Journal of Inequalities and Applications

J Inequal Appl. 2017; 2017(1): 312.
Published online 2017 December 19.
PMCID: PMC5736795

# Commutators associated with Schrödinger operators on the nilpotent Lie group

## Abstract

Assume that G is a nilpotent Lie group. Denote by L = −Δ + W the Schrödinger operator on G, where Δ is the sub-Laplacian, the nonnegative potential W belongs to the reverse Hölder class Bq1 for some $q1≥D2$ and D is the dimension at infinity of G. Let ℛ = ∇(−Δ+W)−½ be the Riesz transform associated with L. In this paper we obtain some estimates for the commutator [h, ℛ] for $h∈Lipνθ$, where $Lipνθ$ is a function space which is larger than the classical Lipschitz space.

Keywords: commutator, Lipschitz space, nilpotent Lie groups, reverse Hölder inequality, Riesz transform, Schrödinger operator

## Introduction

Assume G to be a connected and simply connected nilpotent Lie group and 𝔤 to be its Lie algebra identified with the space of left invariant vector fields. Given X = {X1, …, Xl} ⊆ 𝔤, a Hörmander system of left invariant vector fields on G. Let $Δ=∑i=1lXi2$ be the sub-Laplacian on G associated with X and the gradient operator be denoted by ∇ = (X1, …, Xl). Following [1], one can define a left invariant metric d associated with X which is called the Carnot-Carathéodory metric: let xy ∈ G, and

d(xy) = inf{δ∣∃γ:[0, δ] → Gγ(0) = xγ(δ) = y},

where γ is a piecewise smooth curve satisfying

$γ′(s)=∑i=1lai(s)Xi(γ(s)),∀s∈[0,δ],∑i=1l|ai(s)|2≤1.$

If x ∈ G and r > 0, we will denote by B(xr) = {y ∈ Gd(xy) < r} the metric balls. Assume dx to be the Haar measure on G. Then, for every measurable set E ⊆ G, |E| denotes the measure of E. Suppose e to be the unit element of G. Note that V(r) = |B(er)| = |B(xr)| for any x ∈ G and r > 0. It follows from in [2] or [3] that there exists a constant C1 > 0 such that

$C1−1rd≤V(r)≤C1rd,∀0≤r≤1;$
1

$C1−1rD≤V(r)≤C1rD,∀1≤r<∞,$
2

where d and D denote the local dimension and the dimension at infinity of G, and there is D ≥ d > 0. At this time, the Lie group G is also called a Lie group of polynomial growth. If G is a stratified Lie group, then Dd (cf. [1]). Also, there exist positive constants C2C3 > 1 such that

$C2−1(Rr)d≤V(R)V(r)≤C2(Rr)D,∀0
3

V(2r) ≤ C3V(r).
4

Throughout this paper, we always assume that d ≥ 2 .

Let L = −Δ + W be the Schrödinger operator, where Δ is the sub-Laplacian on G and the nonnegative potential W belongs to the reverse Hölder class Bq1 for some $q1≥D2$ and D > 3. The Riesz transform associated with the Schrödinger operator L is defined by

ℛ = ∇(−Δ+W)−½.
5

Let b be a locally integrable function on G and T be a linear operator. For a suitable function f, the commutator is defined by [bT]fbT(f)−T(bf). Many researchers have paid attention to the commutator on n. It is well known that Coifman, Rochberg and Weiss [4] proved that [bT] is a bounded operator on Lp for 1 < p < ∞ if and only if b ∈ BMO(ℝn), when T is a Calderón-Zygmund operator. Janson [5] proved that the commutator is bounded from Lp(ℝn) into Lq(ℝn) if and only if b ∈ Lipν(ℝn) with $ν=(1p−1q)n$, where Lipν(ℝn) is the Lipschitz space. Sheng and Liu [6] proved the boundedness of the commutator [b, ℛ] from the Hardy space $HLp(Rn)$ into Lq(ℝn) when b belongs to a larger Lipschitz space. Comparatively, there has been much less research on the commutator on nilpotent Lie groups. The goal of this paper will be to obtain some estimates for the commutator related to the Schrödinger operator on nilpotent Lie groups. The complicated structure of nilpotent Lie group will bring some essential difficulties to our estimates in the following sections.

Note that a non-negative locally Lq integrable function W on G is said to belong to Bq (1 < q < ∞) if there exists C > 0 such that the reverse Hölder inequality

$(1|B|∫BW(y)qdy)1q≤C(1|B|∫BW(y)dy)$
6

holds for every ball B in G.

We first introduce an auxiliary function as follows.

### Definition 1

Let W ∈ Bq for some $q≥D2$. For x ∈ G, the function m(xW) is defined by

$1m(x,W)=ρ(x):=supr>0{r:r2V(r)∫B(x,r)W(y)dy≤1}.$

Now we define the space $Lipνθ(G)$ on the nilpotent Lie group.

