• Nie Znaleziono Wyników

On equality of minimal and maximal realizationsAbstract. One considers minimal and maximal closed extensions of linear pseudo-differential

N/A
N/A
Protected

Academic year: 2021

Share "On equality of minimal and maximal realizationsAbstract. One considers minimal and maximal closed extensions of linear pseudo-differential"

Copied!
21
0
0

Pełen tekst

(1)

ROCZNIKI POLSKIEGO TOWARZYSTWA MATEMATYCZNEGO Séria I: PRACE MATEMATYCZNE XXX (1991)

Jouko Tervo (Jyvâskylà)

On equality of minimal and maximal realizations

Abstract. One considers minimal and maximal closed extensions of linear pseudo-differential operators in appropriate spaces. Specifically, the identity of the extensions in question from the Lebesgue space L2 into L2 (and also from the Hormander space Bp l into the Lebesgue space Lp) is examined. Furthermore, a sufficient condition for the identity of the extensions from Bp k into

is obtained.

1. Introduction. Denote by L(x, D) a linear pseudo-differential operator from С* into the Schwartz class S. The minimal closed extension of L(x, D) in the space BPtk is denoted by L~k. The corresponding maximal closed extension is denoted by L*tk. Here Bp k, p e [ l ,

o o [,

k e K ', is the linear subspace of the dual S’ of S such that the Fourier transform Fи belongs to l}™(R") and that Fu- k e Lp: = Lp (R") for any

и e

Bp k (К ' denotes an appropriate collection of weight functions R"->R). Bp k is equipped with the norm ||м||р,к:= (27i)~"/p||Fw 'k\\Lp. Similarly one defines minimal and maximal closed extensions L~,q and I/Д from Bp<ko to Lq (here k

0

eK ', k

0

= 1). Furthermore, if L(x, D) maps Co into C°°(G) and the formal transpose L'(x,D): Co(G)->-S exists, the extensions L~k(G) and L'*k(G) from Bp k to Bl°ck(G) can be defined. Here G is an open subset of R" and Bl°k(G) is the appropriate local distribution space corresponding to Bpk.

One knows that always L~k <= L'*k. If L~k = L'*k, we say that L(x, D) is essentially maximal in Bp<k. Some classes of operators are known to be essentially maximal in appropriate spaces (cf. [6], [4], [9], [3], and [7]).

We establish a decomposition for the convolution iJ/*(0L(x, D)(p) (cf.

Lemma 3.1, Theorem 3.5). This decomposition is applied to show a sufficient criterion for the essential maximality of L(x, D) in L2, where L(x, D) belongs to the Beals and Fefferman class of operators (cf. Corollary 3.6). Further­

more, we prove that L~k(G) = L'*k(G) if

sup|(DJBfL)(x, «I ^ С,,„л M £ ):= C.j,iJE(l + |{|2)1/2,

x e K

where K <è G, p e [ l ,

oo[

and k e K ' (cf. Corollary 4.8). Finally, the relation Lpp, = L'pfp-, p e ] l , 2], is obtained if

sup I(D*xDlL)(x, 0 | ^ C ^ k ^ ) .

xeR "

(2)

We remark that the operator L(x, D) defined (in R2) by (L(x, D)cp)(x) = [eXl(l + \x\2)(Dl + l)(eXl(l + M2)<?)](x) is not essentially maximal in L2: By partial integration one sees that

Re(L(x, D)(p, (p)b2 = Re((Dl + l)(exl(l + \x\2)(p), eXl(l + \x\2)q>)L2

^ \\eXi(l + \x\2)(p\\L2 ^ \\eXl(p\\L2, for all (peC$(R2), and so one easily sees that the kernel N(L

2

,ko) = {0} (note that B2,ko = L2). Oil the other hand, observe that the function и defined by u(x) = 1/(1 + |x|2) belongs to L2 and that (L(x, D)u)(x) = 0 for any x e R 2. Hence ueN(L'2fko)\{0}, which shows that L~2tko ф L'2*ko.

2. Preliminaries

2.1. Let K' be the collection of positive continuous functions к: R"->R so that

(2.1) Щ + fi) < С к М Щ ) for all £, n eR",

where C and R are positive constants and kR{Ç):= (1 + |£|2)я/2. Replacing in (2.1) £ with Ç + r] and rj with —rj one sees that k _1eK ' whenever k e K '. More generally, kse K ' for s e R if k e K '. In addition, the function fcv defined by k'/ (Ç) = k( — Ç) lies in K' for any k e K '. Let p be in the interval [1, oo[.

Denote by BPfk the completion of Co := C

q

(R”) with respect to the norm (2.2) M \ p,k = ((2

jc

) - f \ ( F № № ) \ pdi)1"’,

R "

where F is the Fourier transform from the Schwartz class S into S. Let S' be the dual of S (that is, S' is the space of all tempered distributions). Then Bp<k is essentially that subspace of S' whose elements и satisfy: The Fourier transform F u e L l°c(Rn) and Fu ■ k e L p. In addition,

(2.3) IMU = ((2 я)"” J \{Fu-k){(rdi)11-.

R "

Choose 0from C

q

such that 0(x) — 1 for all x e B ( 0, 1):= (

x g

R"| |x| < 1}.

Define functions 9', 9t and 0J

g

C

q

by

(2.4) '9' = (2к)~п9 \ 9t(x) = 9(x/l), 0J(x) = 9'(x/l), where le N. Furthermore, let

(2.5) = F9'l.

Then F[j/l = (2n)n(9’iy = 9t and so |(Fi/ff)(£)| < C and (Fij/^Ç)-* 1 as /-►

00

. Hence by the Lebesgue Dominated Convergence Theorem

(2.6)

\\ф,*и-и\\р,к =

((2я)-" J

\ (W im F u )(0 -(F u )(t;))k (l;fd l;y " ’

R n

— > 0 as / ->

00

.

(3)

Furthermore,

(2.7)

m

— w||p,fc — > 0 as l-rco,

since ||0,и||р,* ^ С \Щ

1

гкл\\и\\Ргк < C||0||1>fcJ u ||p,fc (cf. [5], p. 42).

