Free field operators
Citation for published version (APA):
Eijndhoven, van, S. J. L., & Graaf, de, J. (1985). Free field operators. (Eindhoven University of Technology : Dept of Mathematics : memorandum; Vol. 8505). Technische Hogeschool Eindhoven.
Document status and date: Published: 01/01/1985
Document Version:
Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)
Please check the document version of this publication:
• A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website.
• The final author version and the galley proof are versions of the publication after peer review.
• The final published version features the final layout of the paper including the volume, issue and page numbers.
Link to publication
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal.
If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:
www.tue.nl/taverne Take down policy
If you believe that this document breaches copyright please contact us at: openaccess@tue.nl
providing details and we will investigate your claim.
Department of Mathematics and Computing Science
Memorandum 1985-0~
January 1985
FREE FIELD OPERATORS
by
S.J.L. van Eijndhoven
J. de Graaf
Eindhoven University of Technology
Department of Mathematics and Computing Science
PO Box 513, Eindhoven
Summary
by
S.J.L. van Eijndhoven and J. de Graaf
In the present paper we give a mathematical rigorization and interpretation
of the canonical (anti-) commutation relations which occur in the theory
of the quantized free field. In our set-up, with respect to the physics,
we feel inspired by the paper [Ro] of B. Robertson. His paper covers all
essential elements of the subject under consideration. The main part of
[Ro] will be mathematically justified and even generalized. In this
connec-tion we menconnec-tion also the monograph [Sh] of A.S. Shvarz. With respect to
the involved mathematics the papers [Co] of J.M. Cook and [KMP] of P.
Kristensen et al. have been a source of inspiration.
Our set-up is closely related to two of our earlier papers [EG 1-2] in which
we presented a mathematical interpretation of Dirac's formalism. This
inter-pretation is based on two new mathematical concepts: the concept of Dirac
basis and the concept of bracket. The bracket is no longer regarded as a
"number" but as an analytic function on the open right half of the complex
plane. As in [EG 1-2], we use the theory of generalized functions which has
Quantum theory and distribution theory seem to be dissolubly connected.
One might wonder whether the choice of the distribution theory plays an
essential role in a mathematical interpretation of Dirac's formalism and
related physical concepts. The answer is threefold: There are strong
indi-cations that similar interpretations can be derived in a very wide class of
distribution theories (which include Schwarz's theory of tempered
distri-butions). This class has been described in our papers [EK] and [EGK].
On the other hand, the distribution theory employed here, introduces
gene-ralized functions in a way very close to the physical intuitive view on
im-proper functions. Moreover, it seems natural to adapt the used distribution
theory to the concrete quantum mechanical system under consideration.
O. Introduction
The material of this paper is presented in the abstract setting of
functi-onal analysis. Thus we expect to be able to avoid notatifuncti-onal problems and,
more importantly, to clarify all mathematical concepts in the best possible
way.
We start with a separable Hilbert space X in which we choose a fixed
ortho-normal basis (Vj)jE~' Let X ®a X denote an algebraic tensor product of X.
Then the bilinear mapping ® sends each pair (vk,V£) to v
k ® v£' In a natural way the set {v
k ® v£
I
k,£ E ~} becomes an orthonormal basis for a Hilbert space X(2) which is a completion of X ®a X. X(2) may be called the two-foldHilbert tensor product of X. Proceeding inductively, the k-fold Hilbert
tensor product X(k) of X can be defined. The Fock space F is their countable direct sum,
(0,1) F
=
$ X(k)k=O
where we take X(O) = t and X(l) = X.
The Hilbert space
F
is the middle component of the Gelfand tripleSF,H
c F cTF,H'
HereSF,H
is called the Fock analyticity space andTF,H
the Fock trajectory space. Further, H denotes a well-known positiveself-adjoint unbounded operator in F,
The one-particle state is represented by the trajectory space Tx,A' where
A is the positive self-adjoint operator in X satisfying
(0.2) Av.=jv. ,
For each ~ E Tx A we define the annihilation operator
,
a(~) and the crea-tion operator e(~). The operator a(~) is a continuous linear mapping fromSf,H into Sf,H which 1s not extendible to Tf,H' 1n general. The operator
e(~) 1s a continuous linear mapping from Tf,H into Tf,H; its restriction to Sf,H does not map Sf,H into Sf,H'
Let (G~)tEM denote a Dirac basis in Tx,A' We present a mathematical inter-pretation of the following heuristic formulas
(0.3) a(~) =
)
<\1>I
Gt
>
a(G~) d)l~ e(\1» =J
<Gt
I
~> e(G~) dtl~ • M1 (d2 2)
,
We note that for X = L
2
(m),
A =2'
- 2
dx + x + 1 and Gt
=
0t' ~ €m,
the Formulas (0.3) are the usual relations for the so-called free field
operators.
Next we consider the symmetric Fock space f(+) and the antisymmetric Fock
space f(-). Both f(+) and f(-) are closed subspaces of f. If p(+} and p(-)
denote the corresponding projections, then we define
(0.4)
a
(±) (~) = p(±} a(\p) pet) ,We give a rigorous interpretation of the following heuristic (anti-)
commu-tat ion relations
(0.5)
[a(±) (\P) ,
e
(±) (IJI)]=
0 etc.+
We also devote attention to the concept of second quantization. For a
linear operator
V
fromTx,A
intoTx,A
we rigorize the following expression for the second quantization ofV
Contents
Q±(V)
=
If
dll~
dlln C (±) (Gl;) <Gt
I
VI
Gn>
ct (±) (Gn)M2
1. The Gelfand triple Sy,B e Y e Ty,B'
2. Some results on continuous linear mappings.
3. The concepts: Dirac basis and bracket •
4. The Fock space •
5. Introducing the machinery.
6. Annihilation and creation operators.
7. The symmetric and anti-symmetric Fock space.
8. A mathematical interpretation of the field operator formalism in
quantum mechanics •
1. The Gelfand triple Sy B
.
e Y e Ty B,
Let Y be an infinite dimensional separable Hilbert space and let (en)nEm
be an orthonormal basis in Y. Each element y E y is represented by the
co
R.2-sequence «y,en»nE:fi' We extend Y by considering formal series
I
n=lw e
n n
where the sequence (Wn)nE:fi does not necessarily belong to t2' Therefore
we introduce a sequence (An)nE:fi of positive numbers with the property
(1.1) lim e
-A
n t=
0 , n-+<»(1.2)
I
e n -A t1 12
w<"".
n=l n
Because of (1.1) the series
I
n=l the operator8
by 00 (1.3) D(B) ={fEylI
n=l and 8f=
I
n=l A (f,e)e , n n n w en n may be divergent in X. We introduce
f E D(8) .
