Skip to main content

Advertisement

Log in

On Doob’s inequality and Burkholder’s inequality in weak spaces

  • Published:
Collectanea Mathematica Aims and scope Submit manuscript

Abstract

Let X be a Banach function space over a nonatomic probability space. We associate with X the weak space \({\mathrm {w}}\text {-}{X}\), which is a quasi-Banach space defined in a natural way. We give necessary and sufficient conditions for Doob’s inequality in \({\mathrm {w}}\text {-}{X}\) to be valid, and necessary and sufficient conditions for Burkholder’s inequality in \({\mathrm {w}}\text {-}{X}\) to be valid.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Similar content being viewed by others

Notes

  1. In [11] the rearrangement-invariance of the space X and the rearrangement-invariance of the norm \(||\,\cdot \, {||}_X\) are distinguished.

References

  1. Antipa, A.: Doob’s inequality for rearrangement-invariant function spaces. Rev. Roum. Math. Pures Appl. 35, 101–108 (1990)

    MathSciNet  MATH  Google Scholar 

  2. Bennett, C., Sharpley, R.: Interpolation of operators. Pure and Applied Mathematics, vol. 129. Academic Press, Boston (1988)

  3. Bergh, J., Löfström, J.: Interpolation spaces. An Introduction. Grundlehren der Mathematischen Wissenschaften, vol. 223. Springer, Berlin (1976)

  4. Burkholder, D.L.: Martingale transforms. Ann. Math. Stat. 37, 1494–1504 (1966)

    Article  MathSciNet  MATH  Google Scholar 

  5. Burkholder, D.L.: Distribution function inequalities for martingales. Ann. Probability 1, 19–42 (1973)

    Article  MathSciNet  MATH  Google Scholar 

  6. Chong, K.M., Rice, M.N.: Equimeasurable rearrangements of functions. Queen’s Papers in Pure and Appl. Math., vol. 28. Queen’s Univ., Kingston (1971)

  7. DeVore, R.A., Lorentz, G.G.: Constructive Approximation. Springer, Berlin (1993)

    Book  MATH  Google Scholar 

  8. Doob, J.L.: Stochastic Processes. Wiley, New York (1953)

    MATH  Google Scholar 

  9. Garling, D.J.H.: Inequalities: A Journey into Linear Analysis. Cambridge University Press, Cambridge (2007)

    Book  MATH  Google Scholar 

  10. Johnson, W.B., Schechtman, G.: Martingale inequalities in rearrangement invariant function spaces. Isr. J. Math. 64, 267–275 (1988)

    Article  MathSciNet  MATH  Google Scholar 

  11. Kikuchi, M.: Characterization of Banach function spaces that preserve the Burkholder square-function inequality. Il. J. Math. 47, 867–882 (2003)

    MathSciNet  MATH  Google Scholar 

  12. Kikuchi, M.: On some mean oscillation inequalities for martingales. Publ. Mat. 50, 167–189 (2006)

    Article  MathSciNet  MATH  Google Scholar 

  13. Kikuchi, M.: Uniform boundedness of conditional expectation operators on a Banach function space. Math. Inequal. Appl. 16, 483–499 (2013)

    MathSciNet  MATH  Google Scholar 

  14. Kikuchi, M.: On some martingale inequalities for mean oscillations in weak spaces. Ric. Mat. doi:10.1007/s11587-015-0224-1

  15. Kreĭn, S.G., Petunin, Yu.I., Semenov, E.M.: Interpolation of linear operators. Translations of Mathematical Monographs, vol. 54. American Mathematical Society, Providence (1982)

  16. Novikov, I. Ya.: Martingale inequalities in symmetric spaces. Sib. Math. J. 34, 99–105 (1993)

  17. Stein, E.M., Weiss, G.: Introduction to Fourier analysis on Euclidean spaces. In: Princeton Mathematical Series, vol. 32. Princeton University Press, Princeton (1971)

  18. Zaanen, A.C.: Integration. Wiley, New York (1967)

    MATH  Google Scholar 

Download references

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Masato Kikuchi.

Additional information

M. Kikuchi was supported by Japan Society for the Promotion of Science (JSPS) KAKENHI Grant Number 25400129.

Appendix

Appendix

As the following example shows, there is a quasi-concave function \(\varphi :[0,1]\rightarrow [0,\infty )\) such that both (3.1) and (3.6) hold for some \(A>1\) but do not hold for all \(A>1\).

Example

Define a function \(\varphi :[0,1]\rightarrow [0,\infty )\) by letting

$$\begin{aligned} \varphi (t)=\left\{ \begin{array}{ll} 2^{n-1}t &{}\quad \text {if }n \in \mathbb {N}\hbox { and }2\cdot 10^{-n}<t\le 10^{-n+1},\\ 5^{-n} &{}\quad \text {if }n \in \mathbb {N}\hbox { and }10^{-n}<t\le 2\cdot 10^{-n},\\ 0 &{}\quad \text {if }t=0. \end{array} \right. \end{aligned}$$

Then \(\varphi \) is quasi-concave. It is easily seen that

$$\begin{aligned} \varliminf _{t\rightarrow 0+}\frac{\varphi (2t)}{\varphi (t)}=1 \quad \text {and} \quad \varlimsup _{t\rightarrow 0+}\frac{\varphi (2t)}{\varphi (t)}=2, \end{aligned}$$

while

$$\begin{aligned} 1<5=\varliminf _{t\rightarrow 0+}\frac{\varphi (10\,t)}{\varphi (t)} \quad \text {and} \quad \varlimsup _{t\rightarrow 0+}\frac{\varphi (10\,t)}{\varphi (t)}=5<10. \end{aligned}$$

On the other hand, we have the following:

Proposition

Suppose that a quasi-concave function \(\varphi :[0,1]\rightarrow [0,\infty )\) is concave. Then:

  1. (i)

    If (3.1) holds for some \(A>1\), then (3.1) holds for all \(A>1\).

