x)-f(x)|<\epsilon$
Which proves uniform convergence.
Is there any mistake in above reasoning. I would be grateful for anyone to point it out.