Then
B
is a nonnegative self-adjoint operator inY.
Because of Condition -t8(1.1) the semi group (e )t>O consists of compact operators on Y. We note
that the parameter t, we use in this paper, must not be regarded as the
time parameter.
To the sequence (wn)nE~ defined in (1.2) we link the mapping F from (0,"")
into Y defined by 00
-x
t e n w e n n t > 0 . F(t)=
L
n=lObserve that F(t + 1') = e -1'B F(t), t,1' > O. Conversely, for a mapping G from
(0,00) into Y which satisfies G(t + 1') = e -1'8 G(t) for all t, l' > 0, the
se-A t -quence (Yn)nE~' defined by Y
n
=
e n (en,G(t», has property (1.2). We have 00 G(t)=
I
n=1 -Ant e y e . n n(1.6) Definition.
The trajectory space
Ty,B
is the vector space of all mappings F with the property"'t>O "'PO F(t + "C)
=
e-"CB F(t) . The elements ofTy,B
are called trajectories.Besides the trajectory space
Ty,B
we introduce the analyticity spaceSy,B'
(1.7) Definition.
The analyticity space
Sy,B
is defined byu
t>O
-tB
I
=
{e y y E y, t > O} •We mention the following important relations
(1.8)
'"
FETy
B'"
t>O .•
F(t)
E Sy,B
On the space
Ty B
,
we impose the locally convex topology generated by the seminorms F 1+ IIF(t)lI, t > O.Ty,B
is a Fr@chet space with this topology.On
Sy,B
we impose the inductive limit topology brought about by the spaces-tB
e (Y). We mention that both
Sy,B
andTy,B
are inductive and projectivelimits of Hilbert spaces. The spaces
Sy,B
andTy,B
are bornological, barreled and Montel. If in addition the sequence (An)nEfi satisfies(1.9)
I
e n -A t < <X>n=l
(1.10) Definition
Let F E
Ty,B
and let g ESy,B'
Then the complex number <g,F> is defined by.B
<g ,F>
=
(e g ,F(.»where • > 0 has to be taken so small that e·B g E Y. The definition of <g,F> does not depend on the choice of • > O.
With the pairing <-,.>,
Sy,B
andTy,B
can be seen as each other's strong duals.If Y
=
L2(M,~) where M is any measure space with a-finite measure ~, then for any nonnegative self-adjoint operatorB
iny,
the spaceTy,B
consists of generalized functions onM.
SO the theory of analyticity and trajectory spaces presents a functional analytic model for a particular type ofdis-tribution theories.
For more details and examples of this theory we refer to [G], [EGm] and
[EG 3-4].
2. Some results on continuous linear mappings
For the purposes of this paper some knowledge of operator theory for
ana-lyticity and trajectory spaces is required. We present here the necessary
material.
Let Z be a separable Hilbert space with orthonormal basis (gn)nEm' Further,
let the sequence (Un)nEm of nonnegative numbers satisfy Condition (1.1),
i.e. lim
e-~nt
= O. Accordingly we define the operatorC
in Z byn-+<»
00
D (C)
=
{z E ZI
I
u!
I
(z, gn)12
< <X>}and
QO
Cz =
I
l1n (z,gn) gn 'n=l
z E D(C) •
Five types of linear mappings occur: the continuous linear mappings from
Sy,B
into Tz C' from Ty,B into,
Sz,C'
from Sy,
B
intoSz
.
C and from Ty,
B
into Tz,C and, also, the extendible linear mappings, 1.e. the mappings from Sy,
B
intoSz,C
which can be extended to a continuous linear mapping from Ty,B into Tz,C' These five types admit the following characterizations.(2.1) Theorem.
A linear mapping
V : Sy,B
+ Tz,C is continuous iff-tC -tB
V t>O : e V e Y + Z is bounded .
A linear mapping
M :
Ty,B +Sz.C
is continuous iff ,CM .B
3 e e Y + Z is bounded .
• >0
A linear mapping L : Sy
,
B +Sz
,
C is continuous iffV 3
t>O .>0 e'C L e -tB y + Z is bounded .
A linear mapping R : Ty
,
B
+ Tz,
C is continuous iff -tC R.B
e e : Y + Z is bounded .
A linear mapping E :
Sy,B
+Sz,C
is extendible iff E* mapsSz,C
intoSy,B'
Its extension t : Ty,B + Tz,C is defined by<f,tG>
=
<E*f,G> , Both E andE
are continuous.""
coI
-A tI
e-llnt < "" for all t > O. Then bothSy,B
Suppose that e n < co and
n=1 n=1
and Sz Care nuclear. In this case the above mentioned types of continuous
,
linear mappings have characterizations based on infinite matrices. NowTheorem (2.1> can be replaced by
(2.2> Theorem.