  2. (ii)

    If (3.6) holds for some \(A>1\), then (3.6) holds for all \(A>1\).

Proof

  1. (i)

    Suppose that \(A>B>1\). It suffices to show that if (3.1) holds for A, then it holds with A replaced by B (cf. Remark 3.3). Since (3.3) holds for all \(c>1\), it suffices to show that if

    $$\begin{aligned} \varlimsup _{t \rightarrow 0+}\frac{\varphi (Bt)}{\varphi (t)}=B, \end{aligned}$$
    (6.1)

    then

    $$\begin{aligned} \varlimsup _{t \rightarrow 0+}\frac{\varphi (At)}{\varphi (t)}=A. \end{aligned}$$
    (6.2)

    Assume that (6.1) holds. Then there exists a decreasing sequence \(\{t_n\}_{n\in \mathbb {N}}\) of numbers in (0, 1 / B] such that \(t_n \rightarrow 0\) and \(\varphi (Bt_n)/\varphi (t_n)\rightarrow B\) as \(n \rightarrow \infty \). For each \(n\in \mathbb {N}\), let \(\psi _n\) be the affine function whose graph passes through the points \((t_n,\varphi (t_n))\) and \((Bt_n,\varphi (Bt_n))\); that is,

    $$\begin{aligned} \psi _n(t)=\frac{\varphi (Bt_n)-\varphi (t_n)}{(B-1)t_n}\,t +\frac{B\varphi (t_n)-\varphi (Bt_n)}{B-1}. \end{aligned}$$

    Note that since \(\varphi \) is a concave function, we have \(\psi _n(t)\ge \varphi (t)\) for \(t\in (0,t_n]\). In particular, \(\psi _n(t_n/A)\ge \varphi (t_n/A)\), and therefore

    $$\begin{aligned} \sup _{0<t\le t_n/A}\frac{\varphi (At)}{\varphi (t)} \ge \frac{\varphi (t_n)}{\varphi (t_n/A)} \ge \frac{\varphi (t_n)}{\psi _n(t_n/A)} =\frac{A(B-1)\varphi (t_n)}{(1-A)\varphi (Bt_n)+(AB-1)\varphi (t_n)}. \end{aligned}$$

    Since \(\varphi (Bt_n)/\varphi (t_n)\rightarrow B\) as \(n \rightarrow \infty \), the right-hand side converges to A as \(n\rightarrow \infty \). Hence

    $$\begin{aligned} \varlimsup _{t\rightarrow 0+}\frac{\varphi (At)}{\varphi (t)}\ge A. \end{aligned}$$

    Since the reverse inequality always holds, this implies (6.2).

  2. (ii)

    Suppose that \(A>B>1\). It suffices to show that if

    $$\begin{aligned} \varliminf _{t \rightarrow 0+}\frac{\varphi (Bt)}{\varphi (t)}=1, \end{aligned}$$
    (6.3)

    then

    $$\begin{aligned} \varliminf _{t \rightarrow 0+}\frac{\varphi (At)}{\varphi (t)}=1. \end{aligned}$$
    (6.4)

    Assume that (6.3) holds. Then there exists a decreasing sequence \(\{t_n\}_{n\in \mathbb {N}}\) of numbers in (0, 1 / A] such that \(t_n \rightarrow 0\) and \(\varphi (Bt_n)/\varphi (t_n)\rightarrow 1\) as \(n\rightarrow \infty \). Again, let \(\psi _n\) be the affine function whose graph passes through the points \((t_n,\varphi (t_n))\) and \((Bt_n,\varphi (Bt_n))\). Note that \(\varphi (t)\le \psi _n(t)\) for \(t\in [Bt_n,1]\). Then we have \(\varphi (At_n)\le \psi _n(At_n)\), and hence

    $$\begin{aligned} \inf _{0<t\le t_n}\frac{\varphi (At)}{\varphi (t)} \le \frac{\varphi (At_n)}{\varphi (t_n)} \le \frac{\psi _n(At_n)}{\varphi (t_n)} =\frac{A-1}{B-1}\cdot \frac{\varphi (Bt_n)}{\varphi (t_n)}+\frac{B-A}{B-1}. \end{aligned}$$

    Since \(\varphi (Bt_n)/\varphi (t_n)\rightarrow 1\) as \(n\rightarrow \infty \), the right-hand side converges to 1 as \(n \rightarrow \infty \). Thus

    $$\begin{aligned} \varliminf _{t\rightarrow 0+}\frac{\varphi (At)}{\varphi (t)}\le 1, \end{aligned}$$

    which implies (6.4). \(\square \)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Kikuchi, M. On Doob’s inequality and Burkholder’s inequality in weak spaces. Collect. Math. 67, 461–483 (2016). https://doi.org/10.1007/s13348-015-0153-z

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s13348-015-0153-z

Keywords

Mathematics Subject Classification

Navigation