Let G be an open subset of R" and let {Gj} be a sequence of relatively compact open subsets of G such that Gj a G

j+ 1

and (Jj°=i G} = G. Denote by BlpC k(G) the completion of C

q

(G) with respect to the quasinorm

00

b,k(<p)=

I

wn\®j<P\ \P,k/(i+\\®j<p\\p,k)’

7=1

where {6L} c Ccf(G) is a sequence such that supp 0 . c= Gj+l and 0^(х) = 1 for xeGj. Then Bpj^G) is essentially that subspace of D'(G) whose elements и obey: F(\J/u)-keLp for any ^

g

C

q

(G). In addition, the quasinorm

00

(2.8) qp,k(u) = X (l/

2

j)\\&ju\\P'k/(l + \\&jU\\P'k)

7 = 1

defines the topology in Blpf(G) and so the topology of Blpf(G) is equivalent with the topology defined by the seminorms pPxj( u)'- = 11®,-м|1р,ь j e N.

2.2. Let L be a linear operator from Co into S such that the formal transpose L : C

q

-+S exists, that is, there exists L' such that

(2.9) (Ь(р)(ф): = J (L(p)(x)i//(x)dx = (p(L'\l/) for (p, феСо.

R "

Define dense operators Lpk. Врк-+Вр>к and L *k: Bpk-^Bpk as follows:

|D (L M) = C

q

,

\ L Pjk(p = L(p for (p e C

q

,

rD(L'*k) = {ueBp>fc| there exists / e B pk such that

J u(L'(p) = f((p) for all (peCg},

l L'*ku = /.

НФ)\ < IM U M Ip, 1/kv for u e B Ptk, феБ

(here p 'e ] l ,

oo],

l/p + 1/p' = 1), LPtk is closable and L'*k is closed. In addition, Lp k c= LPtk. Let LPtk: BPtk^ B Pfk be the smallest closed extension of LPtk. Then

T ~ r- T'*

L'p, k Eip<k.

Let ре [1,

oo

[ and denote (as above) by Lp the Lebesgue space in R". We find that ||u||p = ||Fuv||P)jto for u e L p and so

ll<Ai*M-w||P = IIFuv - W 'FMv||p>Jto = ||F

mv

-0? -Ft/v||Mo

\\u — 0

jm

||p->O as /-►

oo.

(2.10)

(

2

.

11

)

Since (

2

.

12

)

0 as

l

-> oo,

(4)

For p,qe[_ 1, oo[ we can define a dense closable (resp. dense closed) operator LPtq: Bp<ko-*Lq (resp. L*q: BPtko-+Lq) by

(2.13) \ DiL^ ~ >

[Lp<q(p = L(p for (p

e

C

q

,

(2.14)

D (L’p*q) = {u e B p<k01 there exists / e L q such that u(Lq>) = f((p) for (peCo}, L’

p

> = /•

Denote by L~e: BpM-+Lq the smallest closed extension of Lp>9. One sees that L c L*

upa

'

Suppose that L is a linear operator Co -►C00(G) such that the formal transpose L': Co^G)-+S exists. Define dense operators L pk(G): Bpk->Blpk(G) and L*k(G): BpJt->Bft{G) by

(2.15) D(Lp>k(G)) = C%,

Lp,k(G)(p = Lq> for (peCo,

j D(Lp*k(G)) = {u e BPtko I there exists / e B lp k(G) such that

(2-16) < u{Lcp) = f(<p) for (peCo(G)},

lL'p*k(G)u = /.

Then Lpk(G) is closable (cf. [10], p. 77), L'p* (G) is closed and Lpk(G) c L'*k(G).

Let Lpk(G) be the smallest closed extension of Lpk(G). One has

L~Ptk(G ) ^ L'p*k (G ).

2.3. Let (Ф, ф) form a pair of weight functions in the sense of [1]. Choose M and m from R. We say that the smooth function L(-, •)eC 00(R" x R") belongs to the class S if

(2.17) KDÏDfO(*, «I « 9Ф ” - М(х, a

for all x, { e R ‘ . Define the linear pseudo-differential operator L(x, D) by (2.18) (L(x, D)<p)(x) = (2я)-« f L(x, <peS,

R "

P

where L(-, ')eS%£. The class of pseudo-differential operators is defined as follows:

Ljfr" = {L(x, D)\L(x, D) is defined by (2.18), where L(-,

Let L(x, D) be in . Then L(x, D) maps S into itself and the formal

transpose L (x , D): S-+S exists. Furthermore, many rules of computation are

valid in {JM,mL%$ (cf. [1]). We recall

(5)

Th e o r e m

2.1. Let sé be a subset of S%’§, such that

(2.19) p°>°(L(-f •)):= sup £)</>'“'(.x , (DJDfL)(x, 01}

x ,(*e R "

< C«., < oo for all L{-, -)estf, a, /leNô.

Then there exists a constant C > 0 such that

(2.20) ||L(x, D)<p\\:= ||L(x, D)(p\\Ll ^ C\\<p\\ for all (peS,

where C is independent of L(x, D) on the set {L(x, D)\L(‘, -)es/}. ■ For the proof, cf. the proof of Theorem 2 in [1].

Let L(*, •) in C°°(R"xR") be such that

(2.21) sup I(

d

;

d

?

l

)(

x

, 01 « C ^ , KkKt + m (()

x e K

for any K C G (here G is an open set in R") and (a, f ) e N l n. Then L(x, D) maps S into C°°(G) and the formal transpose L'(x, D): C

q

(G)->S exists (cf. [8]; the formal transpose can be defined by L'(x, D)<p = (фЬ)'(х, D)<p for (peCo (G), where i^eCo(G), ф(х) = 1 in supp<p).

3. On minimal and maximal realizations in L2 3.1. Define a seminorm p ^ f

1

on by

РЙГ(£<-> •)) = sup {ф|«-м (х, ()ф'-'~т(ху 01 (DÎDÏD(x, 01}.

x ,£ e R "

Then S$f;” becomes a Fréchet space with the locally convex topology defined by the seminorms p ^ f 1, a, /J

g

N

q

. If феБ we denote the pseudo-differential operator generated by the symbol (Fi/r)(-) by \f(D).

Let Oj, j e N, be as in Section 2.1. Furthermore, let Lj( -, •) = 0; L(-, •). We have

Le m m a

3.1. Suppose that L(*, ‘)

g

C°°(R"

x

R") such that for some Ca>/? > 0

and Na>fie R

(3.1) sup \ m D lU x , O K C .jk H .J t) f o r t e R".

x e s u p p fy

Then for each N e N and for any ф е$

(3.2) ф*Ь](х, D)q> = X ( ( -i)M/y'){DyxLj)(x, 0)((хуф)*(р)

\ y \ < N

+ R j^ tN(x, D)(p, for all (peS, where

(3.3) R ^ .iï( x , 0 = (2 it)-" } lV (l-ï)''-1 £ (1/y!) J J S W ) ( i + «(4-{))

0 \ y \ = N R " R "

x (DlLjiy, £)ei{x~y,n~®dydr\.

