Physica 124A (1984) 311-316 North-Holland, Amsterdam
31 l
BOUND STATES IN N-BODY QUANTUM SYSTEMS
R. FROESE Department of Mathematics, University of Virginia, Charlottesville, VA 22903 I would like to talk about some results on the large Ixl solutions,
behavior of L2
~(x) , to the Schrodinger equation
H~ = E~
where H is an N-body quantum Hamiltonian. bounds for ~.
I w i l l concentrate on finding lower
The connection between the permissible rates of decay for
¢ and
the spectrum of H w i l l show that for a large class of H positive eigenvalues do not occur. These results were obtained j o i n t l y with Ira Herbst4'5 and are closely r e l a t ed to results which we obtained earlier in collaboration with Maria and Thomas Hoffmann-Ostenhof.6 The class of Schr6dinger operators under consideration is actually somewhat }N larger than the class of N-body Hamiltonians arising in physics. Let {~i i=l be a set of orthogonal projections onto the subspaces {xi}Ni=l of ~Rn . Suppose vi(Y) is a real valued function of y ~ Xi which "goes to zero" as [yl ÷ ~ in a sense to be made precise later. in L2(• n) given by
Then H is the operator acting
N
H : -L + ~ v i ( ~ i ( x ) ) i=l with domain D(H) = D(z~). Two sets of conditions on V w i l l be used in what follows: Let Ai and y denote the Laplacian and variable corresponding to Xi .
Then the f i r s t set
o f conditions is
(a)
v i ( - ~ i + I ) -I (-mi + l ) - l y
is compact on L2(Xi ) • v Vi(-A i + l ) -I
is compact on L2(Xi ) .
Presented at the V l l t h INTERNATIONAL CONGRESS ON MATHEMATICAL PHYSICS, Boulder, Colorado, 1983.
312
R. FROESE
Let ~. = dim (Xi) and Pi = max {2, ~i - I } . 1 is
(b)
Then the second set o f c o n d i t i o n s
v i e LPi(xi ) + L~(Xi ) vi =
v1(l . ) + vi(2)
and f o r each
with
c > 0
(I + I Y l ) v i ( 1 )
y • vvi(2)
e k
Pi
<__c(-zl) + b E
(X i ) + L (X i ) f o r some
b
In the above conditions the gradients on v i are taken in the distribution sense; i t is not assumed that classical derivatives exist.
A bounded measurable func-
tion with o ( I x l - l ) decay at i n f i n i t y satisfies both (a) and (b).
On the other
hand i f classical derivatives exist and have suitable decay, these conditions are also satisfied by long range potentials, for example vi(Y) = lyl -~ holds are defined as follows.
Thres-
For any proper subspace Y C R n of the form
Y = span {Xi} i cJ consider the operator Hy acting in L2(y) given by
Hy = -Ay +
~ vi ( ~ i (y)) l xi~Y
where ~y is the Laplacian for Y.
The set of thresholds is defined to be all
eigenvalues of such operators and is denoted %. The spectrum of such an N-body operator looks like:
v
;
v .. 0
X
ei genval ues
@
thresholds
.,,
continuous spectrum
I w i l l describe in a moment how to prove that there are no positive thresholds or eigenvalues. Let H be as above and suppose ~ is an L2-solution to the Schr~dinger equation H~ = E~. We ask the question: when is exp(~Ixl) ~ ~ L2(~n ) ? Consider increasing ~ in exp(~Ixl)~ starting from ~ = O.
For ~ = 0 we have
BOUND STATES IN N-BODY QUANTUM SYSTEMS
exp(mlxl)~ = ~ ~ L2(~ n ) by hypothesis. some p o i n t mO~exp(mlxl)~
313
I t might happen t h a t as we move past
changes from being in L 2 ( ~ n )
to not being in L2(~n).
Theorem 1 says t h a t i f V s a t i s f i e s (a) t h i s can only happen i f m02 + E is a threshold.
