Bielecki norm method (source code)

= Bielecki norm method
{c}

The <Bielecki norm method> changes the <norm> used for an integral <contraction mapping>, allowing a larger time interval without changing the underlying continuous functions. On $C([a,b],X)$ for a <Banach space> $X$, use
$$
\|u\|_\beta=\sup_{a\leq t\leq b}e^{-\beta(t-a)}\|u(t)\|,\qquad\beta>0.
$$
For $Tu(t)=u_0+\int_a^t f(s,u(s))\,ds$ with a uniformly $L$-<Lipschitz continuous> right-hand side,
$$
e^{-\beta(t-a)}\|Tu(t)-Tv(t)\|\leq\frac L\beta(1-e^{-\beta(t-a)})\|u-v\|_\beta.
$$
Thus $\beta>L$ makes $T$ a <contraction mapping>. On a fixed finite interval this weighted <supremum norm> is equivalent to the usual one. The <Banach fixed-point theorem> applies when $T$ maps the chosen complete function space into itself.

For integration from $1$ to $t\in[1,b]$, the weight $t^{-c}$ gives the alternative bound $Lb/(c+1)$ for $c\geq0$. Both versions reduce the integral operator bound by choosing a weight, rather than by claiming that a local existence theorem automatically gives a global solution for an arbitrary nonlinear vector field.