(6)

P ro o f. In virtue of the Fourier inversion formula we obtain (3.4) (ф * Lj(x, D)<p)(x) = (2ic)" :" f F(\j/ * Lj{x, D) <p)(ij)ei("'x) dr\

R "

= (2n)~n J (F\l/)(rj)F(Lj(x, D)(p){rj)ei(n'x)dri

R "

(2rc)- n f

(F\lf){ri)(

J

(Lj{y, D)(p)(y)e~i{y',,)dy)ei(,t’x)dri Rn R"

= (2it)-" { W M 1 {((2it)-" J Lj(y, i)(F<p)(i)ei™ d t ) e - ,<™'dy}ei"-1‘4r,.

Rn R» Rn

Since F\/f and F(peS and since suppL^-, rj) c supp^-, the order of integration in (3.4) may be changed to obtain

(3.5) (i//*Lj(x, D)<p)(x)

= (2*)-" f ((2it)-" f J (Рф)(пЩ(У, i)e‘<*-™-ï’dyd4)(F<p)(Ç)e,™ d ( .

R " R " R "

In virtue of the Taylor formula we get (3.6) (2 к ) - J J W M L j f y , i)e**-™-l4ydr,

R " R "

= ( 2 к ) - Y ( W ) f J ду(Р ф )а )((-ч У ^ (у , Ç)ePx~’*~adydri + RK(x, Ç)

| y | < N R " R "

= (2

k

) - Y ( W ) f f 0’(*'Л Ю (В Д О '. i)e«*-™-«dydr,+R„(x, 0 .

\ y \ < N R " R "

where

(3.7)

Rn(x,

0

=

(271)-" J J

N S i l - t f - 1

X Z (VyW4Fÿ)(Z + t(q-Ç))(DyLj)(y> Q e^-M -^dydri =:Rjt^ N(xt £).

M = N

Furthermore,

(3.8) ( 2 к ) - J (2

k

) - £ (1/y!) f f dy(F>l/)(i)(DyyLj)(y, i)

R " |y| <iV R " R "

x e‘(JC - ™ ~ « dy drj • (F<p)(É) ei M dil

= (2k) - y

( W ) s (S (ПЩЬМУ’ i)e‘<x~™~l)dy)d>i)

[ y | < N R " R " R ”

x ( — i)171F ((xy ф) * (p) (£) ei(4,x) dÇ, where

J (f (D}Ly)(y, £)е-«™-*ЫуУх«-*Чг1

R " R "

= f F((DlLj}(-, Q){ti-№*«-*>dri = (2n)n(DyxLj)(x, 0 .

(7)

Hence

(3.5)—(3.8)

imply the assertion. ■

Furthermore, we need the following lemma:

Lemma 3.2.

Suppose that L(-, •) belongs to S a n d that is the subset of S such that

(3.9) p8S((FM -)):= s u p W F M Q \ * m {x, É) ^ E, for all фе®.

X , i

Let •) be defined by (3.3). Then •)eS® JN,M-JV and

(3.10)

0) < C for all \J/e£%,jeN,

where C >

0

is independent of j and ф.

P ro o f. In virtue of (3.9), M is a bounded subset in 5 ^ . Furthermore,

(3.11)

|(Z>ÏZ>!Lj)(x,

£)|

= | Z 0 (D * -”^)(x)(D;Df L)(x, £)|

u<a

= I Z 0 ( i

/ / |“ - “ |) (с“ - “й) (х/ / ) (й ; о ?ц (х ,

а |

« £ O sup|(D «-“0)(x)|C^<fM-l'"(x, 0

x

«S Z C)sup|(D“-"0)(x)|CM Cl* -“l^ M- |'l|(x, а < Г “ |“|(х, 0 and thus p l’J(Lj( - , - ) H Q „ that is, {ТД-, -)}j6N is a bounded set in

In virtue of Theorem 1 in [1] (cf. also the proof),

(3.12)

(ifi(D)oLj(x, D))(x, ()

= z (wmimamLjXx, а + а д , a

\ y \ < N

= Z

( ( - 1 ^ / у Щ х У ф т ( ^ хЩ х , Q + R j,t A * > 0

\ y \ < N

= Z ( ( -/)м/у!)((1)3;^)(х, D )o(xV)A(D))(x, ^ ) + ^ , N(x, ^),

\ y \ < N

where {Rj^,it}j.^ is bounded in S

0

^ N,m~N. Since

$(D)oLj{x, D)(p){x) = (ф*Ь]{х, D)(p)(x)

one gets the assertion from (3.12) (note that one has (L(x, D)o\j/{D))(p = L(x, D)№*(p)). л

Lemma 3.3.

Let фг be defined by

(2.5).

Then

PoliHWiM')) < Е/ for all l eN .

(3.13)

(8)

P ro o f. We have = F(F9'i) = (2

tu

)"(09

v

= fy. Since 6

e

C

q

we find constants Cp > 0 such that

KD^Kfll *5 C^l + lfD-1' 1

and so (recall that Ф(х, fl ^ C(l + |{|))

Po:?(e,(-)) = sup |(а<'0,)(й|Ф'<,'(х, 0 = sup(l/ll,l|) |(дев){(/1)\Ф|д|(х, fl

x,£ e R n

^ C,sup(l//W)(/W/(/ + | ^ l ) C ^ ( l + | ^ ' ^ C^C^I, which completes the proof. ■

3.2. In the sequel we shall assume that there exists N e N such that (3.14) Фм ~"{х, £)фт~*{х, £) ^ С .

The inequality (3.14) is always valid with M = N when M ^ m:

Фм " м(х, £)фм ~т{х, <*) ^ CM“m.

It is also valid for any M, m eR (with iV large enough) when Фф ^ ckx with some c > 0, x > 0.

The following lemma is easily seen:

Le m m a

3.4. Let L(-, •) be in S%;™. Then there exist re N and C > 0 such

that

(3.15) ||L(x, D)(p\\ ^ C\\(p\\kr,

(3.16) ||L(x, D)q>\\k_r < C||<p|| for all (peS.