Let both Sy,B and
Sz,C
be nuclear.A linear mapping
V :
SyB
+ Tz C is continuous iff
,
,
sup (e-tllm e-tAn I<g
,Ve
>1> < "" .m n
n,mElN
A linear mapping M : Ty
,
B
+ Sz C is continuous iff,
sup (eL)..lm e'rAn I (Me ,g > I> < "" •n m
n ,mElN
A linear mapping L : S + S is continuous iff
y,B z,C
V 3
t>O T>O sup (e- tAn e Tllm I(Ve,g >1> n m < c o .
n,mElN
A linear mapping R : Ty
B
+ Tz C is continuous iffJ ,
V 3
t>O T>O sup (e -t II m e n <g, R e > > T A I m n 1
n,mElN
< co •
A linear mapping
E
is extendible iff-tll 'rA -t>.. Tll
I
I
sup (e m e n + e n e m
(E
en,gm> > n,mElN< co •
We denote the space of extendible linear mappings by E(Sy,B'SZ,C>' For
the other spaces of continuous linear mapping we employ the usual notations
L(Sy,B,TZ,C)' L(Ty,B'Sz,C)' L(Sy,B'SZ,C) and L(Ty,B,Tz,C)' In our
mono-graph [EGm] we have extensively studied the algebras L(Sy,B)' L(Ty,B)
and E(Sy
,
B); and in particular their algebraic and topological structure.3. The concepts: Dirac basis and bracket
In [EG1] the notion of Dirac basis has been introduced. Each orthonormal
basis is a Dirac basis. In general, a Dirac basis may consist of an
un-countable number of elements.
Let Ty B be a nuclear trajectory space. The definition of Dirac basis is
,
the following.(3.1) Definition.
Let M be a measure space with nonnegative a-finite measure ~. Let
G : a ~ Ga, a E
M,
be a mapping from M in Ty,B' Suppose the functionsa~ <ek,G
a> are ~-measurable and satisfy
I
<ek,Ga> <et,Ga>d~a
=
°kt • k,t Eli.M
Then the quadruple (M,~,G,Ty,B) is called a Dirac basis. Instead of
(M,~,GJTy,B) we mostly write (Ga)aEM'
Remark.
It is appropriate to reserve the notion of Dirac bases for equivalence
classes. However, we will stick to a fixed "canonical" representant. Cf.
In [EG1] we have proved the following expansion theorem.
(3.2) Theorem.
Let f €
SY,B'
ThenTB
<e f,G > G (.) dJ.l
ex. ex. ex.
.B
where • > 0 has to be chosen so small that e f € Sy, B' The integral ex-pression does not depend on •• Further, the integral
H
I
I <e TB f,G ex. > I IIG ex. (T) II dJ.l ex. exists.Next we give an outline of our paper [EG1].
For the ket space we take a nuclear trajectory space Ty
,
B' The choice of Y andB
may depend on the quantum mechanical system under consideration. In accordance with Diracfs formalism we denote the kets by IG>. SO anyelement IG> can be written as
where the sequence (w
k) satisfies
and where Ie?, k € 1N, denotes the function
t > 0 •
con-tinuous linear fUnctional
y
on Y by y(x)=
(x,y) , xE
Y •By the Riesz representation theoren, the mapping y» y is an antilinear
norm-preserving isomorphism from Y into
Y.
The norm in the Banach spaceY
originates from the inner product (.,-) defined by
So
Y
is a Hilbert space. InY
we define the nonnegative self-adjoint ope-ratorB
byfollowed by the natural linear and self-adjoint extension. Then the
trajec-tory space
Ty,S
is well-defined. Each element ofTy,S
is in 1-1 correspondence with an element ofTy
,
S' viz.co co
Ty,S
3a
L
-AL
GE
Ty,s
= wkek ++ w k ek =.
k=1 k=1The correspondence G ++
a
is antilinear, i.e.Now for the bra space we take the space
Ty
,
S'
Its elements are denoted by<GI.
It is clear that each bra<GI
is in 1-1 correspondence with the ketIG>.
Moreover, this correspondence is antilinear. We note that each bra<GI
can be written aswhere the sequence (wk)kE:tf satisfies
:;~
(I
wkl e -Akt) < "' and where <ek1
denotes the function
t > 0 .
To the elements f of Sy B
,
there correspond the so-called test kets and test bras. The functions <fl and 1 f> from (0,"') into X can be extended tofunc-tions from (_~,w) into X for some ~ > 0 dependent on the choice of f.
The bracket <-1-> is defined as follows.
(3.3) Definition.
The bracket of <FI and IG> denoted by <F I G> is the complex valued function
on (O,w) defined by
<F
I
G> t 1+ <FI (I
G> (t» •Definition (3.3) makes sense because <FI is a continuous linear functional
on Sy,B and
10>
(t) E Sy,B:We observe that the function <F
I
G> can be extended to an analytic function on the open right half of the complex plane. Thus it can be seen as analmost periodic distribution along the imaginary axis.
We mention the following relations
(3.4) <FI (t)
=
<F(t)I
(0) , I G> (t)=
1 G(t» (0)<F I
0>
(t)=
<F(t) 1 G> (0)=
<FI
G(t» (0) <FIG>=<GIF>For each test bra <f 1 the function <f I G> can be extended to
Next we present a mathematical interpretation of Dirac's expansion
theo-rem for kets with respect to continuous basis.
(3.5) Theorem.
Let
(1a»aEM
be a Dirac basis inTy,B'
and letIF>
be a keto Then we have<a
I
F>
I a>
dll ato be interpreted in the following sense
IF>
(t)=)
<aI
F>
('r)Ia>
(t - T) dll ,a t > 0 .
The integral expression does not depend on the choice of T, where 0 < T < t.
(3.6) Example.
Let
I j>, j
E IN, be an orthonormal bas is in X and letI F>
be a ket. ThenIF>
=L
<j
I
F> I j>
in the sense thatIF>
(t)=
jElN
t >
o.
I
<j I F>
(t)I
j>
(0) ,jElN
With our concept of bracket we can interprete Dirac's orthogonality
rela-tions for continuous bases,
We only give a rough sketch of this interpretation here (cf. [EG1]). The
evaluation functional
Os :
f ~ f(a),a
EM,
belongs to the spaceT
L2(f~,l.l)-
for a certain operatorB
which is unitarily equivalent toB.