### Definition 2

Let θ > 0 and 0 < ν < 1, the space $Lipνθ(G)$ consists of the functions f satisfying

$|f(x)−f(y)|≤Cdν(x,y)(1+d(x,y)ρ(x)+d(x,y)ρ(y))θ$

holds true for all xy ∈ G, x ≠ y. The norm on $Lipνθ(G)$ is defined as follows:

$∥f∥Lipνθ(G)≜supx,y∈G,x≠y(|f(x)−f(y)|dν(x,y)(1+d(x,y)ρ(x)+d(x,y)ρ(y))θ)<∞.$

It is easy to see that this space is exactly the Lipschitz space when θ = 0 if G is a stratified Lie group (cf. [7] and [8]).

We also introduce the following maximal functions.

### Definition 3

Let $f∈Lloc1(G)$. For 0 < γ < D, the fractional maximal operator is defined by

$Mγf(x)=supx∈Brγ|B|∫B|f(y)|dy,x∈G,$

where the supremum on the right-hand side is taken over all balls B ⊆ G and r is the radius of the ball B.

### Definition 4

Given α > 0, the maximal functions for $f∈Lloc1(G)$ and x ∈ G are defined by

$Mρ,αf(x)=supx∈B∈B1|B|∫B|f(y)|dy$

and

$Mρ,α♯f(x)=supx∈B∈B1|B|∫B|f(y)−fB|dy,$

where ρ,α = {B(yr):y ∈ G and r ≤ αρ(y)}.

We are in a position to give the main results in this paper.

### Theorem 1

Assume W ∈ Bq1 for some $q1≥D2$, where D denotes the dimension at infinity of the nilpotent Lie group G. Let

7

Denote the adjoint operator of by $R˜=(−Δ+W)−12∇$. Then, for any $h∈Lipνθ(G)$, 0 < ν < 1, the commutator $[h,R˜]$ is bounded from Lq(G) into Lp(G), $1p=1q−νD$, if q2 < p < ∞.

We immediately deduce Corollary 1 by duality.

### Corollary 1

Assume W ∈ Bq1 for some $q1≥D2$, where D denotes the dimension at infinity of the nilpotent Lie group G. Let

Then, for any $h∈Lipνθ(G)$, 0 < ν < 1, the commutator [h, ℛ] is bounded from Lp(G) into Lq(G), $1p=1q−νD$, if $1.

Throughout this paper, unless otherwise indicated, C will be used to denote a positive constant that is not necessarily the same case at each occurrence and it depends at most on the constants in (3) and (6). We always denote $δ=2−Dq1$. By A ∼ B, we mean that there exist constants C > 0 and c > 0 such that $c≤AB≤C$.

## Estimates for the kernels of ℛ and $R˜$

In this section we recall some estimates for the kernels of Riesz transform and the dual Riesz transform $R˜$, which have been proved in [3].

### Lemma 1

W ∈ Bq is a doubling measure, that is, there exists a constant C > 0 such that

B(x,2r)W(y) dy ≤ CB(x,r)W(y) dy.

### Lemma 2

There exists C > 0 such that, for 0 < r < R < ∞,

$r2V(r)∫B(x,r)W(y)dy≤C(Rr)Dq−2r2V(R)∫B(x,R)W(y)dy.$

### Lemma 3

If rρ(x), then

$r2V(r)∫B(x,r)W(y)dy=1.$

Moreover,

$r2V(r)∫B(x,r)W(y)dy∼1if and only ifr∼1m(x,W).$

### Lemma 4

There exist constants Cl0 > 0 such that

$1C(1+d(x,y)ρ(x))−l0l0+1≤ρ(x)ρ(y)≤C(1+d(x,y)ρ(x))l0.$

In particular, ρ(x) ∼ ρ(y) if d(xy) ≤ Cρ(x).

### Lemma 5

There exist constants C > 0 and l1 > 0 such that

$∫B(x,R)d2(x,y)W(y)V(d(x,y))dy≤CR2V(R)∫B(x,R)W(y)dy≤C(1+Rρ(x))l1.$

Using Lemma 4, we immediately have the following lemma.

### Lemma 6

There exist l0 > 0, C > 0 such that, for any x and y in G,

$1+d(x,y)ρ(x)+d(x,y)ρ(y)≤C(1+d(x,y)ρ(x))l0+1.$

Let Γ(xyλ) denote the fundamental solution for the operator −Δ + Wλ, namely, [−Δ + Wλ]Γ(xyλ) = δ(y−1x), where δ is the Dirac function and λ ∈ [0, ∞). Markedly, Γ(xyλ) = Γ(yxλ).

### Lemma 7

Let N be a positive integer.