P ro o f. Due to (i) (cf. the properties (i)-(iv) of (Ф, ф) in [1]), k - r(-)eS%;™

n S ^ ,ф’~т for r large enough. Hence the application of Theorem 2.1 with L(x, D )ok-r(D) and /c_r(D)oL(x, D) yields the assertion. ■

Since L{x, D): S-y Sis continuous we can define the continuous extension L: S ' S ' of L(x, D) by

(3.17) (Lu)((p) = u(L'(x, D)q>).

We show

Th e o r e m

3.5. Suppose that (3.14) is valid and that L(-, feS^;™. Then there

exists a constant C > 0 such that

(3.18) ||ih * L u - £ ( ( - 0 |у|/7!)(^Ь)((^«Аг)*м)||^С |М |

\ y\ <N

for all u e L 2.

(9)

P ro o f. A. First we establish that

(3.19) ||«^*L(x,Z))<p- X ( ( - 0 |у|/у !)(^ Ь )(х ,/))((х ^ г)*Ф)||

\ У \ < И

^ C\\(p\\ for all (peS and le N.

For any le N one gets by (3.13), (3.14) and (3.10)

(3.20) p%?(RM ,.s C> ')) = sup<p№(x, £№'*'(x, й |D i D l R j ^ H(x, {)|

*■!

< C 'su p $ N- A, + ''’l(x, + W(x, O lD lD lR j^ x , f)I

= С'р«<Г№ -” - ,'(Л ^ „ я(-> •)) ^ C'C.

Due to Theorem 2.1 there exists a constant C > 0 such that (3.21) № j.* A x , B)<p\\ < C|MI for all

where C is independent of j, le N. Hence we obtain by (3.2) (3.22) | |

D)(p- X ({ -i )M/y^(DyxLj)(x, Я)((хг(А,)*<р)||

|y|<N

^ C||<p|| for all <peS.

Furthermore,

(3.23) ||^*L ;(x, D)(p — ij/l*L(x, D)(p\\

^ IM oo.i||(b/x, D ) - L { x , D))<jo||

= Ih M c o .ill^ x , D)(p — L(x, D)(p\\->0 as j -ю о, where we have set

(3.24) IMIoojk = sup|(F^)(£)fc(£)| for (peS and к e K ' . In addition,

(3.25)

DlLj(x,D)((xy<l/,)*q>)(x)

= (2

)-" J D’((0;L)(

x

,

R"

= (2я)-" j X © (1//“') f (В"0)(х//)(/)Г“Ц(х,

R " « ^ y R n

м < у

and so

(3.26) ||д а ь ;)(х, D)((xyiA{)*(p)-(Z)];L)(x, jD)((xy^j)*ç))||->0 as ;->oo.

Hence we* obtain (3.19) from (3.22) by letting j-* oo.

(10)

B. Let и be in L2. Choose a sequence {(pn} c C

q

such that

||ç>„ —u II -»0 as n-* oo. Let r be in N such that (3.16) holds. Then

\\L(x, D)q>n — Lu\\k_r-+Q as oo and so (3.27) Цф^Цх, D)(pn-il/l*Lu\\

< \\^i\\oo,kr\\L{x, D)(pn-Lu\\k_r-*0 as n->co.

Furthermore, (хуф1)* и е Н к/.= B2)ks and

(3.28) W(xyil/i)*(pn- ( x y\J/l)*u\\ks < ||xy^rI||00,Jk,||<

3

t>B —

m

|| — *>0 as n — > со, where seN such that

(3.29) ||(В Д (х , 2>MI ^ C\\<p\\ks

(note that (DyxL)(-, and use Lemma 3.4). From (3.29) we get (3.30) ||(B ÏL )((x W * u )-(D ’xL){x, B )((x'«*V ,)||

^ C||(xV,)*M -(xvi/q)*(pJ|*s-->0 as n — > oo.

Since by (3.19) for any N

e

N one has

||./q*L(x, £)(/>„- £ ((-O w/y!)(l)iL)(x,D )((x^I) * ç » J ^ C M ,

|у|<ЛГ

we obtain (3.18) from (3.27) and (3.30) by letting n-»oo. This completes the proof. ■

We show the following criterion for L~ := L~2M = L'# : = L'2^0.

Co r o l l a r y

3.6. Suppose that L ( -,-)

e

S$;™ and that there exists

such that

(3.31) ^ ) ^ с Ф м - 1( х , а г _1( х , а

(3.32) D(L'*)czHq: = B 2<q.

Then

(3.33) L~ = L'* .

P ro o f. The operator (Log-1)(x, D):= L{x, D)oq~1(D) belongs to Ьф\ф (note that by (3.31), q~l ( ‘) lies in S®,Jf + 1,-m + 1). In the inequality (3.14) is always valid and so by Theorem 3.5 (with N = 1) we obtain for any u e b 2

(3.34) \\i//l*(Loq~1)u — L o q ~ 1(\l/l*u)\\^: C\\u\\.

Let и be in D(L’*). Then by (3.32), q u e L 2. Furthermore,

(11)

(3.35) (Loq

1

(qu))((p) = (qu)((Loq ■ 1)'(x, D)q>)

= (qu)((#vГ 1 (D)oL'(x, D)<p) =

m

v

(^)

o

v

) - 1(D)

o

L'(

x

, D)<p)

= (L'*u)(cp) and

(3.36) F(Ÿi*qu) = (Fil/JFiqu) = F(q№t*u)).

As in (3.35) one sees by (3.36) that

(3.37) L o q ~

1

(\j/l*qu) = L o q ~

1

(q({f/l*u)) = L(^,*n).

Since ф ^ и е Н ^ (where re N such that (3.15) holds) we see that ^,*ue£)(L~).

Hence by (3.34), after replacing и with qu and applying (3.35) and (3.37) we obtain

and then (3.38) where

||^ * (L '# M)- L ~ ( ^ * M)||^ C |N |,

||L~(i/f|*u)|| ^ C ||m||9 + ||i^i||00>1||L,#uH,

M U

i

= sup |(Fty,)(£)| = sup|0j(£)| ^ sup |0(£)| <

00

.