So 013 is a trajectory
and <a
I
13>
= 0B(a) has to be interpreted as <aI
8>
(t) = 0e(a;t) , t>
0 •4. The Fock space
General vector space theory yields the existence of an algebraic tensor
product of Y and Z, we denote it by Y ® Z. The bilinear mapping ® sends a
the pair [en,gm]
N nations
I
a e n=1 n n to e ® g and more n M m andL
13m gm into m=l usual identifications,generally the pair of linear combi-N 11
I I
a 8 (e ® g ) with the n=1 m=l n m n mIn Y ® Z we define a sesquilinear form through a
Then Y ® Z becomes a pre-Hilbert space with orthonormal basis
a
(e n ® g) m n,
mE-'
... , Its completion is denoted by Y ® Z; it is called the Hilbert tensor product of Y and Z, Naturally, Y ® Z consists of all serieswith
Let X be a separable infinite dimensional Hilbert space with orthonormal
(4.1) with Ax
=
I
j (x, v j) Vj jEIi D(A)=
{x E X 1I
j2 1 (x,V J.)12 <~}
jEIi(E. g. we can take X
=
L2 (Il) and A=
t (-
d 22 + x 2
+ 1).) Then the spaces dx
Sx
,
A and Tx,
A are nuclear.Next, we can inductively define the Hilbert spaces X(k)
=
X ® ••• ® X(k times). X(k) is called the k-fold Hilbert tensor product of X. An
ortho-normal basis in X(k) is established by the vectors v. ® •••
J 1
j
=
(j1"'" jk) E lik. The space X(k) consists of all series(4.2) where (4.3) where
I
Wj (v. jElNk J1I
Iw.12 < ~ jElNk J ® ••• ® v . ) JkThe inner product in
(1) (2)
I
(1) (2) (w ,w )X(k)=
Wj Wj jElNk (i)I
(i) (v. ® ••• ® w=
Wj v. jElNk J1 Jk X(k) is given by ) i=
1,2.
® v where jk Further, if w l ' ... , wk belong to X, then the sequence (w j) jElNk with W
=
(w1' v. ) ••• (wk,v. ), j E lN k
, uniquely determines an element of X(k).
j J1 J
k
This element is denoted by w 1 ® tensor product of (w
1' ••• ,wk).
® w
k and it is called the k-fold
In X(k) we define the positive self-adjoint operator A(k) by
(4.4) ® ••• ® v. )
Jk
® ••• ® v. ) Jk
followed by linear and unique self-adjoint extension, i.e.
A (k)
=
A ® I ® ••• ® I + I ® A ® I ® ••• ® I + ••• + I ® I ® ••• ® I ® A •So we take
Ijl
=
j1 + j2 + + jk'Since
jJ~k
e-tljl < ro for all t > 0, the spaces SX(k),A(k) and TX(k),A(k)are nuclear. We observe that
- t A(k) (
e v. ® ••• ®v.)
J 1 J k
It leads to the following results.
(4.5) Proposition. tt+- <v ,F > j1 1 - tj 1 -tjk = (e v.. ) ® ••• B (e v. ) J1 Jk <v ,F > (V j jk k 1
is a member of TX(k),A(k). This mapping is denoted by Fl ® " . ® Fk ,
The space TX(k).A(k) is a k-fold topological tensor product of Tx,A'
Let
The
f1.···,f k € Tx,A' Then f1 ® ••• ® fk € X(k) belongs to SX(k),A(k). space SX(k),A(k) is a k-fold topological tensor product of Sx,A'
Proof: Cf. [EGmJ. Ch. III.
(4.6) Lemma.
Let Y1'Y2"" be Hilbert spaces, and let
B
1
,B
2, ••• be positive self-adjointoperators in Y
1,Y2, ••• , respectively. Then we define the operator diag(Bk)
by
00
The operator diag(B
k) is positive and self-adjoint in
co
We put F
= •
X(k) and H=
diag(A(k» with A(l)=
A, X(l)= x,
A(o)=
0 k=Oand X(O)
=
t. The spaceSF,H
is called the Fock analyticity space and TF,
H
the Fock trajectory space. The operatorH
has a complete orthonormal basis of eigenvectors with the set of natural number as its spectrum. Themultiplicity of 0 equals mO
=
1, and of N::: 1,2, • •• equals mN :::f
(:=~)
=
N-l
(N-1)
k-1
=
2 • (We note that k-l is the number of ways to write N as a sum of k natural numbers, e.g. 5=
1 + 1 + 3=
1 + 3 + 1=
3 + 1 + 1 = 2 + 2 + 1=
2 + 1 + 2 = 1 + 2 + 2. ) It follows that the operators e-tH
t > 0, are compact, but they are not Hilbert-Schmidt for o < t < log 2. So the spacesSF,H
andTF,H
are Montel but not nuclear.5. Introducing the machinery
In the remaining of this paper we stick to Dirac's bracket notation
be-cause of its expressive nature. The technical results we will need at the
=
end of this paper, are contained in this section. We will rigorize a number
of heuristic formulas which are in use by theoretical physicists. We start
with some additional notations.
The trajectory space TX(k),A(k) represents the k-particles kets and bras;
<F ;
11
we shall also writeI
F> and<Fl.
The "kets" and "bras" related toT F , H are denoted by
{I
F ; k>}~
and{<F ;
kI }
~
.
As observed in tbe previous section the set {vj
1
an orthonormal basis in X(k) and each Vj ® ••• ® V. is an eigenvector of
1 Jk
A(k). These eigenvectors are denoted by
(5.1)
It follows that
(5.2) IF; k>
=
r
<j ; kiF ; k>I
j ; k> .jElNk
with the usual interpretation as given in Section 3,
The relation
(5.3)
leads to the following considerations: Associated to the k-tuple
From Theorem (4.5) it follows that
So it makes sense to write
and for the corresponding bra <G ; k I •
(5.5) (cf. [Di], p. 81) •
Similarly, for a k-tuple of test kets (lf
1> , ••• , Ifk» the k-particles ket
is a k-particles test ket, and <g ; k I ,
a k-particles test bra.
