In this paper, we prove the John-Nirenberg theorem of the bmop martingale spaces for the full range 0<p<∞. We also consider the John-Nirenberg inequality on symmetric spaces of martingales.
Citation: Changzheng Yao, Congbian Ma. An extension of the classical John-Nirenberg inequality of martingales[J]. AIMS Mathematics, 2023, 8(3): 5175-5180. doi: 10.3934/math.2023259
[1] | Tao Ma, Jianzhong Lu, Xia Wu . Martingale transforms in martingale Hardy spaces with variable exponents. AIMS Mathematics, 2024, 9(8): 22041-22056. doi: 10.3934/math.20241071 |
[2] | Meijiao Wang, Qiuhong Shi, Maoning Tang, Qingxin Meng . Stochastic differential equations in infinite dimensional Hilbert space and its optimal control problem with Lévy processes. AIMS Mathematics, 2022, 7(2): 2427-2455. doi: 10.3934/math.2022137 |
[3] | Mingquan Wei, Xiaoyu Liu . Sharp weak bounds for discrete Hardy operator on discrete central Morrey spaces. AIMS Mathematics, 2023, 8(2): 5007-5015. doi: 10.3934/math.2023250 |
[4] | Pengyan Wang, Jiahao Wang . Hardy type identities and inequalities with divergence type operators on smooth metric measure spaces. AIMS Mathematics, 2024, 9(6): 16354-16375. doi: 10.3934/math.2024792 |
[5] | Tohru Nakamura . Asymptotic stability of degenerate stationary solution to a system of viscousconservation laws in half line. AIMS Mathematics, 2018, 3(1): 35-43. doi: 10.3934/Math.2018.1.35 |
[6] | Suriyakamol Thongjob, Kamsing Nonlaopon, Sortiris K. Ntouyas . Some (p, q)-Hardy type inequalities for (p, q)-integrable functions. AIMS Mathematics, 2021, 6(1): 77-89. doi: 10.3934/math.2021006 |
[7] | Muhammad Sheraz, Vasile Preda, Silvia Dedu . Non-extensive minimal entropy martingale measures and semi-Markov regime switching interest rate modeling. AIMS Mathematics, 2020, 5(1): 300-310. doi: 10.3934/math.2020020 |
[8] | Rabha W. Ibrahim, Dumitru Baleanu . Fractional operators on the bounded symmetric domains of the Bergman spaces. AIMS Mathematics, 2024, 9(2): 3810-3835. doi: 10.3934/math.2024188 |
[9] | Mohammed Shehu Shagari, Saima Rashid, Fahd Jarad, Mohamed S. Mohamed . Interpolative contractions and intuitionistic fuzzy set-valued maps with applications. AIMS Mathematics, 2022, 7(6): 10744-10758. doi: 10.3934/math.2022600 |
[10] | Wenjuan Liu, Zhouyu Li . Global weighted regularity for the 3D axisymmetric non-resistive MHD system. AIMS Mathematics, 2024, 9(8): 20905-20918. doi: 10.3934/math.20241017 |
In this paper, we prove the John-Nirenberg theorem of the bmop martingale spaces for the full range 0<p<∞. We also consider the John-Nirenberg inequality on symmetric spaces of martingales.
Let (Ω,F,P) be a probability space and (Fn)n≥0 an increasing sequence of sub-σ-algebras of F with the associated conditional expectations (En)n≥0. A sequence f=(fn)n≥0 adapted to (Fn)n≥0 is said to be a martingale if E(|fn|)<∞ and En(fn+1)=fn for every n≥0. For the sake of simplicity, we assume f0=0. Let 1≤p<∞. The quasi-Banach spaces bmop are defined as follows:
bmop={f=(fn)n≥0:‖f‖bmop=supn‖En(|f−fn|p)‖1p∞<∞}. |
Here, the notation f in |f−fn|p stands for f∞. It follows from [7] that
‖f‖bmop=supnsupa∈Lp(Fn),‖a‖p≤1‖(f−fn)a‖p. |
Before describing our main results, we recall the classical John-Nirenberg inequality in the martingale theory (see [6,7]).
Theorem 1. If the stochastic basis {Fn}n≥0 is regular, then for 1≤p<∞ we have that
bmop=bmo1 |
with equivalent norms.