? « « ■

Using (3.38) and the Banach-Saks Theorem we observe that there exists a subsequence {ф^*и}

c=

{ф^и} such that for some g e b

2

II(IA) z L~('l'ii * u )-g \\^ Q as s —* со.

j = i

Since

and

(1/s) £ Ь~{ф

1

л*и) = L~((l/s) £ №ij*u))

J

=i

j

=i

||(l/s) Y

j

Ф^*и-и\\->0 as s —> со,

j = i

we see that ueD(LT) and L~ и = g (= L * u). This finishes the proof of (3.33). ■ R em ark 3.7. A. For M = m = 1, we can choose q = 1. The inclusion (3.32) is also trivially valid since H Q = L

2

for q = 1. Hence L~ = L '# when

' ) е ^Ф,^'

В. Corollary 3.6 shows that any operator L(x, D)el%>\™ from B

2

q to L2 is essentially maximal (the definitions of minimal and maximal realizations are as in 2.2). Specifically, any operator L(x, D): B

2

tkm_(

6

_e)->L

2

*s essentially maxi­

mal when L(-, •) belongs to the Hormander class of symbols.

(12)

4. On essential maximally in BPtk

4.1. Let L(-, •) be in C°°(R"xR") such that

(4.1) sup № Щ Щ х , «I «

■ xesuppOj

where > 0 and

N p e

R. Then

Lj(x, D)

defined by (2.16) maps S into S and the formal transpose Ц(х, D): exists. Furthermore, one shows as in

3.1

that for

l, j e

N

(4.2) ^i*Lj(x, D)(p = Ь;#,*<р) + Ял,(х, D)q>, where

(4.3) K,,,(x,{) = } X

S

! (M M e + t t i - t y i D ’LjHy, ()e‘<*-™-«dyd

4

dt

O |y| = l R " R"

= f X i (dWl)(i + t4)(i(DyyL1) ( y ^ ) e - ‘^ d y ) e i^">dt,dt

O Ivl = 1 R " R"

= } I | ( ^ 0 г) ( ^ + ^ ) т ( ( ^ ^ ) ( - л ) ) ( / / ) ^ ^ ^ .

О |y| = 1 R "

Lemma 4.1.

Suppose that L(-, •)

g

C00(R"

x

R") obeys

(4.1)

and that Rjyi{',

•)

is defined by

(4.3).

Then

RJt/( - ,

-)

g

C°°(R" x R") and

(4.4) m D l R j ^ x , 0\ = Ca,PJkNfi(t) for x , £

g

R", where Caypyj is independent of

le

N and Np : = maxu<cp{Nu — 1}.

P ro o f. A. Since

i))(n) = J (D’LjKy, ()e~‘™ d y

supp Oj

one sees that

(4.5) DI[(F(D;L

j

)(-, = F((DyyD^Lj}(-, 0)(rj).

We recall that (cf. the proof of Lemma 3.3)

(4.6) \ ( 1 г е № + щ ) \ ^ с ак - м в + щ ) . Furthermore, since (for t > 0 and for a fixed £0

g

R")

l (8-’e,)(( + tn)F((DlLj)(-, t y W ^ d q

R "

J (d'0№ + trj)F((DyLj)(-, t y W e ^ d r i

(l/ t)(is u p P0 + B (£ o ,l))

for all (x, <^)

g

R "

x

B(<^0, 1) we see that -)e C°°(R" x R") and (4.7) (DlD^Rjj)(x, i)

= î Z J

DlDillSW M + triF liD lL jH ^ ty W ^ -id q d t.

О I у I — 1 R "

(13)

B. In view of (4.1), (4.5) and (4.6) we obtain for any a, jfreNô and

tg

N

ô

, |

t

| ^ n +

2

,

\ Ч ' В Щ W Ш + t4)F((D’ L j)( -, « ) W e '(« )]|

= I Z (Î) ( - i ) l“l(5"+ '0,)(« + t '! ) f ( ( B r " + , C f-" L J.)(-> 0 )fo K < * '1

U*Z0

Z Z © C +; +,,)C I, + yC„,(,- „ J ||D '+“+' - ' ’0J.||I.l J:. l„+rl({ + tr,)/cK„ . B« )

U^0 V^x + tx + y and so

(4.8) ID>iD\[(d4i(t + t

4

)F((DlL)(-, O)0f)e'(JU,,]|

^ X

k - \ u\ - 1(Ç + trj)k Nll_u( Ç ) k - i n+2)(ii).

We decompose the integration over R" in (4.7) into the two parts

f DaxD ll № = J E&Dtl ]<fy + J D iD il Ш

R" Ul<2|i»| |«|>2|ir|

Bt . If |£| ^ 2|f/|, one has

к - м -

1

в + »

1

)к*'-ы( № - ь + г М <

and so we obtain by (4.8)

(4.9) I J />“/>?[ ] ^ | < 2 С ; (/,)Уа / с _ (п+1)( ^ » / ) ^ ( а .

1«1^2|Ч| R"

B2.

Suppose that |£| ^

2|f/|.

Then

(4 .1 0 ) \ z + t n \ > |£| - ф | ^ № - \ ч \ > ( 1 /2 ) |£ |

and so

(4.11) k - M-i ( ^ + tri)kNll_u(^)k-(n+2)(ri) ^ k -i( ^ + tti)kNf_u{ ^ k - (n + 1){ri)

<

2

к _ ! (£)kNp_u(Ç)k-{n +

1

)(r]).

This implies by (4.8) that

I J

D%Dll Щ 2

Q w ( J

k . in+1)m n ) k N^ ) .

\S\Z2\t,\ R "

Thus the estimate (4.4) follows from (4.7).

4.2. Since Rjtl( -, •) obeys the estimate (4.4) we know that the formal transpose R'jti(x, D): S-+S of RJti(x, D) exists. Furthermore, we have for p e ] l , oo[

Le m m a 4 .2 .

Suppose that

s f

is a subset of

C ® (R " x R " )

such that for any

*(•, •) € s / and (a, (1) € N

q

" there exist constants Ca P > 0 and Nlt e N so that (4.12) sup ID%D\R(x, 0 | « CKtks ,(0-

jceR"

(14)

Furthermore, assume that

(4.13) I \F(e,RC, -Ч ))(Ч -Ф (Ч )А ! МЩ ),

R "

for £

g

R",

îg

N and (4.14) j и м о ,

- i , ) ) ( 4 - o \ / m d i

«

к / т ,

R "

for rjeR",

îg

N and R(-, -)

g

<

s

/ . 77zen t/iere exists a constant C > 0 such that

(4.15) HR'(x, D)(p\\Ptk ^ C\\(p\\Ptk for (peS and R ( v ) e ^ >

where

(4.16) C ^ (2n)-nK 1/p'M llp ^ (2n)~nm<ix{K, M}.

For the proof, cf. the proof of Theorem 4.2 in [8].