Now let I~> be a I-particle keto Identifying SX(k),A(k) with the space of all k-particles test kets. we define the operator dk «<I>
I)
on SX(k). A(k) by(5.5)
Remark.
For I fl> , ••• , I fk> test kets we have
(5.6) Lemma.
Let Iw> be a normalizable ket, i.e. X - lim Iw> (t)
=
Iw> (0) exists. ThenHO
dk«wl) extends to a Hilbert-Schmidt operator from X(k) into X(k - 1).
Proof: We compute the matrix of dk«wl) with respect to the orthonormal
So
Thus the assertion has been proved.
o
(5.7) Lemma.
Let I~> be a ket, and let t > O. Then
k-l Proof: We observe that for all n,m
E
~ and n1
E
~=
<qJ
I n1>
(t) d =<m;
k - l l a.k«~1 (t»I
n1,n; k> (0) •, nm
o
(5.8) Theorem.
The operator
ak«~I)
maps SX(k),A(k) continuously into SX(k-l).A(k-l)·Proof: Combine Lemma (5.6) and Lemma (5.7), look at Theorem (2.2).
o
(5.9) Theorem.
The operator ak«~I) is extendible iff I~> is a test keto Proof: Following Theorem (2.5) we have to prove that V
Let t > O. If I~> is not a test ket, then for all T > 0 the expression
<*)
is infinite. If I~> is a test ket, then we can take 0 < T < t so smallthat I ~> (-T) E X and hence the expression (*) is finite for such T. 0
We denote the dual mapping of ak«~I) by Ck(I~», Then Ck(I~» maps the (k - 1) -particles ket space T X(k-1) , A(k-1) cont inuously into the k-particles ket space TX(k),A(k), Easy computation shows that
(5.10)
and also that
We have the following corOllary,
(5.11) Corollary.
Let I~ be a test keto Then ~(I~) is a continuous linear mapping from SX(k-1),A(k-l) into SX(k),A(k) with Hilbert space adjoint
Ck(I~)*
=
= dk(<tp\).
Remark,
The mappings
and
\~ a test ket
Now let (1~»~€M be a Dirac basis. For each ket l~> we want to interprete the following heuristic expressions
and
a
k«411)
==J
<41
I
s>
a
k «l;I )
dllE; M C k(141))
==J
<l;I
41>
Ck(Il;»
d 1.\ MTo this end, we first prove the following lemma.
(5.12) Lemma.
Let I~ be a test keto Then ak«~I) can be expressed by
co
I
«P
I JI.> (0) ak «JI. I) with convergence in the following sense JI.=1
Proof: Let t > O. Then by Lemma (5.6) and (5.7)
-R,t
=
e Henceco
I
I
«P
I
JI.> (0)1
lie t A(k-1) ak «9.
I)
e -t A(k) II :;; JI.=1So for all t > 0 we have
co
=
I
«P
I
JI.> (0)ak«Jl.1
(t» .JI.=l
Employing the expansion theorem for Dirac bases, we derive ==
1:
<lP
I t> (0) dk «R.I) == t==1 ==~
(J
<lP
I~>
(0)<~
I t> (0)dll~)
dk «tI)
=
R.-1M
=
I
(J
<lP
I~>
(-'1:)<~
! R.> (T)dll~)
dk«tl)=
R,=1 M<:)
J
<lP
I~>
(-T)(~ <~
I
t> (T) dk«t!»)dll~
=
M
t-1 ==J
<lP
I~>
(-T)dk«~1
(T»dll~
• M*) It is allowed to interchange summation and integration:
(~
f)
1<lP
Ii;>(-T)<~
I D (T)llIdk«R.I)1Idll~ ~
R,-1
M
~
CIl
.-2h)1
(MI
I«I!
1
,>
(-T)12
d",f .
We come to the following theorem.(5.13) Theorem.
Let I~> be a keto Then
dk
«~
I)=
f
<~
I t;> d k«~
I )
dll~
M
in the sense that
V t>0 e t A(k-1) d
k
«4>
I)
e -t A(k)=
J
<~
I
t;> (t - T) dk«~
I
(T»d1J~
•Proof: Observe that
et A(k-l) ak«m 'i'
I)
e-t A(k)=
a <I
k ( ~ ( t
»
and that <q, I (t) is a test ket.
Similarly, by taking adjoints
(5.14) Corollary.
Let Iq,> be a keto Then
C
k ( I
~»
=
f
<t;I
~>
Ck (I
f;»
dllt;M
in the sense that
Vt>o e -t A(k) C
k
(I
W»
e t A(k-l)=
f
<i;I
q,> (t - T) Ck(I
t;> (T» dllt;o
M 0
We finish this section with the presentation of the technicalities which
are involved in the concept of second quantization.
Let
V
EL(T
x
,
A)' ThenV
satisfies -tA
V TA
e e : X + X is bounded •
So for all i; E M we obtain
We can compose the operators Ck(V I;» and ak«~I). The product Ck(V 1~»ak«~I) is a continuous linear mapping from SX(k),A(k) into TX(k),A(k)' Consider the following estimation
:ii lie -t A(k) C
k
(V
1~»
e t A(k-l) II lie t A(k-l) ak«~
I)
e -t A(k) II :iiFor all t > 0 and T > 0 sufficiently small we thus obtain
Mf
lie -t A(k)
c.
(VI
~»
a
«~
I)
e -t A(k) II dj.l :iik k F,:
Hence the integral expression
M
J
e -t A(k)
c.
(VI
~»
a
«F,:I)
e -t A(k) dJJk k ~
denotes a bounded linear operator from X(k) into X(k). Now the expression
(5.14)
acts in the following way
(5.14')
(J
Ck(V IF;» ak«F;I) dlJF;) • Itp;k>
M
(5.15) Proposition.
