Abstract
This article deals with the equivariant non parametric robust regression estimation for stationary ergodic processes valued in where
is a semi-metric space. We consider a new robust regression estimator when the scale parameter is unknown. The principal aim is to prove the almost complete convergence (with rate) for the proposed estimator. Unlike in standard multivariate cases, the gap between pointwise and uniform results is not immediate. So, suitable topological considerations are needed, implying changes in the rates of convergence which are quantified by entropy considerations.
Acknowledgments
The authors would like to thank the anonymous reviewers for their valuable comments and suggestions which improved substantially the quality of this article. The authors extend their appreciation to the Deanship of Scientific Research at King Khalid University for funding this work through General Research Project under grant number (G.R.P-125-40).
Appendix
The proof of Lemma is very close to that of Lemma
in Boente and Vahnovan (Citation2015).
According to and
it is clear that if
(A1)
(A1)
In the situation when the combination of
and
allows to get the same result. From now on, in order to simplify the notation, we set
Proof of Theorem 4.2.
We consider the decomposition:
(A2)
(A2)
where
Therefore, Theorem 4.2 is a consequence of the following intermediate results.
Lemma A.1.
Under hypotheses and
-
, we have
Corollary A.2.
Under the hypotheses of Lemma A.1, we have
Lemma A.3.
Under the hypotheses and
, we have
Lemma A.4.
Under the assumptions of Theorem 4.2, we have
Proof of Lemma A.1.
Let be an
net for
and for all
one sets
One considers the following decomposition:
Let us study F1. By using (A1) and the boundness of K, one can write
Let us first consider the case Because K is Lipschitz on
in this case, it comes
with, uniformly on x,
A standard inequality for sums of bounded random variables with allows to get
and it suffices to combine
and
to get
Now, let In this situation K is Lipschitz on
One has to decompose F1 into three terms as follows:
with
One can follow the same steps (i.e., case ) for studying F11 and one gets the same result:
Following same ideas for studying one can write
with
and by using again
and the same inequality for sums of bounded random variables, we get
Similarly, one can state the same rate of convergence for To end the study of
it suffices to put together all the intermediate results and to use
for getting
Now, concerning
we have, for all
Using the fact that is a triangular array of martingale differences according to the σ-fields
to apply the inequality of Lemma 1 in Laïb and Louani (Citation2011). To do that, we must evaluate the quantity
Indeed, under and
we have, for
Hence,
Thus, by using the fact that and by choosing η such that
we have
(A3)
(A3)
Because we obtain that
For
it is clear that
and by following a similar proof to the one used for studying
it comes
Proof of Corollary A.2.
It is easy to see that
such that
We deduce from Lemma A.1 that
Consequently,
Proof of Lemma A.3.
One has
Hence, we get
Thus, with hypotheses
and (A1) we have
the proof of this lemma is then completed.
Proof of Lemma A.4.
This proof follows the same steps as the proof of Lemma A.1. For this, we keep these notations and we use the following decomposition:
Condition and result (A1) allow to write directly, for G1 and
Now, as for one considers
(i.e., K Lipschitz on
) and one gets
with
The main difference with the study of F1 is that one uses here an inequality for unbounded variables. Note that one has
which implies that
Now, allows to get
(A4)
(A4)
If one considers the case (i.e., K Lipschitz on
), one has to split G1 into three terms as for F1 and by using similar arguments, one can state the same rate of almost complete convergence. Similar steps allow to get
(A5)
(A5)
For similarly to the proof of Lemma A.1, we have,
The rest of the proof is based on the application of the exponential inequality of Lemma 1 in Laïb and Louani (Citation2011) on the triangular array of martingale differences.
Indeed, let
For this, we must evaluate the quantity The latter can be evaluated by using the same arguments as were invoked for proving Lemma 5 in Laïb and Louani (Citation2011), allowing us to write, under
Thus, we are now in a position to apply the aforementioned exponential inequality and we get: for all
Thus, by using the fact that and by choosing η such that
we have
As we obtain that
(A6)
(A6)
Now, Lemma A.4 can be easily deduced from Equation(A4)(A4)
(A4) –Equation(A6)
(A6)
(A6) .
Proof of Theorem 4.3.
We use the fact that ψx is monotone w.r.t. the second component. We give the proof for the case of an increasing decreasing case being obtained by the same manner. For this consideration, we write,
Because is increasing, it follows that,
We have
where ξn is between
and θx. Therefore, all we have to prove, is to study the convergence rate of
and to show that
(A7)
(A7)
To do that, we write
where
Therefore, Theorem 4.3 is a consequence of the following intermediate results.
Lemma A.5.
Under Hypotheses and
, we have,
Corollary A.6.
Under Hypotheses of Lemma A.5, we have,
Lemma A.7.
Under Hypotheses and
Lemma A.8.
Under Assumptions and
exists a.s. for all sufficiently large n and there exists
such that
The proof of Lemma A.5 and Lemma A.7 are, respectively, very close to those of Lemma 1 and Lemma 2 in Laïb and Louani (Citation2011). Moreover, the proof of the first part of Lemma A.8 is similar to the Lemma 5 in Azzedine, Laksaci, and Ould-Said (Citation2008) and its second part is a direct consequence of the regularity assumption on
Proof of Corollary A.6.
It is clear that, under there exists
Hence,
It is obvious that the previous Lemma and allows to get
which gives the result.