In 2014, Yi et al. [8] proved the John-Nirenberg inequality on the rearrangement-invariant Banach function space E with 1≤pE≤qE<∞. In 2019, Li [4] considered the John-Nirenberg theorem on Lorentz space bmop,q with 1<p<∞ and 0<q<∞.
In this paper, we first prove the John-Nirenberg inequality of bmop martingale spaces for 0<p<∞, extending Theorem 1 via a new interpolation method. Then, we extend this result to a wider class of the symmetric quasi-Banach function space E with 0<pE≤qE<∞.
Let us first recall some basic facts on the symmetric quasi-Banach function spaces. Let ((0,∞),F,P) be the Lesbegue measure space and L0(0,∞) be the space of all Lesbegue measurable real-valued functions defined on (0,∞). Let E be a quasi-Banach subspace of L0(0,∞), simply called a quasi-Banach function space on (0,∞) in the sequel. A quasi-Banach function space E on (0,∞) is called symmetric if for any g∈E and any measurable function f with μt(f)≤μt(g) (μt(f) and μt(g) respectively represent the non-increasing rearrangement of f and g) for all t≥0, f∈E and ‖f‖E≤‖g‖E. E is said to have the Fatou property if for every net (xi)i∈I in E satisfying 0≤xi↑ and supi∈I‖xi‖E<∞ the supremum x=supi∈Ixi exists in E and ‖xi‖E↑‖x‖E.
The K¨othe dual of a symmetric Banach function space E on (0,∞) is given by
E×={f∈L0(0,∞):∫∞0|f(t)g(t)|dt<∞:∀g∈E}, |
with the norm ‖f‖E×:=sup{∫∞0|f(t)g(t)|dt:‖g‖E≤1}. The space E× is symmetric and has the Fatou property. Refer to [1,5] for more details.
For a quasi-Banach function space E on (0,∞), the lower and upper Boyd indices pE and qE of E are respectively defined by
pE:=lims→∞logslog‖Ds‖ and qE:=lims→0+logslog‖Ds‖, |
where the dilation operator Ds on L0(0,∞) is defined by (Dsf)(t)=f(t/s) for all t∈(0,∞). For a symmetric quasi-Banach function space E on (0,∞), Ds is a bounded linear operator on E for every s>0 and 0≤pE≤qE≤∞ (see [2, Lemma 2.2]).
Given a quasi-Banach function space E on (0,∞), for 0<r<∞, E(r) will denote the quasi-Banach function space on (0,∞) defined by E(r)={x:|x|r∈E} and equipped with the quasi-norm ‖x‖E(r)=‖|x|r‖1rE. Note that
pE(r)=rpE, qE(r)=rqE. | (2.1) |
Let Ei be a quasi-Banach function space on (0,∞) for i=1,2. The pointwise product space E1⊙E2 is defined by
E1⊙E2={f∈L2(0,∞):f=f1f2,fi∈Ei,i=1,2} |
with the functional ‖⋅‖E1⊙E2 being defined by
‖f‖E1⊙E2=inf{‖f‖E1‖f‖E2:f=f1f2,fi∈Ei,i=1,2}. |
We need the following lemmas (see Theorem 2.1 in [1]).
Lemma 1. Let E and F be two symmetric Banach function spaces on (0,∞).
(ⅰ) If 0<p<∞, then (E⊙F)(p)=E(p)⊙F(p).
(ⅱ) L1(0,∞)=E⊙E×.
Lemma 2. Let E be a symmetric quasi-Banach function space on (0,∞) with the Fatou property. If pE>p, then E(1p) can be renormed as a symmetric Banach function space.
Proof. By (2.1), we have that pE(1p)=1ppE>1. Thus E(1p) is an interpolation space for the couple (L1(0,∞),L∞(0,∞)) (see [3, Lemma 3.6]). Therefore, according to Lemma 2.2 in [1], we get that E(1p) can be renormed as a symmetric Banach function space.