The linear mapping
defined in (5.14) and (5.14') belongs to L(SX(k),A(k),TX(k),A(k»' Further,
for all t > 0 and 0 < L < t, we have
=
f
<s'
IVI
s>(t-L)Ck<ls'>('r»e-2tA(k-lak«sl<t»dlJ~
® lJF;' .MxM
Proof: Continuity of
MI
Ck (V Is» ak «I;
I)
dlJl; follows from the considerations which led to (5.14).The other assertion in the proposition can be derived by the following
straightforward computation
=
Ck«V Is» (t»=
J
<I;'
I V IF;> (t -L) ck<II;'> ('r» dlJkM
6. Annihilation and creation operators
In Section 4 the Fock analyticity space
Sf.H
and the Fock trajectory spaceT f • H have been introduced. We shall denote their elements by {I g ; IV} ~, { IF; IV} ~. etc. With the aid of the operators ~ «~ I). k E 1N J introduced
in the previous section, we define the operator a«~I) on
Sf.H
by(6.1)
or, equivalently. by the operator matrix
(6.2)
o
v'1
a
1 «4>1 )0
o
12
a2«~I)
o
13
a
3 «4>1 ).
o
• • (6.3) Theorem.Let 14» be a keto Then the linear operator a«~I) maps
Sf.H
continuously intoSf.H'
Proof: Let t > 0 and let 0 < , < t. Then by Lemma (5.6) and (5.7)
sup (ilik e' A(k-1) a
k«4> 1 ) e- t A(k)lI) ;:;i; k€1N
;:;i; sup
(Ik
e -k(t-1") lie' A(k-1) ak «~ 1 ) e -, A(k) II) ;:;i;
k€lN
;:;i; 1l1<I»(,)11 sup (lke-k(t-,» < 0 0 .
k€lN
Hence the opera tor e'
H
a
«~
1 ) e -tH :
f -+ f is bounded.(6.4) Theorem.
Let I~ be a test keto Then a(~I) is extendible. Proof: We have to proof that
-t H I T H
"'t>O 3
T>0 : e a(~ ) e : F -+- F is bounded . Let t > 0 and 0 < T < t so small that I~ (-T) E X. Then
lie -t H
a(~I)
e't' Hii~
supnlk
e -t A(k-l)ak(~I)
e't' A(k) II~
kE:N::i e t - T II Iq:»(-"n Ii sup
(Ik
e-k(t-T»kElN
We denote the dual mapping of a«~ I) by c. ( lIP». Then C. ( I ~» maps T F. H
con-o
tinuously into
TF H'
Further, ifI~
is a test ket, thenc.(I~)rs
belongs,
F ,H
to
L(SF
.
H)' The linear mapping c.(I~) acts onTF H
,
as followswhere we put IF; -V = 0,
c.
o
(I ~»=
O. It is represented by the operator matrixo
o
11
c. 1 ( lIP» 012
c. 2 ( lIP» 013
c. 3 ( lIP» 0o
• •We mention that for any normalizable ket Iw>
Moreover, the operators (()(
I
w»=
a«wI )
+ c.(I
w» and ({)t (
I
w»=
a«wl) - c.( Iw»
12
=
are self-adjoint inf.
i12
From the previous section we obtain the following propositions which
eventually will lead to a rigorization of some heuristic formulas used
in the formalism of boson field operators and fermion field operators.
(6.5) Proposition.
Let I~> be a keto Then we write
with the following interpretation: Vt>O VO<t <t <t
2 1
e 1: 1 H
a«~
I )
e - t H ==
J
<~
I
f;> (t1 - t2)
a.«~
I
(1:2»
e -(t-tl)Hdjl~
M
where each of the integrals converges absolutely in
L(f).
Proof: The proof is a consequence of Theorem (5.13), and the estimation
Remark.
= sup
(Ik
lIa.k«~1
(t2»lIlIe-(t-t 1 )A(k)lI)
~
kEm ~1111;>
(t 2) II sup(Ik
e -<t-t1)k) kENo
(6.6) Corollary,
Let I~> be a ket, Then we write
c.(
I
~»
=
I
<~
I
~>
c.(I
~»
dllE;M
with the following interpretation: V V
t>O 0<1"2<1"1 <t
Each of the integrals converges absolutely in L(F).
Further, we give an interpretation of the integral expression
) C(V
1<»
a
(<I;I)
d"<
where V is a member of L(TX,A),
From the previous section for all E; E M the following estimation can be derived
~ 1I1E;>(t)III1I~>(1")lIl1e-tAVe1"AIi
sup (k e-t(k-l» ,kElN
H ere L > 0 must be taken so small that the operator e -t A VeL A ].' s bounded,
So as in the previous section it follows that for all t > 0
J
e -t H C.(VI
;»
a«~I
)
e -t H dll~M
converges absolutely and hence can be regarded as a bounded operator on
f. It follows that the integral expression
J
C.(VI
;»
a«~
I )
dll~
M
denotes a continuous linear mapping from Sf,H into Tf,H' Its action on an
element
{I
f ; IV} E Sf ,H is given by(6.7) t 1+
(f
e-tHc.(VI~»a«~I)e-1"Hd)lf;){lf ;IV(-T)}~
M
where
{I
f ; IV (_or)}~ belongs to f for 1" > 0 sufficiently small.(6.8) Proposition.
The linear mapping
defined by (6.7) belongs to L(Sf,H,Tf,H)'
=
J
<~
tI
v
I
~>
(t - r) c(IE;
,> (1:» e -2t H a«sI
(t» dlls dlJt;;' •MxM
Each of the integral converges absolutely in
L(f).