So i f we define
m(~) : sup{m 2 + E :m ~ 0 and exp(mIx I) @ ~ L 2 ( R n ) } then we have Theorem I :
If V satisfies
(a) then
T(~) c Z U { + ~} This theorem gives an L2 upper bound, namely i f T O is the f i r s t above E then e x p ( ~ I x l ) ~E L 2 ( ~ n )
for all a < ~
the possibility that T(@) = +~ , i.e. that exp(~Ixl) ~ Theorem 2: I f V s a t i s f i e s
.
L2(~ n) for all
~.
(b) then
Thus i f V s a t i s f i e s both (a) and (b) we know t h a t T(~) c I . Corollary I.
threshold at or
Theorem 2 e l i m i n a t e s
If V satisfies
NOW we can prove
(a) and (b) then H has no p o s i t i v e eigenvalues
or thresholds. The proof o f t h i s c o r o l l a r y is by i n d u c t i o n on the set of subspaces Y o f the form Y = span { X i } i ~ j
.
The induction step goes as f o l l o w s :
each W of t h i s form s t r i c t l y
thresholds. Then, by d e f i n i t i o n , and 2 ~(~) f o r any e i g e n f u n c t i o n not p o s i t i v e .
Suppose f o r
contained in YoHw has no p o s i t i v e eigenvalues o f Hy has no p o s i t i v e thresholds.
By Theorems 1
@ with eigenvalue E is a threshold and thus
But from the d e f i n i t i o n o f T(~) i t is c l e a r t h a t E ~ T ( ~ )
.
Thus E ~ 0 . Now we can also prove C o r o l l a r y 2: I f V s a t i s f i e s (a) and (b) then
exp(~Ixl) ~ # L2( ~Rn) for ~ > ~ E To prove t h i s c o r o l l a r y note t h a t i f f o r some ~
with 2
+ E > O.
i t is f a l s e we have e x p ( ~ I x l ) @ ~ L2 (IRn)
Thus T(@) >__2
+ E > 0 , which is a c o n t r a d i c t i o n
since T(~) is a threshold. C o r o l l a r y 2 can be viewed as a crude s o r t o f lower bound and the question
314
R. FROESE
n a t u r a l l y a r i s e s : can b e t t e r lower bounds be obtained?
If
~ is the ground
s t a t e then d e t a i l e d p o i n t w i s e l o w e r bounds have been obtained by Carmona and Simon. 2
If
9
is not the ground s t a t e , then the p o t e n t i a l l y
complicated na-
t u r e o f the set where ~(x) = 0 i n d i c a t e s t h a t p o i n t w i s e lower bounds f o r I~(x)l Merigot,
would be very d i f f i c u l t to o b t a i n . I n s t e a d , f o l l o w i n g Bardos and 1 we consider the average o f ~ over spheres. Let S(R) be the sphere
o f radius R and d e f i n e I~l R by
I~IR 2 = ~
I-~12 ds
S(R) Then we have the f o l l o w i n g theorem, which gives exponential upper and lower bounds f o r
I~IR when E l i e s below the e s s e n t i a l spectrum o f H.
Theorem 3. lira R-I
R-~o:
If V satisfies
(a) and (b) and E < z = i n f ~ess(H) then
log(IC, IR) = -%
where a 0 c Z . When E does not n e c e s s a r i l y l i e
below the infimum o f the e s s e n t i a l spectrum
then the theorem remains t r u e w i t h I~IR replaced by the L2 norm o f
~ over a
s p h e r i c a l shell w i t h r a d i u s R and thickness 6(R) where ~(R) does not decrease
too q u i c k l y .
For a precise statement and p r o o f of these r e s u l t s ,
see r e f s . 5
and 7. To conclude t h i s t a l k I w i l l o f Theorem l .
very roughly sketch the main step o f the p r o o f
The many o m i t t e d d e t a i l s can be found in r e f . 3.
generator of dilations,
Let A be the
i.e.