Now we define the Hardy spaces and BMO spaces of martingales. For a martingale f=(fn)n≥0, we denote its martingale difference by dfi=fi−fi−1 (with convention f0=0). Then the conditional quadratic variation and the square function are defined by
sn(f)=(n∑i=1Ei−1|dfi|2)1/2, s(f)=(∞∑i=1Ei−1|dfi|2)1/2, |
Sn(f)=(n∑i=1|dfi|2)1/2, S(f)=(∞∑i=1|dfi|2)1/2. |
Let 0<p<∞. Define
Hsp={f=(fn)n≥0:‖f‖Hsp=‖s(f)‖p<∞}, |
HSp={f=(fn)n≥0:‖f‖HSp=‖S(f)‖p<∞}, |
bmop={f=(fn)n≥0:‖f‖bmop=supnsupa∈Lp(Fn),‖a‖p≤1‖(f−fn)a‖p<∞}, |
BMOp={f=(fn)n≥0:‖f‖BMOp=supnsupa∈Lp(Fn),‖a‖p≤1‖(f−fn−1)a‖p<∞}. |
Here, the notation f in |f−fn−1|p stands for f∞.
A stochastic basis (Fn)n≥0 is said to be regular if, for n≥0 and A∈Fn, there exists B∈Fn−1 such that A⊂B and P(B)≤RP(A), where R is a positive constant independent of n. A martingale is said to be regular if it is adapted to a regular σ-algebra sequence. This means that there exists a constant R>0 such that fn≤Rfn−1 for all nonnegative martingales (fn)n≥0 adapted to the stochastic basis (Fn)n≥0. We refer the reader to Long [6] and Weisz [7] for the theory of martingales.
In what follows, unless otherwise specified, for two nonnegative quantities A and B, by A≲B we mean that there exists an absolute constant C>0 such that A≤CB, and by A≈B that A≲B and B≲A.
In this section, we first establish the John-Nirenberg theorem of the bmop spaces for 0<p<1.
Theorem 2. If the stochastic basis (Fn)n≥0 is regular, then, for any f∈bmo1
‖f‖bmop≈‖f‖bmo1 0<p<1. | (3.1) |
Proof. From H¨older's inequality it follows that
‖f‖bmop≤‖f‖bmo1. |
To prove the converse we choose 1<p1<∞ and 0<θ<1 such that 1=(1−θ)/p+θ/p1. Fix n, and for any 0<r<∞, let Tn:Lr(Fn)→Lp(F) be a linear operator with Tn(a)=(f−fn)a. Then by the definition of bmop, we have the following inequalities:
‖Tn‖Lp→Lp=supa∈Lp(Fn),‖a‖p≤1‖(f−fn)a‖p≤‖f‖bmop, |
‖Tn‖Lp1→Lp1=supa∈Lp(Fn),‖a‖p1≤1‖(f−fn)a‖p1≤‖f‖bmop1. |
Thus by interpolation, we have that
‖Tn‖(Lp,Lp1)θ→(Lp,Lp1)θ≤‖f‖1−θbmop‖f‖θbmop1. |
Noting that (Lp,Lp1)θ=L1 with equal norms and using the inequality
‖f‖bmoq≤Cq‖f‖bmo1 for 1≤q<∞, |
(see [7, Corollory 2.51]) we reduce that
‖Tn‖L1→L1≤(Cp1)θ‖f‖1−θbmop‖f‖θbmo1 |
which implies that
‖f‖bmo1≤(Cp1)θ1−θ‖f‖bmop. |
Remark 1. (i) If, in the proof of Theorem 2, we replace f−fn with f−fn−1 and bmop and bmo1 with BMOp and BMO1 then
‖f‖BMOp≈‖f‖BMO1 for 0<p<1. |
(ii) According to Theorem 1, bmop coincides with bmo1 for 1≤p<∞. While for 0<p<1, if a priori we assume that f∈bmo1. Theorem 2 tells us the norms of bmop and bmo1 are also equivalent.
Recall that if (Fn)n≥0 is regular, then Hs1=HS1 which follows that their dual spaces bmo2 and BMO2 are equivalent. Hence, by Theorem 2, Theorem 1, (i) of Remark 1 and [7, Theorem 2.50], we obtain the following result.
corollary 1. Let 0<p<∞. If the stochastic basis (Fn)n≥0 is regular, then for any f∈BMO1 and f∈bmo1
‖f‖bmop≈‖f‖bmo1≈‖f‖BMOp≈‖f‖BMO1. |
Now we present the John-Nirenberg inequality of martingale spaces associated with symmetric quasi-Banach function spaces, generalizing the results obtained in [4,8].