Proof: Cf. Proposition (5.15). (6.9) Definition. The equality
f
c(ls'» <SfI
VI
s>a«E;
I
>
dlJ s dlJt;'MxM
can be interpreted as: "t>o "O<'t<t :
e-tH (
f
C(V It;» a«sl) dlJt;)e-tH=
M=
f
<E;'
I
VI
s> (t - T) c(I
Sf> (T» e -2t H a(<l;;I
(t» dllt; dlll;' .MxM
7. The symmetric and anti-symmetric Fock space
The set of vectors
(7.1)
establishes an orthonormal basis in X(k). Let P
k denote the permutation
group of order k. For each (]
e:
Pk we define the linear operator
a :
X(k) ~ X(k) by (7.2)...
The operator
a
is unitary on X(k). Next we define(7.3) p(+) k = -1
I
a...
k!
aEP
k
Then (p(+»2
=
p(+)=
(p(+»*,k k k i.e.
P~+)
is an orthogonal projection. Therange of p!+) is called the k-fold symmetric Hilbert tensor product of X;
it is denoted by X(+)(k). For completeness we take
P~+)
=
1.(7.4) Definition.
The orthogonal projection p(+) on F is defined by
Further we define F(+)
=
P(+)(F). The Hilbert space F(+) is called the Boson Fock space.DP
We have F(+)
=
e X(+)(k). k=OIt is not hard to see that
P~+)
(A(k»=
A(k) p!+), and We put A(+) (k) = p!+) A(k) p!+), and H(+)=
p(+) Hp(+).(7.5) Lemma. SF(+) H(+)
=
p(+)(S )•
F,H andT
F (+),
H(+)=
p(+) (T F,H ) (+) (+) hence P H = H P •The eigenvalues of the self-adjoint operators H(+) are the numbers N
=
=
0,1,2, . . . . Because of the symmetrization the multiplicity~+)
is strongly diminished in comparison with the multiplicity ~. In fact theinteger summands without regard to order". (E.g. 5
=
1 + 4=
2 + 3=
=
1 + 1 + 3=
1 + 1 + 1 + 2=
1 + 2 + 2=
1 + 1 + 1 + 1 + 1, whence m!+)=
7.) The asymptotics of~+)
are given in [AS]. p. 825,(7.6) ~ (+) ,..., - -I exp('Jf';~
12
vN) • r.: 4N13(lO
\' (+) -t N
So the series I.. mN e converges for all t > O.
N=1
(7.7) Lemma.
The spaces SF(+).H(+) and TF(+),H(+) are nuclear. -t H(+)
Proof: Observe that the operators e • t > 0. are Hilbert-Schmidt.
0
Similarly. we introduce the anti-symmetric Fock space F(-). Let
e : P
k + {-1.1} denote the function which is one for even permutations and minus one for odd permutations. We introduce
(7.8) £(0)
cr
Then
p~-)
is an orthogonal projection from X(k) onto X(-)(k)=
p~-)(X(k».
X(-)(k) may be called the k-fold anti-symmetric Hilbert tensor product ofX. For completeness we take
p~-)
=
1. (7.9) Definition.The orthogonal projection p<-) on F is defined by
Moreover F(-)
=
P(-)(F). The Hilbert space F(-) is called the Fermion Fock space.(-)
It is not hard to check that P
k A(k)
=
p(-)
H =
H
p(-). PutH(-)
=
p(-)H
p(-) .(7.10) Lemma.
SF(-)
,
H(-)=
p(-)(SF
,H )and
A(k) p(-) and hence that
k
The eigenvalues of H(-) are the numbers N
=
0,1,2, . . . . The multiplicity(-)
~ of N is equal to Uthe number of decompositions of N into distinct
integer summands without regard to order", (E,g. 5
=
1 + 4=
2 + 3, whence m!-)=
3.) The asymptotics of~-)
for large N are given in [AS], p, 826,~-)
"" YsN3
exp(1TIi
IN) .
(7.11) Lemma.The spaces
SF(-),H(-)
andTF(_),H(-)
are nuclear,8. A mathematical interpretation of the field operator formalism in quantum
mechanics
The mappings a«~I) and c(I~» do not satisfy the wanted canonical (anti-) commutation relations. In order to obtain these CAR and CCR we modify the
(8.1) Definition.
Let I~> be a keto We introduce the following operators
From the properties of the linear mappings
a«~I), e(I~»,
and pet) we obtain(8.2) Lemma.
a (±) «~
I)
= a«IP I )
P (±) ,From Sections 6 and 7 we obtain the following results.
(8.3) Theorem.
Let
I'>
be a ket, and let I~ be a test keto The following statements are valid- a(±)«'I) is a continuous linear mapping from SF(±),H(±) into
SF(±) H(±);
,
a(±)(<qJI) 1s extendible.- e(±)(I'» is a continuous linear mapping from TF(±),H(±) into
T F (±) ,H(±); \.;. ~(±)(I,n') '¥'
=
~±(~nl)*. ... ""VProof: Cf. Theorem (6.3) and (6.4).
The mappings
a
(±)«~
I)
ande
(±)(I
~», ~
E M, can be regarded as "basic operators" because we have the following expansion results.(8.4) Theorem.
Let I~> be a keto Then the formal expansion
a(±)
«~I)
=
J
<~
I s>
a(±)«s I)
d lls
Mcan be interpreted as follows: V t>O V 0<'[2 <'[ 1 <t
H (±) (+) H(±)
eLl
a -
«~I )
e -t=
=
Each of the integrals converges absolutely in
L(F(±».
Proof: Cf. Proposition (6.5) and Lemma (8.2).(8.5) Corollary.
Let I~> be a keto Then the formal expansion
can be interpreted as follows: V V t>O 0<L
2<T1<t
Next we present a distributional interpretation of the so-called canonical
(anti-) commutation relations.
o
Let I~> be a ket and let I~ be a test keto Following Theorem (8.3) the (anti-) commutators
(8.6)
are well-defined continuous linear mappings on TF(±),H(±) and SF(±),H(±)'
respectively. Similarly, for all pairs of kets I~» I~> the (anti-) commu-tators
(8.7) [c. (±) (
I
~»,
c.
