Return to Revisions

The answer is: of course not. Indeed, let $a=1$, $b=0$, and take any real $m>0$. If $f$ were analytic in $x,y$ such that $-m\le\Re x,\Re y,\Im x,\Im y\le m$ then it would be analytic in $x$ at $x=0$ for $y=0$. Then the function $$\mathbb R^2\ni(s,t)\mapsto g(s,t):=\frac{Q(s)}2\,\ln\frac{Q(s)}2$$ would satisfy the Cauchy–Riemann equations at $(s,t)=(0,0)$ -- which it does not, because the partial derivative of $\Re g(s,t)$ in $s$ at $(s,t)=(0,0)$ is $\dfrac{\ln(4/e)}{2\sqrt{2\pi}}\ne0$, while the partial derivative of $\Im g(s,t)[=0]$ in $t$ is $0$ everywhere.