Completeness of $\Lambda^\alpha$ in $D([0,\infty))$

65 Views Asked by At

I was going through this paper in which the weak $L_\alpha$ norm $$\Vert X\Vert_{\alpha,w}^\alpha = \sup_{\lambda>0} \lambda^\alpha \mathbb P\bigg[\sup_{0\leq t\leq T} |X_t|>\lambda\bigg]$$ was defined.

The space $$\Lambda^\alpha :=\bigg\lbrace X \text{ cadlag and adapted } : \Vert X\Vert_{\alpha,w}<\infty\bigg\rbrace $$ was introduced.

Now it was claimed that $\Lambda^\alpha$ is a complete linear metric space for all $0<\alpha<2$.

It's easy to show that this space is indeed linear, but i have some trouble verifying that it is complete. Here is my attempt to far

Let $(X^{(n)})_{n\in\mathbb N}$ be a Cauchy sequence (w.r.t. $\Vert\cdot\Vert_{\alpha,w}$) in $\Lambda^\alpha$, meaning for all $\epsilon>0$ and sufficiently large $M$ we have $$\epsilon> \Vert X^{(n)}-X^{(m)}\Vert_{\alpha,w}^\alpha = \sup_{\lambda>0}\lambda^\alpha \mathbb P\bigg[\sup_{0\leq t\leq T} |X_t^{(n)}-X_t^{(m)}|>\lambda\bigg]\geq \sup_{\lambda>0} \lambda^\alpha \mathbb P[|X_t^{(n)}-X_t^{(m)}|>\lambda]$$ for $n,m>M$ and each $t\in[0,T]$. This implies that the finite-dimensional projections of $X^{(n)}$ converge weakly. Now i would have to verify some tightness criterion of the sequence $X^{(n)}$ to show that it converges weakly in $D([0,T])$ and show that its limit $X$ satisfies $\Vert X\Vert_{\alpha,w}<\infty$.

For the tightness criterion, one might use the characterization using the modulus of continuity $w'$, meaning one has to show that $$\lim_{\delta\to 0}\limsup_{n\to\infty} \mathbb P[w'(X^{(n)},0,T,\delta)>\epsilon]=0.$$ Now by the triangle inequality we get $$\mathbb P[w'(X^{(n)},0,T,\delta)>\epsilon]\leq \mathbb P[w'(X^{(n)}-X^{(m)},0,T,\delta)>\epsilon]+\mathbb P[w'(X^{(m)},0,T,\delta)>\epsilon] \hspace{1cm} (*)$$ which i think is sufficient to show that $\lim_{\delta\to 0}\limsup_{n\to\infty} (*)=0$.

Is this proof correct? Is there an easier way to show the completeness?

1

There are 1 best solutions below

2
On BEST ANSWER

This is essentially correct, but the "I think" at the end can be made more precise. Indeed, first, one has $$ \omega'\left(X^{(n)}-X^{(m)},0,T,\delta)>\varepsilon\right)\leqslant 2\sup_{0\leqslant t\leqslant T}\left\lvert X_t^{(n)}-X_t^{(m)}\right\rvert. $$ For a fixed $\eta$, choose $N$ such that for each $m,n\geqslant N$, $\lVert X^{(n)}-X^{(m)}\rVert^\alpha_{\alpha,w}\leq \eta$. Then for $m,n\geqslant N$, $$ \mathbb P[w'(X^{(n)},0,T,\delta)>\varepsilon]\leqslant\varepsilon^{-\alpha}\eta+\mathbb P[w'(X^{(m)},0,T,\delta)>\varepsilon]. $$ Taking $\limsup_{n\to\infty}$, keeping $m\geqslant N$ fixed, and using tightness of $X^{(m)}$ gives that $$ \lim_{\delta\to 0}\limsup_{n\to\infty}\mathbb P[w'(X^{(n)},0,T,\delta)>\varepsilon]\leqslant\varepsilon^{-\alpha}\eta. $$ Don't forget to show that the limit $X$ is adapted.