(±) (I
~»I
+
are well-defined. We have the following result.
(8.8) Theorem.
Let I~> and I~> be kets, and let I~ be a test keto Then we have
[a.(±) «'VI ) •
c.(±)(1~)1
=
<'V I~
(0)1 •
+[a.(±)
«~I) ,a.±«~1)1
= 0+
Proof: We only prove the first commutator relation. The others can be dealt
with similarly.
Consider the following computations, where k E IN, j E lNk
k p(±)
c. (Iq:i»
a.«'VI)
p(±)I
j . k>=
k k k k '
k
k = (k + 1) a < I [ 1 1"''- p(±) I' ,-, + 1 \' (I' > k+1 ( '1' ) k + 1 'V'" k J ; IV' - k + 1 L J JI, • JI,=l JI, = <'1' I <P> (0)
p~±)
I j ; IV ±,2:
<'1' I jJl,> (0)p~±)
(IllY I j1"" ,jJl,-1 ' jJl,+l"",j~)
J=lThus we obtain with (*) and (**)
From these relations (***) we get
+ +
[a-«'1'I) , c.-(I<p»l = <'1' I <p> (0) I .
o
+
For arbitrary kets I~> and 1'1'> we thus obtain
+ +
(8.9) [a-«'1'I) ,c.-(I~> (t»l = <'1' I ~> (t) I
+
This observation leads to the following definition.
(8.10) Definition.
denotes the operator valued function on (OJ~) defined by
+ +
t ~ [a-«lfl) ,c-(I~>
(t»l '
t > 0 .+
(8.11) Corollary.
in the sense that for all t > 0
[a±«lfl)
,c(±)(I~>
(t»l
=
<If
I~>
(t) I+
For the members of the Dirac basis (I~»~EM we obtain
[a (±)
«~
I) ,
c
(±) (1
n»
1
=
<~
I n>
I=
+ 15 (~) 1 • n (Cf. Section 3.) Remark.The canonical (anti-) commutation relations can also be interpreted with
the aid of operator valued distributions. Here we give some heuristic
ar-guments.
valued distributions. For fixed n E
M
we haveHence
1
<.p) t+ [a (±)«n
I) ,
c
(±) (1<1»)]
is the operator valued distribution =+In Proposition (6.8) and Definition (6.9) we presented an interpretation
of the formal integral expression
MxM
I
dill; dill;' c.(It·»
<I;'
I
VI
1;>a
«I;
I) .
Multiplying both sides by p(±) should give a similar relation in the
Boson (Fermion) case.
(8.12) Definition.
Let V E
L(T
x,,4.)' LetMJ
C.(VI
E;»a
«I;I)
dllE; denote the operator inL(Sf,H.Tf,H)
as defined in (6.7). Then the second quantization Q(±)(V)of V is defined by Q (±) (V) = P (±) {
f
C.(V11;»
a«E;
I)
dill;} P (±) • M (8.13) Theorem. Let V EL(T
x
,4.), Then•
in the sense that for all t > 0:
-t H(±) (±) -t H(±)
e Q (V) e
=
Further,
o
(±) (V)=
f
c.
(±)(I
~
'»
<~'
I
VI
~>
a. (±)«~
I )
dll~ dll~,
MxM
in the sense that V t>O V O<T<t :
-t H(±) ± -t H(±)
e 0 (V) e
=
=
J
MxM
(±)
where the integral converge absolutely in L(F ).
Remark.
+
Appendix
In the spirit of this paper it can be shown that continuous operators
between Fock analyticity (c.q. trajectory) spaces can be represented in
their normal formal (cf. Sh)
c
=
I
m,n
in a rigorous way with respect to any Dirac basis.
The nature of the coefficients and the modus of convergence can be fully
interpreted within our theory.
It is very likely that most results in Shvarts' book [Sh] which are
semi-mathematical/semi-physical can be given a rigorous meaning with
References
[Co] Cook, J.M., The mathematics of second quantization. Trans. AMS,
74 (1953), pp. 222 - 245.
[Di] Dirac, P.A.M., The principles of Quantum Mechanics. Fourth edition,
Clarendon Press, Oxford, 1958.
[EG1] Eijndhoven, S.J.L. van, and J. de Graaf, A mathematical interpretation
of Dirac's formalism, Part a: Dirac bases in trajectory spaces.
[EG2]
[EG3]
[EG4]
[EGm]
To appear in Rep. Math. Phys ..
---, Part b: Generalized eigenfunctions in trajectory spaces. To
appear in Rep. Math. Phys ..
---, Analyticity spaces subjected to perturbations with applications
to Hankel invariant distribution spaces. To appear in SIAM J. of
Math. Anal..
---, On distribution spaces based on Jacobi polynomials. EUT-Report
84-WSK-01, Eindhoven, 1984.
---, Trajectory spaces, generalized functions and unbounded
opera-tors (with applications to Dirac's formalism). Monograph, to
appear 1985.
[EGK] Eijndhoven, S.J.L. van, J. de Graaf and P. Kruszynski, A dual system
of projective-inductive limits of Hilbert spaces. Preprint.
[EK] Eijndhoven, S.J.L. van, and P. Kruszynski, On Gelfand triples
generated by algebras of unbounded operators. EUT-Report 84-WSK-02,
[G] Graaf, J. de, A theory of generalized functions based on
holo-morphic semigroups. Proceedings Koninklijke Nederlandse
Akade-mie van Wetenschappen, A 86 (4) (1983), A 87 (2) (1984.
[KMP]
Kristensen, P.,L.
Mejlbo and P.T. Poulsen, Tempered distributions in infinitely many dimensions I. Commun. Math. Phys. 1 (1965).pp. 175 - 214.
[Ro] Robertson. B .• Introduction of field operators in quantum mechanics.
Amer. J. Physics 41 (1973).
[Sh] Shvarts. A.S .• Matematicheskie osnovy kvantovoj teorii polya.