Theorem 3. Let E be a symmetric quasi-Banach function space on (0,∞) with 0<pE≤qE<∞ that has the Fatou property. If (Fn)n≥0 is regular, then for any f∈bmo1,
‖f‖bmoE≈‖f‖bmo1, | (3.2) |
where
bmoE={f=(fn)n≥0:‖f‖bmoE=supnsupa∈E(Fn),‖a‖E≤1‖(f−fn)a‖E<∞}. |
Proof. Choose p and q such that 0<p<pE≤qE<q<∞. Then by Lemma 2, E(1p) can be renormed as a symmetric Banach function space; so, we assume that E(1p) is a symmetric Banach function space. By (ii) of Lemma 1, we have that L1(0,∞)=E(1p)⊙E(1p)×. It follows that
Lp(0,∞)=E⊙F, | (3.3) |
where F=(E(1p)×)p (see (i) of Lemma 1). Fix n. Take a∈Lp(Fn) with ‖a‖p≤1. Then by (3.3), there exist a1∈E and a2∈F such that a=a1a2 and ‖a1‖E,‖a2‖F≤1. Thus we have that
‖(f−fn)a‖p=‖(f−fn)a1a2‖p≤‖a2‖F‖(f−fn)a1‖E≤‖f‖bmoE, |
which implies ‖f‖bmop≤‖f‖bmoE. Therefore, by Theorem 2, ‖f‖bmo1≤‖f‖bmoE.
Now we turn to the converse inequality. Fix n. Similar to the definition of the operator Tn in Theorem 3.1, we can view f−fn as an operator from Lp(Fn) to Lp(F) and from Lq(Fn) to Lq(F); then, we get that
‖f−fn‖Lp→Lp≤‖f‖bmop and ‖f−fn‖Lq→Lq≤‖f‖bmoq. | (3.4) |
By Lemma 3.6 in [3], we have that E is an interpolation space in (Lp(0,∞),Lq(0,∞)) which implies that
‖f−fn‖E→E≤Cmax{‖f−fn‖Lp→Lp,‖f−fn‖Lq→Lq}, | (3.5) |
where C>0 is a constant depending only on p and q. Putting (3.4) and (3.5) together and using Corollary 1, we obtain that
‖f−fn‖E→E≤Cmax{‖f‖bmop,‖f‖bmoq}≤C‖f‖bmo1. |
It follows that ‖f‖bmoE≤C‖f‖bmo1. This completes the proof.
Remark 2. When E=Lp(0,∞) for 0<p<∞, (3.2) implies that
‖f‖bmop≈‖f‖bmo1. |
The second author was supported in part by NSFC No.11801489.
The authors declare that they have no conflicts of interest.
[1] |
T. Bekjan, Z. Chen, M. Raikhan, M. Sun, Interpolation and John-Nirenberg inequality on symmetric spaces of noncommutative martingales, Studia Math., 262 (2021), 241–273. https://doi.org/10.4064/sm200508-11-12 doi: 10.4064/sm200508-11-12
![]() |
[2] | S. Dirksen, Noncommutative Boyd interpolation theorems, T. Am. Math. Soc., 367 (2015), 4079–4110. |
[3] |
S. Dirksen, B. dePagter, D. Potapov, F. Sukochev, Rosenthal inequalities in noncommutative symmetric spaces, J. Funct. Anal., 261 (2011), 2890–2925. https://doi.org/10.1016/j.jfa.2011.07.015 doi: 10.1016/j.jfa.2011.07.015
![]() |
[4] | L. Li, A remark John-Nirenberg inequalities for martingales, Ukrainian Math. J., 770 (2019), 1571–1577. |
[5] | J. Lindenstrauss, L. Tzafriri, Classical banach spaces, Berlin: Springer, 1979. |
[6] | R. Long, Martingale spaces and inequalities, Bei Jing: Peking University Press, 1993. |
[7] | F. Weisz, Martingale Hardy spaces and their applications in fourier analysis, Berlin: Springer, 1994. |
[8] |
R. Yi, L. Wu, Y. Jiao, New John-Nirenberg inequalities for martingales, Statist. Probab. Lett., 86 (2014), 68–73. https://doi.org/10.1016/j.spl.2013.12.010 doi: 10.1016/j.spl.2013.12.010
![]() |