• (i)
Suppose W ∈ Bq1 for some $q1≥D2$. Then there exists a constant CN > 0 such that, for x ≠ y,
$|Γ(x,y,λ)|≤CN(1+d(x,y)λ12)N(1+d(x,y)/ρ(x))Nd(x,y)V(d(x,y)).$
• (ii)
Suppose W ∈ Bq1 for some $q1≥D2$. Then there exists a constant CN > 0 such that
$|∇xΓ(x,y,λ)|≤CN(1+d(x,y)λ12)N(1+d(x,y)/ρ(x))N×(d2(x,y)V(d(x,y))∫B(y,d(x,y))W(z)d(z,y)dzV(d(z,y))+d(x,y)V(d(x,y))).$
Particularly, if W ∈ Bq1 for some q1 ≥ D, then there exists CN > 0 such that, for x ≠ y,
$|∇xΓ(x,y,λ)|≤CN(1+d(x,y)λ12)N(1+d(x,y)/ρ(x))Nd2(x,y)V(d(x,y)).$

By the functional calculus, we may write

$(−Δ+W)−12=1π∫0∞λ−12(−Δ+W+λ)−1dλ.$
8

Let $f∈C0∞(G)$. From (−Δ+W+λ)−1f(x) = ∫GΓ(xyλ)f(y) dy, it follows that

f(x) = ∫G𝒦(xy)f(y) dy
9

where

$K(x,y)=1π∫0∞λ−12∇xΓ(x,y,λ)dλ.$
10

Similarly, the adjoint operator of is defined to be

$R˜f(x)=∫GK˜(x,y)f(y)dy,$
11

where

$K˜(x,y)=1π∫0∞λ−12∇yΓ(y,x,λ)dλ.$
12

We recall estimates of the kernels for and $R˜$ (cf. [3]).

### Lemma 8

Suppose W ∈ Bq1 for some $D2≤q1. For any integer N > 0, there exists CN > 0 such that

$|K(x,y)|≤CN(1+d(x,y)/ρ(x))N(d(x,y)V(d(x,y))∫B(y,d(x,y))W(z)d(z,y)dzV(d(z,y))+1V(d(x,y)))$
13

and

$|K(x,y′)−K(x,y)|≤CNdδ(y,y′)(1+d(x,y)/ρ(x))Nd1−δ(x,y)V(d(x,y))(∫B(y,d(x,y))W(z)d(z,y)dzV(d(z,y))+1d(x,y))$
14

for some δ > 0 and $0. If W ∈ Bq1 for some q1 ≥ D, then

$|K(x,y)|≤CN(1+d(x,y)/ρ(x))N1V(d(x,y))$
15

and

$|K(x,y′)−K(x,y)|≤CN(1+d(x,y)/ρ(x))Ndδ(y,y′)V(d(x,y))dδ(x,y).$
16

### Lemma 9

Suppose W ∈ Bq1 for some $D2≤q1. For any integer N > 0, there exists CN > 0 such that

$|K˜(x,y)|≤CN(1+d(x,y)/ρ(x))N(d(x,y)V(d(x,y))∫B(y,d(x,y))W(z)d(z,y)dzV(d(z,y))+1V(d(x,y)))$
17

and

$|K˜(x′,y)−K˜(x,y)|≤CNdδ(x,x′)(1+d(x,y)/ρ(x))Nd1−δ(x,y)V(d(x,y))(∫B(y,d(x,y))W(z)d(z,y)dzV(d(z,y))+1d(x,y))$
18

for some δ > 0 and $0. If W ∈ Bq1 for some q1 ≥ D, then

$|K˜(x,y)|≤CN(1+d(x,y)/ρ(x))N1V(d(x,y))$
19

and

$|K˜(x′,y)−K˜(x,y)|≤CN(1+d(x,y)/ρ(x))Ndδ(x,x′)V(d(x,y))dδ(x,y).$
20

## Some technical lemmas and propositions

### Proposition 1

Let θ > 0 and 1 ≤ s < ∞. If $h∈Lipνθ(G)$, then there exists a positive constant C such that, for all BB(xr) with x ∈ G and r > 0,

$(1|B|∫B|h(y)−hB|sdy)1/s≤C∥h∥Lipνθ(G)rν(1+rρ(x))(l0+1)θ.$

### Proof

Since $h∈Lipνθ(G)$, then

$(1|B(x,r)|∫B(x,r)|h(y)−hB|sdy)1s≤(1|B(x,r)|∫B(x,r)|h(y)−h(x)|sdy)1s+(1|B(x,r)|∫B(x,r)|h(x)−hB|sdy)1s≤2C(1|B(x,r)|∫B(x,r)|∥h∥Lipνθ(G)dν(x,y)(1+d(x,y)ρ(x)+d(x,y)ρ(y))θ|sdy)1s≤C∥h∥Lipνθ(G)rν(1+rρ(x)+rρ(y))θ≤C∥h∥Lipνθ(G)rν(1+rρ(x))(l0+1)θ,$

where we have used Lemma 6 in the penultimate inequality.

Similar to the proof of Proposition 1, we immediately get the following.