,~ = ½ (X o D + D o x) where D is the g r a d i e n t o p e r a t o r . [H,A] = - 2 & The f i r s t
Then the commutator o f H w i t h A is given by
- x • v V
i n g r e d i e n t in the p r o o f is the Mourre estimate 4'8 f i r s t
body H by Perry, S i g a l , and Simon. then there e x i s t s an open i n t e r v a l
E(1) [H,A] E(1) ~ c
I t says t h a t i f
•
proved f o r N-
is a non-threshold p o i n t
I c o n t a i n i n g ~ such t h a t
E(1) + K
where E(1) i s the s p e c t r a l p r o j e c t i o n f o r H, c is p o s i t i v e and K is compact.
BOUND STATES IN N-BODY QUANTUM SYSTEMS
The second ingredient is some equations satisfied by @ ~ exp(F) .~ F A computation shows that
315
where
F is an increasing function of Ixl . H~ F : (E + (VF)2) ~F - B~F
where B is an antisymmetric f i r s t order operator.
Now using the formal relation
: 0 with A = exp(F)A exp(F) we obtain <~F,EH,A] ~F> : -4
II
g½A~FII2 + <~F' G ~F)
where g is defined by vF = xg and G is a function which w i l l be small for large [xl
for the choice of F that we make. What we want to show in this step of the proof is that T(~) cannot be a non-
threshold point greater than E.
Assumethe contrary.
with 2 + E < T(O) < (~ + y)2 + E where ~ and y
Then we can find ~ > 0
are such that the Mourre est-
imate holds for some I containing [ 2 + E, (~ + ¥)2 + E]. exp(~Ixl) 0 ~ L2(Rn) while exp((~ + y ) I x l ) ~ L 2 ( R n ) .
Furthermore To interpolate we de-
fine F~ by exp(Fx) = exp(~[xl) (l + y l x l / x )~ Then exp (Fx) ~ ~ L2(~ n) for each
but
l im llexp(Fh)VII= ~ , Define
Then i t is easily seen that for a bounded set f i l l 2 dnx ~ 0 as
~~
so that ~X ~ O . The next step is to show that IIB~II ÷0. We omit the proof of t h i s . Now using this and the fact that (VF~)2 is approximately 2 we have for large 1 HI~ = (E + c=2) I~
316
R. FROESE
From t h i s i t f o l l o w s t h a t , roughly speaking, E(1).
Here we are ignoring an e r r o r term.
~X converges i n t o the range of From the Mourre estimate we have
<~E), E(1)[H,A] E(1) I)> _> c liE(1) ~)~II2 + ,,K !i)> In the l i m i t
~ ÷ = we can more or less ignore the E(1), and the l a s t term van-
ishes since K is compact and ~x ~ O.
So up to a small e r r o r which goes to zero
with Y, lim inf <~,
[H,A] I X )
>
c
On the other hand using the f a c t t h a t -411 g½A~ll 2 < 0 we obtain
(~),[H,A] I~) <__( I x , G IX) Now f o r l a r g e Ixl G is small on the order of y . I x l regions as I x l + =
Since T~
is moving to large
we have, up to an O(y) e r r o r term
lim sup ~I), [H,A] ~>)
<
0
This c o n t r a d i c t s the previous equation ( f o r small enough y ) and completes the proof. REFERENCES I ) C Bardos and M Merigot, Proc. Roy. Soc. Edin. 76A, 323-344 (1977). 2) R Carmona, B Simon, Comm. Math. Phys. 80 (1981) 59. 3) R Froese and I Herbst, Comm. Math. Phys. 87 (1982) 429-447. 4) R Froese and I Herbst, Duke Math. J. 49 (1982) 1075-1085. 5) R Froese and I Herbst, Exponential Lower Bounds f o r Solutions of the Schr~dinger Equation: Lower Bounds on the Shperical Average, p r e p r i n t . 6) R Froese and I Herbst, M Hoffman-Ostenhof and T Hoffman-Ostenhof, Comm. Math Phys. 87 (1982) 265-286. 7) R Froese, Lower Bounds for Solutions of the SchrSdinger Equation, U. o f V i r g i n i a D i s s e r t a t i o n , August 1983. 8) P Perry, I M S i g a l , and B Simon, Ann. Math. 114 (1981) 519-567.