From Lemma 4.2 we obtain for р с[1 , oo[ (recall that Rm( - ,- ) : = e mR(-,-))

Th e o r e m

4.3. Suppose that

s tf

is a subset of C°°(R"xR") such that (4.12)

holds and

(4.17) sup \(D%R)(x, 01 < Ca,m for |a| ^ N k + n + 1, R(-, -) e s t,

x e s u p p6rn

where N ke N such that

(4.18) kK + ^ C t y i |) k ( f ) .

Then there exists a constant Cm > 0 such that

(4.19) \\Rm(x, D)(p\\Ptk^ Cm\\(p\\p>k for (peS and R{-, -)ej t f (where Cm is also independent of p).

P ro o f. A. First we shall verify that

(4.20) J H W . - ч ) ) ( ч - 0 \ / к ' ' Ш ч ^ М т/к ''(0 for ÉeR*.

R "

(4.21) S № * „ ( - . - Ч ) ) ( П ~ Ф ' ' Ш < K »*v(4) for 4 eR-,

R "

where Mm and K m are independent of / and R(-, )

gjs

/.

From (4.18) one gets k( — Ç) = k( — rj — (Ç — rj)) ^ CkkNk(Ç — ?/)/c( —

77

) and so Ck/c)vk(^ —^/)//cv(0- For any

tg

N

q

, |

t

| <JVk + w + l, we obtain (4.22) |(>/-{)'F(0IRm(-, ->1))(>1-{)| = |F(DJ(0,R„(-, - ч)))(ч - {)|

« £ © |l/iM |sup|(P“0)(x)| J |(С Г “Л„)(х, -

4

)|dx

ut x R "

« £ © s u p |( B “0)(x)| E С 7 )(1 /т|'’|ДО',0 Д 1 sup |Р Г “- '« ( * , -ч)1

и $ т x v ^ T — u x e s u p p 0 m

(15)

and so one has with C'm := max {СТ)Ш }

|x| ^ Nic + n+ 1

|F(e(R„(-, -4))(4-9|/fcv(l) < C'mCk(feWl(4-0/fcv(«)fc-W,+.+i1('î-{)

= c mc t k_(„+1,(f/-i)/k v( a . Thus

J № «„,(•, -4))(4-0|/fcv(»i)<i4 « c^ct(J fc-„+I)(4))/fc''(fl =-.M/k4i),

R " R "

where M does not depend on R(-,-)estf.

Since fcv (£) ^ ChkNk(rj — £)kv {rj), the validity of (4.21) can be verified similarly.

B. Due to Lemma 4.2 one has for p' e ] 1, oo [ (4.23) \\K (x, D)q>\\

p ' , l / k v ^ C m I M I p \ l / f c v >

where Cm = (27i)""max{Km, Mm}. For all (p,\j/eS we obtain

\(Rm(x, D)if/)((p)\ = |il/(R'm(x, D)(p)I ^ \\ф\\р,к\\К(х> D)<P\lPM/fcv

< с т\ЩР,кМ\рл1к^

and thus for any p e ] l , oo[

(4.24) \\Rm(x,D)il/\\P'k ^ C mm \ p,k.

By letting p -> l and noting that ||<p||p>k-4Mli,k as p-> 1 one sees from (4.24) that (4.19) is valid also for p = 1. This finishes the proof. ■

R em ark 4.4. From the above proof one sees that the constant Cm in (4.19) satisfies

(4.25) Cm«(2it)-*Ct { I I ( 9 ( ';“)sup|(Z>;0)(x)|

H < r v = z — u X

x\\Dvx0 J Ll max {Ca>m}}.

|a|$iVfc + n + l

Combining (4.2), Lemma 4.1 and Theorem 4.3 we obtain

Co r o l l a r y

4.5. Suppose that L(-,-)eC °°(R "

x

R") such that

(4.26) sup \(DlD>t L)(x, £)| < C„,WM 9

xesuppfl j

Then for any j , f e N and k e K '

(4.27) I I

d

)<P\\

p

* < Cjj\\(p\\Ptk

for all (peS, where Cjty is independent of le N and p e [ l , oo[.

14 — Comment. Math. 30.2

(16)

P ro o f. Due to Lemma 4.1 we have

(4.28) I (D*DiRj,d(x, 0 | ^ for all x, £eR", le N

(since by (4.26), N u = 1 for any и ^ ft). Hence we can apply Theorem 4.3 with stf = •) |/e N } and with m = f and so

||

( R j , i ) r { x , D)<p\\Ptk

<

Cjyj'\\(p\\Pik

for all

<peS,

l e N, as desired. ■

4.3. Suppose that L(*, •) obeys (4.26). Then (4.29) \\Lj(x,D)<p\\Ptk^Cj\\<p\\pMl, (4.30)

(cf. [8], p. 49). Since the continuous formal transpose Ц (х, D): S->S exists we are able to define the continuous extensions Lf. S' -+S' of Lj(x, D) (cf. (3.21)).

Furthermore, we obtain (here f e N so that OjOf = Oj) (4.31) L , # , * u) = 0r (ij/t* Lju) + (Ки )г и

for all u e B Ptk: Let и be in BPtk and let {<p„} c S be a sequence such that

I I T h e n

IIt l / ^ L j i x , D)(pn — ij/l * L j ( x , D)(pm\\Ptk

^

H ^ I U J I L ^ x , ^ ( ^ - « р Л р д . ,

< IW oo.^C^I^-tpJlp.fc^O as n, m-MX) and

II

L j ( x ,

D)(^*(pn) - L 7.(x, ÆX^pJllp,* <

C j l l ' J / ^ i V n - f P r J l p M i

< Cj i w oo>kl||(pn-< pj|p)k-^0 as n, m->oo.

Since also

\ \ ' l ' i * < P n - ' l ' i * u \\P, k ^ Q

as n->oo,

(4.27) shows that (4.31) is valid (cf. the proof of Theorem 3.5).

Let j be in N. Choose f e N so that Ofif = 9 у Then IIOr liJ/i*Lj(x,

D)<p]\\pJl <

\\Gj'\\ukNkU i*Lj(x,

D)(p\|p>fc,

where

= (2я)-” f « щ а и т =

(2

*)-"(/)" f

R " R "

= (2я)"" f |(Т0)(^)|^к( ^ //) ^ < ll^lli,kNk < °o.

R n

We are now ready to establish

Theorem 4.6.