### Lemma 10

Let $h∈Lipνθ(G)$, BB(xr) and s ≥ 1. Then there exists a positive constant C such that, for all k ∈ N,

$(1|2kB|∫2kB|h(y)−hB|sdy)1s≤C∥h∥Lipνθ(G)2kνrν(1+2krρ(x))(l0+1)θ.$

### Proposition 2

Let W ∈ Bq1 for $q1≥D2$. Let

Then there exists C > 0 such that, for any $f∈C0∞(G)$,

∥ℛf(x)∥Lp(G) ≤ CfLp(G)

where $1.

### Proposition 3

Let W ∈ Bq1 for $q1≥D2$. Let

Then there exists C > 0 such that, for any $f∈C0∞(G)$,

$∥R˜f(x)∥Lp(G)≤C∥f∥Lp(G),$

where q2 < p < ∞.

For the proofs of Proposition 2 and Proposition 3, one can refer to [3].

### Proposition 4

There exists a sequence of points ${xk}k=1∞$ in G, so that the set of critical balls QkB(xkρ(xk)), k ≥ 1, satisfies

• (i)
kQkG;
• (ii)
There exists N such that, for every k ∈ N, ♯{j:4Qj ∩ 4Qk ≠ ∅} ≤ N.

### Proposition 5

For 1 < p < ∞, there exist positive constants C, α and β such that if ${Qk}k=1∞$ is a sequence of balls as in Proposition  4, then

$∫G|Mρ,αf(x)|pdx≤C∫G|Mρ,β♯f(x)|pdx+C∑k|Qk|(1|Qk|∫2Qk|f(x)|dx)p$

for all $f∈Lloc1(G)$.

The above propositions have been proved in [9] and [10] in the case of a homogeneous space, respectively.

### Lemma 11

Let W ∈ Bq1 for $q1≥D2$, and let

and $h∈Lipνθ(G)$. Then, for q2 < m < ∞, there exists a positive constant C such that

$1|Q|∫Q|[h,R˜]f(x)|dx≤C∥h∥Lipνθ(G)infy∈Q{Mmν(|f|m)(y)}1m$

holds true for all $f∈Llocm(G)$ and every ball QB(x0ρ(x0)), where Mmν is a fractional maximal operator.

### Proof

Throughout the proof of the lemma, we always assume $D2≤q1. Let f ∈ Lp(G) and QB(x0ρ(x0)). For

$[h,R˜]f=(h−hQ)R˜f−R˜((h−hQ)f),$
21

we need to consider the average on Q for each term. By the Hölder inequality with m > q2 and Proposition 1,

$1|Q|∫Q|(h−hQ)R˜f(x)|dx≤(1|Q|∫Q|(h(x)−hQ)|m′dx)1/m′(1|Q|∫Q|R˜f(x)|mdx)1/m≤C∥h∥Lipνθ(G)(ρ(x0))ν(1|Q|∫Q|R˜f(x)|mdx)1/m.$

If we write ff1f2 with f1fχ2Q, due to Proposition 3, we get

$(ρ(x0))ν(1|Q|∫Q|R˜f1(x)|mdx)1/m≤C(ρ(x0))ν(1|2Q|∫2Q|f(x)|mdx)1/m≤Cinfy∈Q{Mmν(|f|m)(y)}1m$
22

for x ∈ Q, and using (17) in Lemma 9, we split $R˜f2(x)$ into two parts

$|R˜f2(x)|=|∫d(x0,z)>2ρ(x0)K˜(x,z)f(z)dz|≤C{I1(x)+I2(x)},$

where

$I1(x)=∫d(x0,z)>2ρ(x0)|f(z)|V(d(x,z))(1+d(x,z)/ρ(x))Ndz$

and

$I2(x)=∫d(x0,z)>2ρ(x0)d(x,z)|f(z)|V(d(x,z))(1+d(x,z)/ρ(x))N∫B(z,d(x,z)4)d(u,z)W(u)V(d(u,z))dudz.$

To deal with I1(x), noting that ρ(x) ∼ ρ(x0) and d(zx) ∼ d(zx0), we split d(zx0) > 2ρ(x0) into annuli to obtain

$(ρ(x0))νI1(x)=(ρ(x0))ν∫d(x0,z)>2ρ(x0)|f(z)|V(d(x,z))(1+d(x,z)/ρ(x))Ndz≤C∑k≥12−Nk(ρ(x0))νV(2kρ(x0))∫d(x0,z)<2kρ(x0)|f(z)|dz≤Cinfy∈Q{Mmν(|f|m)(y)}1m.$
23

