Show that if Frechet (d) derivative is exist this implies that Gateaux (D) derivative is exist , and D f(x)= d f(x), and show that the opposite is not true always and why?
my solution
suppose that the f-derivative exist
$\lim_{\| k \| \to 0} \frac{\| T(x+k)-T(x)-dT(x)k)\| }{\| k\|}=0 \quad Put \quad k=\alpha h \rightarrow $ $\lim_{\| \alpha \| \to 0} \frac{ \| T(x+\alpha h-T(x)-dT(x)\alpha h\|}{\|\alpha h \|}=0 $
By tailor series:
$ T(x+\alpha h)=T(x)+\alpha dT(x)h+h O(\alpha) \leftrightarrow T(x+\alpha h)-T(x)=\alpha dT(x)h+h O(\alpha)$
$\lim_{\alpha \to 0} \frac{T(x+\alpha h)-T(x)-dT(x)\alpha h}{\alpha} = 0$
$\lim_{\alpha \to 0} \frac{T(x+\alpha h)-T(x)}{\alpha} -dT(x)h = 0$
$\lim_{\alpha \to 0} \frac{T(x+\alpha h)-T(x)}{\alpha} =dT(x)h$
and i can't do second term of question "and show that the opposite is not true always and why?" .
Consider the function $f\colon \mathbf R^2 \rightarrow \mathbf R$ defined by $f(0,0)=0,\enspace f(x,y) = \dfrac{xy^3}{x^2+y^6}$. It is Gâteaux-derivable at $(0,0)$, but not Fréchet-derivable. Indeed, if it were Fréchet-derivable, it should be continuous at $(0,0)$. Hint: let $(x,y) \to (0,0)$ along the curve $x=y^3$.