Suppose that L(-, •)e C 00(R"

x

R") such that

(4.32) sup I(D ^ L )(x , Д К

x e s u p p O j

(17)

Then for je N , p e [ l , oo[ and k s K '

(4.33) (L/ Д =

P ro o f. Let и be in D((Lj)'p%). Since ф ^ и е

f ] keK' B p,k

<=

B p Ml ,

(4.29) shows that ф ^ и е D({Lj)~k). Let e be a positive number. Choose tpeS such that

\W~u\\p,k < £• Then by (4.31) (4.34) \\(Lj);^i*u)-(L jyp%u\\p,k

< 11 (Lj)p,k («Ai * (w - <?)) “ (Lj)'p*k (м - <P)\\p,k

+ \\Lj(x, D )(iA ,*f/?)-L/x, D)(p\\Ptk

^ \\6f &А1* е д > - (p) — {Lj)'*k( u - Ф)]I U

+ ||(R;>0/(u-<p)||p,k + ||LJ.(x, Д)(«Аг*ф )-Т ,(х, D)ç>||p,fc

^ P IIi.*wJI«A,* C(L/Pî(t t - <p)] - (L / pJ(ii - <p)IU 4” Cjtf \\u ф||р,к "b ^jll^AI * Ф «Pllp.kfci •

Choosing /0 e N so that for I ^ /0

ll«Ai*[(TjK(w-<p)]-(LJ.);#fc(u-(p)||p>fc < e, ||«A/*<P-<PlUit1 < e, one finds that for l ^ l0

11 (Lj);,k(«Ai *u) - (Lf^uWp' k ^ \ \ 9 \ \ l t k N k s + C j j £ + CjE

and so (L^p^OAi*!*)-*^)^ and фг*и-^и in

B p k.

This proves that

u e D ( ( Lj ) ~ k)

and (Lj)~ku - (Lj)’p*ku, as required. ■

R em ark 4.7. As the proof of Theorem 4.6 shows, (4.35) 11 (Ь])~к(ф1 * и) - (L;);#k и I Ip,к + 1 |*A,* и - u\Ip,fc -> 0 as Z-MX> (for any ueD((Lj)'*k)).

Co r o l l a r y

4.8. Suppose that L(-,-)eC ® (G x R") such that for any

compact subset K a G and (a, /?)eNo"

(4.36) sup\ (DxD^L)(x, 0 | ^ Св.,.* М 0 for ^eR".

x e K

Then (for p e [ l , oo[ and keK ')

(4.37) L~k(G) = L'p#k(G).

P ro o f. Recall that L'(x, Z)): Co (G)->S exists (cf. Section 2.3). Let ф be in C

q

(G) and let Ьф(х,

D) : =

фЦх,

D).

Choose ; e N so that в-ф = ф. Then by (4.36) one has

sup K ^ D f L ^ x , 0 | < C e.,.,M 0

xeR"

and so by (4.35)

(18)

(4.38) 11((L*),)~k( ф ^ и ) - ((L^)j)'*kи |\Ptk + 11фг* и - u\\Pyk -► 0 as l

—*■

со. One finds that

= U(<LM X’

d

)<

p

) = «(£*(*. D)Wjip))

= «(!/(*, D)(iA^ç>)) = (L&(G)

k

)(<M = фир\( в ) и and

( ( ^ ) ; , к(Ф1*и) = ({Ьф)])'р*к{Ф1*и) =

ф

ь'*к(С)(ф t*u) = 11/ь~к(в)(ф1*и).

Thus, by (4.38), Lp>fc(G)(^I*M)->Lpjt(G)M in jBp,k(G) and iJ/t* u ^ u in BPyk, which completes the proof, ш

5. On the equality L~y = Lp*p'.

5.1. Finally, we shall show a sufficient condition for the identity L~ p' = Lp*p'. Our considerations are based on the following lemma:

Lemma

5.1. Suppose that

L ( - ,

•)

g

C 00(R"

x

R") such that (5.1) sup\(D%D^L)(x, 0\ < Q A ( £ ) for all £eR".

x e R "

Then the symbols

Rj,i{%,

•) defined by (3.4) obey

(5.2) sup \{Da xD^Rjj){x, £)| < CM for all j, le N.

x.^ e R ”

P ro o f. A. In view of (4.7) we obtain (5.3) (DlDlRjM x, ()

0 |y| = l R "

= } I I ( f)(-o M ! (&'+сж + ‘ ч т 1 > ;+‘ о { - 'Ц ( - ,

(»(>,)

0 |y| = 1 R"

x el{x,n)dr}dt.

We consider the integral

/ :=

h 4 „ ( x , ( ,

f):= J

( d ^ ’ e i)(i + i ^ F ( ( D y D l - - L ])(-, ф ф « ^ .

R ”

One sees that

1 = j S (dy*vei)(i + tti)(Dl*‘D l-'‘LJ)(y, i ) ^ - ^ d r , d y ,

R " R "

where

J (а' + |’0,)({ + г1/)№ Г ‘ О Г -Lj)(y, dq

(19)

= f ( i / ( i + |x - y |2"))(e'+”ei« + t ),)(Dj;+* B f-’LJ))(y> о

R "

x ( l + ( £

= J (i/d + |x -y |2"))-((i + ( - t 2) " ( i

d]f)d->+°e,)(t+tri)

R " 7 = 1

x(D l+aD l - vLJ)(y, Q e^-w'dri and so

(5.4) ; = J (i + ( - t 2)" (£ д1у)д’**в,в+щ)

R n 7 = 1

x (J ( W l + \ x - y \ 2a))(Dl+°Dllt - ’Ll)(y, i)e«*-’-»dy)dri.