Secondly, we consider the term I2(x). We have, for x ∈ Q,

$(ρ(x0))νI2(x)=(ρ(x0))ν∫d(x0,z)>2ρ(x0)d(x,z)|f(z)|V(d(x,z))(1+d(x,z)/ρ(x))N×∫B(z,d(x,z)4)d(u,z)W(u)V(d(u,z))dudz≤C(ρ(x0))ν∫d(x0,z)>2ρ(x0)d(x,z)|f(z)|V(d(x,z))(1+d(x,z)/ρ(x))N×∫B(z,4d(x0,z))d(u,z)W(u)V(d(u,z))dudz≤C(ρ(x0))ν∑k≥12−Nk2kρ(x0)V(2kρ(x0))∫d(x0,z)<2k+1ρ(x0)|f(z)|×∫B(z,2k+3d(x0,z))d(u,z)W(u)V(d(u,z))dudz≤C(ρ(x0))ν∑k≥12−Nk2kρ(x0)V(2kρ(x0))×∫d(x0,z)<2k+1ρ(x0)|f(z)|I1(WχB(x0,2kρ(x0)))(z)dz.$

Let q2 < m < D. Using the Hölder inequality and the boundedness of the fractional integral 1:Lm ↦ Lq1 with

$1q1=1m′−1β,$

where

(cf. Theorem 1.6 in [11]), we obtain

$∫d(x0,z)<2k+1ρ(x0)|f(z)|I1(WχB(x0,2kρ(x0)))(z)dz≤∥fχB(x0,2kρ(x0))∥m∥I1(WχB(x0,2kρ(x0)))∥m′≤∥fχB(x0,2kρ(x0))∥m∥WχB(x0,2kρ(x0))∥q1.$

Since W ∈ Bq1, we obtain

$∥WχB(x0,2kρ(x0))∥q1≤C(V(2kρ(x0)))−1q1′∫B(x0,2kρ(x0))W≤C(2kρ(x0))−2(V(2kρ(x0)))1−1q1′(2kρ(x0))2V(2kρ(x0))∫B(x0,2kρ(x0))W≤C(2k)l1−2(V(2kρ(x0)))1−1q1′(ρ(x0))−2,$

where in the last two inequalities we have used doubling measure and the definition of ρ, respectively. Therefore,

$I2(x)≤V(ρ(x0))−1/q1′−1∑k≥12k(−N+l1−1−d/q1′)∥fχB(x0,2kρ(x0))∥m.$

Finally, observing that

$∥fχB(x0,2kρ(x0))∥m≤(V(2kρ(x0)))1minfy∈QMmf(y)$

and using that $1q1′−1m=1D$ or $1d$, we have

$(ρ(x0))νI2(x)≤Cinfy∈Q{Mmν(|f|m)(y)}1m$

by choosing N large enough.

So far, we have solved the term (h − hQ)ℛf, now we want to control ∫ℛ[(h − hQ)f] dx by the term $Cinfy∈Q{Mmν(|f|m)(y)}1m$. We still split ff1f2. Choose $q2 and set $t=m˜mm−m˜$. Using the boundedness of $R˜$ on $Lm˜(G)$ and the Hölder inequality, we get

$1|Q|∫Q|R˜(h−hQ)f1(x)|dx≤(1|Q|∫Q|R˜(h−hQ)f1(x)|m˜dx)1/m˜≤C(1|Q|∫Q|(h−hQ)f(x)|m˜dx)1/m˜≤C(1|Q|∫2Q|f(x)|mdx)1/m(1|Q|∫2Q|h(x)−hQ|tdx)1/t≤C((ρ(x0))mν|Q|∫2Q|f(x)|mdx)1/m∥h∥Lipνθ(G)(1+ρ(x0)ρ(x0))(l0+1)θ≤C∥h∥Lipνθ(G)infy∈Q{Mmν(|f|m)(y)}1m,$

where we have applied Proposition 1 to the last but one inequality. Similarly, for x ∈ Q and using (17) in Lemma 9, we have

$|R˜(h−hQ)f2(x)|=|∫d(x0,z)>2ρ(x0)K˜(x,z)(h(z)−hQ)f(z)dz|≤C{I˜1(x)+I˜2(x)},$

where

$I˜1(x)=∫d(x0,z)>2ρ(x0)|(h−hQ)f(z)|V(d(x,z))(1+d(x,z)/ρ(x))Ndz$

and

$I˜2(x)=∫d(x0,z)>2ρ(x0)|(h−hQ)f(z)|V(d(x,z))(1+d(x,z)/ρ(x))N∫B(z,d(x,z)4)d(u,z)W(u)V(d(u,z))dudz.$

We start by observing that for $1≤m˜, $t=m˜mm−m˜$, and by Lemma 6,

$∥[(h−hQ)f]χB(x0,2kρ(x0))∥m˜≤∥fχB(x0,2kρ(x0))∥m∥(h−hQ)χB(x0,2kρ(x0))∥t≤C(∫B(x0,2kρ(x0))|f(y)|mdy)1/m∥h∥Lipνθ(G)(2kρ(x0))ν(1+2k)(l0+1)θV1/t(2kρ(x0))≤C2k(l0+1)θV1/m˜(2kρ(x0))∥h∥Lipνθ(G)infy∈Q{Mmν(|f|m)(y)}1m.$
24

For $I˜1(x)$, using (24) with $m˜=1$, we have

$I˜1(x)≤C∑k≥12−NkV(2kρ(x0))∫d(x0,z)>2kρ(x0)|h(z)−hQ||f(z)|dz≤C∥h∥Lipνθ(G)infy∈Q{Mmν(|f|m)(y)}1m∑k≥12k(−N+(l0+1)θ)≤C∥h∥Lipνθ(G)infy∈Q{Mmν(|f|m)(y)}1m$

if we choose N sufficiently large.

To deal with $I˜2(x)$, we discuss as in the estimate for I2(x) with (h − hQ)f instead of f and and $q1˜$ instead of m and q1, but we cannot avoid to discuss the different cases where 2kρ(x0) ≥ 1 and 2kρ(x0) < 1. Let

Using (24), we similarly have

$I˜2(x)≤C∥h∥Lipνθ(G)infy∈Q{Mmν(|f|m)(y)}1m,$

where we choose N large enough to ensure the above series converges.

### Lemma 12

Let $R˜=(−Δ+W)−12∇$ be the adjoint operator of the Riesz transform . Then there exists C > 0 such that, for any $f∈Llocm(G)$ and $h∈Lipνθ(G)$,

$Mρ,α♯([h,R˜]f)(x)≤C∥h∥Lipνθ(G)({Mmν(|f|m)(y)}1m+{Mmν(|R˜f|m)(x)}1m).$
25

### Proof

Let $f∈Llocm(G)$, x ∈ G and a ball BB(x0r) with x ∈ B and r < ϵρ(x0), ϵ > 0, we need to control $J=1|B|∫B|[h,R˜]f(y)−c|dy$ by the right-hand side of (25) for some constant c, which will be designated later. Let ff1f2, where f1fχ2B and f2f − f1. Then

$[h,R˜]f=[h−h2B,R˜]f=(h−h2B)R˜f−R˜(h−h2B)f1−R˜(h−h2B)f2≐A1f+A2f+A3f.$

Take $c=∫d(x0,z)≥2rK˜(x0,z)(h(z)−h2B)f(z)dz$. Then we have

$J≤1|B|∫B|A1f(y)|dy+1|B|∫B|A2f(y)|dy+1|B|∫B|A3f(y)−c|dy≐J1+J2+J3.$

At first, we consider J1. Note that d(xx0) ≤ r < ϵρ(x0) implies ρ(x) ∼ ρ(x0). By the Hölder inequality, Proposition 1 and Proposition 3, we have

$J1=1|B|∫B|A1f(y)|dy=1|B|∫B|(h(y)−h2B)R˜f(y)|dy≤C|B|[∫B|h(y)−h2B|mm−1dy]m−1m[∫B|R˜f(y)|mdy]1m≤C|B|[∫2B|h(y)−h2B|mm−1dy]m−1m[∫B|R˜f(y)|mdy]1m≤C|B|∥h∥Lipνθ(G)(1+2rρ(x0))(l0+1)θrν|B|m−1m[∫B|R˜f(y)|mdy]1m≤C|B|∥h∥Lipνθ(G)(1+2ϵρ(x0)ρ(x0))(l0+1)θrν|B|m−1m[∫B|R˜f(y)|mdy]1m≤C′∥h∥Lipνθ(G)(rmν|B|∫B|R˜f(y)|mdy)1m≤C′∥h∥Lipνθ(G){Mmν(|R˜f|m)(x)}1m$

for m > q2.

For J2, by the Hölder inequality and Proposition 3,

$J2≤1|B|∫B|A2f(y)|dy≤(1|B|∫B|A2f(y)|m˜dy)1m˜≤C(1|B|)1m˜(∫2B|(h(y)−h2B)f(y)|m˜dy)1m˜≤C|B|1m˜(∫2B|h(y)−h2B|mm˜m−m˜dy)1m˜−1m(∫2B|f(y)|mdy)1m≤C∥h∥Lipνθ(G)rν(1+2rρ(x0))(l0+1)θ(1|B|)1m(∫2B|f(y)|mdy)1m≤C∥h∥Lipνθ(G)rν(1+2ϵρ(x0)ρ(x0))(l0+1)θ(1|B|)1m(∫2B|f(y)|mdy)1m≤C′∥h∥Lipνθ(G)(rmν|B|∫2B|f(y)|mdy)1m≤C′∥h∥Lipνθ(G)((2r)mν|2B|∫2B|f(y)|mdy)1m≤C′∥h∥Lipνθ(G){Mmν(|f|m)(y)}1m,$

where $q2.

Finally, we consider J3.