R "

B. For any

|t|

^ n + 2 one has

(5.5) |, ' j (1/(1 + |x - y |2"))(£>ï+“£>f—’Z,i)(y> i)e‘^ - ^ d y \

R "

« J |0 '((l/(l + |x - y |2"))BJ+,,D f-,’Lj)(y, ()\dy

R ”

< Z (У J |в;(1/(1+|х-у|2”))||(В5+*+*-»в|-'^)(у,

W y

w ^ i R n

= I (У J С„(1/(1 + |х-у|2”))С;+„+,_^_Д 1(07у=:См,„,„/с1(«).

w < T R "

where we have used the fact that

|D ;(l/(l + |x - y | 2 "))|«C „(l/(l + |x - y |2”)) for x . y e R ”

and that

\ W D f Lj){y, |)| ^ Z e')|(D » 6 jm D Ï - D l'L ) ( y , 01 u^a'

=S I (;')(1//I“l)l№“0)(y//)I I W - “D(L)(y, 01

^ X (Ï) sup

I

(Я“

в)(у)\CX' - иЖк± (£).

u ^ a y e R n

Due to (5.4) and (5.5) one gets

(5.6) |/| =s cM,7,„ J(i + (- t 2)"(Z a?)")(S'+“<w+n,)M0k

(n + 2)(fl}df] •

R " 7 = 1

C. We have the estimate (cf. (4.6))

|(i + ( - t 2)"(Z <5?)")(<з'+ч ) « + » / ) I

j=l

(20)

^ Kd’^ M É + t i r t i + K d a?)”a '+-e,)«+ t7)|

J= 1

^ Cy + 0k - \ y + v\(Ç + tri) + CytV'ttk-\y + vi-2n(Ç + tri) ^ C y ^ k - ^ + trj)

and so by (5.6) one gets

1Л < CatPt7tV J k - 1(^ + tti)k-{n + 2)(t])k1^)dri.

R "

As in the proof of Lemma 4.1, Parts B1-B 2, one sees that |/| ^ and so

№ D iR j,d (x , £ )K J z tt/vO I (01 l\dt ^ CM ,

0 |y| = l v ^ p

which finishes the proof. ■

From Lemma 5.1 we obtain (recall that p 'e ] 1, oo], 1/p+ l/p' = 1 and

* o = l) \

Theorem

5.2. Suppose that L (-,

• )

e C00(R" x R") obeys the estimate (5.1) and let ■) be defined by (4.3). Then for p e [ l , 2]

(5.7) \\Rjfix, D)(p\\P' ^ C\\(p\\pM for all (peS and j, le N.

P ro o f. A. Let

Tjft:=

R jfix, D)oF~l . Then by (5.2) we find that WT

j

M I

loo

= sup

\(Tjj(p){x)\

= sup|[Rj-,(x, Z))((F-1 <p))](x)|

JceR” xeR"

= (2n)_"su p |J Rjj(x, Q F (F -1 </>)(№**>di\

x R "

^ C1

J

\<p(£)\d£ = c t l\(p\\Ll

for all

(peS.

R "

B. Let Ф(х, £) = ф(х, Ç) = 1. Then by (5.8) one sees that Rjfi-, - )eS%’^

and that Pa’,p(RjA*» ')) ^ Cx,p- Hence by Theorem 2.1

\\TjM l2 = WRjAx, W » ||La ^

C 2||F-VII

l

2

= C2\\cp\\l2

for all (peS.

In virtue of the Riesz-Thorin interpolation theorem (cf. [2]) one has

\\Tj,i<p\\P’ < C l~ 2lp'C llp,\\(p\\p ^ ( C 1 + C2)\\(p\\p for all <peS and thus

II

R jA x ,

D)<p\\P' = \\TjAF(p)\\p ^ (Cl + C2)\\F(p\\p = (C1 + C2)||<3£>||p,Jk0,

as desired. ■

(21)

Finally we establish

Th e o r e m

5.3. Suppose that L(-, •)eC Q0(R"xR") so that (5.1) holds. Let

pe~\ 1, 2[. Then

P ro o f. Let и be in D(L'*P'). Similarly to the proof of Theorem 5.2 we observe that \\L(x, D)(p\\p> ^ C\\(p\\Ptkl and then iJ/^ueDiL'*^). Furthermore, for any j, le N (cf. (4.31))

Since ||

i

^

j

* w — w||p' — ► 0 as l-> oo, the assertion follows by the same kind of argument as in the proof of Theorem 4.6. ■

[1] R. B e a ls and C. F e ffe r m a n , Spatially inhomogeneous pseudo-differential operators I, Comm.

Pure Appl. Math. 27 (1974), 1-24.

[2] J. B erg h and J. L ô fs tr ô m , Interpolation Spaces, Grundlehren Math. Wiss. 223, Springer, Berlin 1976.

[3] K. O. F r ie d r ic h s , The identity of weak and strong extensions of differential operators, Trans.

Amer. Math. Soc. 55 (1944), 132-155.

[4] P. H e ss, Über wesentliche Maximalitat gleichmassig stark elliptischer Operatoren in L2(R”), Math. Z. 107 (1968), 67-70.

[5] L. H o r m a n d e r , Linear Partial Differential Operators, Grundlehren Math. Wiss. 116, Springer, Berlin 1963.

[6] M. S c h e c h te r , Spectra o f Partial Differential Operators, North-Holland Ser. Appl. Math.

Mech., North-Holland, Amsterdam-London 1971.

[7] J. T e r v o , On maximal and minimal realizations o f linear pseudo-differential operators, Comment. Phys.-Math. 75 (1986).

[8] —, On Bpk-boundedness and compactness o f linear pseudo-differential operators, Z. Anal.

Anwendungen 7 (1) (1988), 41-56.

[9] R. A. W ed er, Spectral analysis of pseudo-differential operators, J. Funct. Anal. 20 (1975), 319-337.

[10] K. Y o s id a , Functional Analysis, Grundlehren Math. Wiss. 123, Springer, Berlin 1974.

(5.8)

References

d e p a r t m e n t o f m a t h e m a t i c s, u n i v e r s it y O F JYVÂSKYLÂ SEMINAARINKATU 15, SF-40100 JYVÂSKYLÂ, FINLAND

Cytaty

Powiązane dokumenty

efficients of the systems satisfy (rather weak) additional smoothness properties, then at either of the branch points an asymptotic expan ­ sion (exists and) can be

If a Banach space X contains an asymp- totically isometric copy of c 0 , then X fails the fixed point property for non- expansive (and even contractive) mappings on bounded, closed

In this paper we study local derivations of Ore extensions in the case when R is the polynomial ring k[x] in one variable over a field k of characteristic zero... Using the

We present a description of all derivations of Ore extensions of the form R[t, d], where R is a polynomial ring in one variable over a field of characteristic zero..

This possibility to bypass media stereotypes, as well as traditional canonical and bureaucratic‑nationalistic conceptions of American Studies, is what permits us to

(They called ideals as deductive systems.) In this paper, we give some characterizations of maximal ideals in

In particular, we showed that implicative ideals are precisely the Boolean ideals and proved that an ideal is maximal if and only if the quotient BL-algebra is a simple

Abstract: Self-adjoint extensions are constructed for a family of boundary value problems in domains with a thin ligament and an asymptotic analysis of a L q -continuous functional