Case of q1 ≥ D: By Lemmas 9 and 6, we have

$(∫2kr≤d(x0,z)<2k+1r|K˜(y,z)−K˜(x0,z)|m′dz)1m′(∫2kr≤d(x0,z)<2k+1r|h(z)−h2B|s′dz)1s′≤CNrδ(1+2kr/ρ(x0))N(∫2kr≤d(x0,z)<2k+1r1dm′δ(x0,z)Vm′(d(x0,z))dz)1m′×(∫d(x0,z)<2k+1r|h(z)−h2B|s′dz)1s′≤CN(1+2kr/ρ(x0))Nrδ(2kr)δV(m′−1)/m(2kr)∥h∥Lipνθ(G)2kνrν(1+2krρ(x0))(l0+1)θV1s(2kr)≤∥h∥Lipνθ(G)2kνrνCN(1+2kr/ρ(x0))N−(l0+1)θrδ(2kr)δV1/m(2kr),$
26

where $1m+1m′+1s=1$ and $1s′+1s=1$. Therefore, via the Hölder inequality,

$J3=1|B|∫B|A3f(y)−c|dy≤1|B|∫B|∫d(x0,z)>2r(K˜(y,z)−K˜(x0,z))(h(z)−h2Q)f(z)dz|dy≤1|B|∫B∑k=1∞∫2kr≤d(x0,z)<2k+1r|(K˜(y,z)−K˜(x0,z))(h(z)−h2Q)f(z)|dzdy≤∥h∥Lipνθ(G)|B|∫B∑k=1∞2kνrνC(1+2kr/ρ(x0))N−(l0+1)θrδ(2kr)δV1m−1s(2kr)×(∫d(x0,z)<2k+1r|f(z)|mdz)1m≤C∥h∥Lipνθ(G)∑k=1∞CN(1+2kr/ρ(x0))N−(l0+1)θrδ(2kr)δV1/m(2kr)(2kr)ν(2k+1r)νV1m(2k+1r)×((2k+1r)mνV(2k+1r)∫d(x0,z)<2k+1r|f(z)|mdz)1m≤C∥h∥Lipνθ(G)∑k=1∞2−kδCN(1+2kr/ρ(x0))N−(l0+1)θ{Mmν(|f|m)(y)}1m≤C∥h∥Lipνθ(G){Mmν(|f|m)(y)}1m$

if we choose N sufficiently large.

Case of $D2≤q1: By Lemmas 9 and 6, we have

$(∫2kr≤d(x0,z)<2k+1r|K˜(y,z)−K˜(x0,z)|pdz)1p(∫2kr≤d(x0,z)<2k+1r|h(z)−h2B|tdz)1t≤CNrδ(2kr)1−δV(2kr)(1+2kr/ρ(x0))N(∫2kr≤d(x0,z)<2k+1r|∫B(z,2k+3r)d(z,y)W(y)V(z,y)dy|pdz)1p×(∫2kr≤d(x0,z)<2k+1r|h(z)−h2B|tdz)1t+CN∥h∥Lipνθ(G)2lνrν(1+2kr/ρ(x0))N−(l0+1)θrδ(2kr)δV1/m(2kr)≤CN∥h∥Lipνθ(G)2lνrν(1+2kr/ρ(x0))N−(l0+1)θ×[rδ(2kr)1−δVt−1t(2kr)(∫B(x0,2k+3r)Wq1(z)dz)1q1+rδ(2kr)δV1/m(2kr)]≤CN∥h∥Lipνθ(G)2lνrν(1+2kr/ρ(x0))N−(l0+1)θ×[rδV1q1(2kr)(2kr)−1−δVt−1t(2kr)((2k+3r)2V(2k+3r)∫B(x0,2k+3r)W(z)dz)+rδ(2kr)δV1/m(2kr)]≤CN∥h∥Lipνθ(G)2lνrν(1+2kr/ρ(x0))N−l1−(l0+1)θ(rδV1q1(2kr)(2kr)−1−δVt−1t(2kr)+rδ(2kr)δV1/m(2kr))≤C∥h∥Lipνθ(G)2lνrν(1+2kr/ρ(x0))N−l1−(l0+1)θrδ(2kr)δV1/m(2kr),$

where $1m+1p+1t=1$ and $1q1=1p+1D$. Therefore, for m > q2,

$J3=1|B|∫B|A3f(y)−c|dy≤1|B|∫B|∫d(x0,z)>2r(K˜(y,z)−K˜(x0,z))(h(z)−h2Q)f(z)dz|dy≤1|B|∫B∑k=1∞∫2kr≤d(x0,z)<2k+1r|(K˜(y,z)−K˜(x0,z))(h(z)−h2Q)f(z)|dzdy≤∥h∥Lipνθ(G)|B|∫B∑k=1∞2kνrνC(1+2kr/ρ(x0))N−l1−(l0+1)θrδ(2kr)δV1m−1s(2kr)×(∫d(x0,z)<2k+1r|f(z)|mdz)1m≤C∥h∥Lipνθ(G)∑k=1∞CN(1+2kr/ρ(x0))N−l1−(l0+1)θrδ(2kr)δV1/m(2kr)(2kr)ν(2k+1r)νV1m(2k+1r)×((2k+1r)mνV(2k+1r)∫d(x0,z)<2k+1r|f(z)|mdz)1m≤C∥h∥Lipνθ(G)∑k=1∞2−kδCN(1+2kr/ρ(x0))N−l1−(l0+1)θ{Mmν(|f|m)(y)}1m≤C∥h∥Lipνθ(G){Mmν(|f|m)(y)}1m,$

if we choose N sufficiently large.

## Proof of the main result

### Proof of Theorem 1

Suppose $h∈Lipνθ(G)$. We choose m such that it satisfies q2 < m < p. We conclude from Proposition 4, Lemmas 5, 11 and 12 that

$∥[h,R˜]f∥Lp(G)p≤∫G|Mρ,α([h,R˜]f)(x)|pdx≤C∫G|Mρ,β♯([h,R˜]f)(x)|pdx+C∑k|Qk|(1|Qk|∫2Qk|[h,R˜]f(x)|dx)p≤C∫G|Mρ,β♯([h,R˜]f)(x)|pdx+C∥h∥Lipνθ(G)p∑k∫2Qk|{Mmν(|f|m)}1m(x)|pdx≤C∥h∥Lipνθ(G)∥(Mmν(|f|m))1m∥Lp(G)p≤C∥h∥Lipνθ(G)∥f∥Lq(G)p,$

where $1p=1q−νD$.

## Conclusions

We prove the Lp → Lq boundedness for the commutator which is generated by the Riesz transform and the function $h∈Lipνθ$. We generalize the corresponding results on the Euclidean space in [6] to the nilpotent Lie group, and they may have some applications in harmonic analysis and PDE on the Lie group.

## Acknowledgements

Research supported by the National Natural Science Foundation of China (No. 11671031, 11471018), the Fundamental Research Funds for the Central Universities (No. FRF-BR-17-004B).

## Authors’ contributions

Authors’ contributions

All authors contributed equally to the writing of this paper. All authors read and approved the final manuscript.

## Notes

### Competing interests

The authors declare that they have no competing interests.

## Footnotes

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

## Contributor Information

Tianzhen Ni, moc.361@32184010871.

Yu Liu, nc.gro.ukp@57uyuil.

## References

1. Varopoulos NT, Saloff-Coste L, Coulhon T. Analysis and Geometry on Groups. Cambridge: Cambridge Univ. Press; 1992.
2. Jiang RJ, Jiang XJ, Yang DC. Maximal function characterizations of Hardy spaces associated with Schrödinger operators on nilpotent Lie groups. Rev. Mat. Complut. 2011;24:251–275. doi: 10.1007/s13163-010-0041-8.
3. Li HQ. Estimations Lp$Lp$ des opérateurs de Schrödinger sur les groupes nilpotents. J. Funct. Anal. 1999;161:152–218. doi: 10.1006/jfan.1998.3347.
4. Coifman R, Rochberg R, Weiss G. Factorization theorems for Hardy spaces in several variables. Ann. Math. 1976;103:611–635. doi: 10.2307/1970954.
5. Janson S. Mean oscillation and commutators of singular integral operators. Ark. Math. 1978;16:263–270. doi: 10.1007/BF02386000.
6. Liu Y, Sheng JL. Some estimates for commutators of Riesz transforms associated with Schrödinger operators. J. Math. Anal. Appl. 2014;419:298–328. doi: 10.1016/j.jmaa.2014.04.053.
7. Folland GB. Subelliptic estimates and function spaces on nilpotent Lie groups. Ark. Mat. 1975;13:161–207. doi: 10.1007/BF02386204.
8. Folland GB. Lipschitz classes and Poisson integrals on stratified groups. Stud. Math. 1979;66:37–55. doi: 10.4064/sm-66-1-37-55.
9. Yang DaC, Zhou Y. Localized Hardy spaces H1$H1$ related to admissible functions on RD-spaces and applications to Schrödinger operators. Trans. Am. Math. Soc. 2011;363:1197–1239. doi: 10.1090/S0002-9947-2010-05201-8.
10. Liu Y, Huang JZ, Dong JF. Commutators of Calderón-Zygmund operators related to admissible functions on spaces of homogeneous type and applications to Schrödinger operators. Sci. China Math. 2013;56(9):1895–1913. doi: 10.1007/s11425-012-4551-3.
11. Bernardis A, Salinas O. Two-weight norm inequalities for the fractional maximal operator on spaces of homogeneous type. Stud. Math. 1994;108:201–207. doi: 10.4064/sm-108-3-201-207.

Articles from Springer Open Choice are provided here courtesy of Springer

 PubMed Central Canada is a service of the Canadian Institutes of Health Research (CIHR) working in partnership with the National Research Council's national science library in cooperation with the National Center for Biotechnology Information at the U.S. National Library of Medicine(NCBI/NLM). It includes content provided to the PubMed Central International archive by participating publishers.