Poisson bracket operator

T. Koide∗
Instituto de Fı́sica, Universidade Federal do Rio de Janeiro,
arXiv:2104.11780v1 [quant-ph] 23 Apr 2021

C.P. 68528, 21941-972, Rio de Janeiro, RJ, Brazil

We introduce the Poisson bracket operator which is an alternative quantum coun-

terpart of the Poisson bracket. This operator is defined using the operator derivative
formulated in quantum analysis and is equivalent to the Poisson bracket in the clas-
sical limit. Using this operator, we derive the quantum canonical equation which
describes the time evolution of operators. In the standard applications of quantum
mechanics, the quantum canonical equation is equivalent to the Heisenberg equation.
At the same time, the quantum canonical equation is applicable to c-number canon-
ical variables and then coincides with the canonical equation in classical mechanics.
Therefore the Poisson bracket operator enables us to describe classical and quantum
behaviors in a unified way. Moreover, the quantum canonical equation is applicable
to non-standard system where the Heisenberg is not applicable. As an example, we
consider the application to the system where a c-number and a q-number particles
coexist. The derived dynamics satisfies the Ehrenfest theorem and the energy and
momentum conservations.

I. INTRODUCTION

We reconsider the relation between the Poisson bracket in classical mechanics and the
commutator in quantum mechanics. In the canonical quantization, the time evolution of
operators are determined by solving the Heisenberg equation which is obtained from the
canonical equation by replacing the Poisson bracket with the commutator,

∂f ∂g ∂f ∂g i i
{f, g}P B = − =⇒ − [f, g] = − (f g − gf ) , (1)
∂x ∂p ∂p ∂x ~ ~

∗
Electronic address: [email protected],[email protected]
 2

where f and g are functions of canonical variables. Therefore the commutator is normally
considered to be a quantum counterpart of the Poisson bracket. This identification is however
not intuitively understandable. For example it is not clear why the derivatives appearing
in the Poisson bracket are replaced with the non-commutativity of operators. Moreover the
commutator is divided by ~ in Eq. (1) and hence we cannot see directly the classical limit.
To clarify the role of the Poisson bracket in quantum mechanics, we often map operators
in the Hilbert space into functions of phase space variables using the Wigner-Weyl transfor-
mation. In this approach, we can define the Wigner function which is the quasi-probability
distribution in the phase space. The time evolution of the Wigner function is characterized
by the Moyal bracket, which is reduced to the Poisson bracket in the classical limit. Thus the
Moyal bracket is regarded as a quantum counterpart of the Poisson bracket in this approach.
This perspective is extended in the deformation quantization [1, 2].
In this paper, however, we do not consider the phase space representation of quantum
mechanics. Instead we introduce the Poisson bracket operator as another quantum coun-
terpart of the Poisson bracket. This operator is defined through the operator derivative
formulated in quantum analysis proposed by Suzuki [4–10]. One of the advantages of our
approach is that the Poisson bracket operator has a clear classical correspondence with the
Poisson bracket because the operator derivative is equivalent to the standard derivative in
the classical limit. Using this operator, we derive the quantum canonical equation which de-
scribes the time evolution of operators. In the standard applications of quantum mechanics,
the quantum canonical equation is equivalent to the Heisenberg equation. At the same time,
the quantum canonical equation is applicable to c-number canonical variables and then co-
incides with the canonical equation in classical mechanics. Therefore classical and quantum
behaviors are described in a unified way by introducing the Poisson bracket operator and
this is another advantage of the Poisson bracket operator. Moreover, the quantum canoni-
cal equation is applicable to non-standard system where the application of the Heisenberg
equation is not clear. As an example, we consider the application to the system where a
c-number and a q-number particles coexist. The derived dynamics satisfies the Ehrenfest
theorem and the energy and momentum conservations.
This paper is organized as follows. In Sec. II, the operator derivative is introduced.
Three mathematical formulae are derived in Sec. III. These are used to show the relation
between the Poisson bracket operator and the commutator. The Poisson bracket operator
 3

and the quantum canonical equation are introduced in Secs. IV and V, respectively. As a
non-standard application of the quantum canonical equation, we consider the system where
a c-number and a q-number particles coexist in Sec. VI. Section VII is devoted to concluding
remarks.

II. DEFINITION OF OPERATOR DERIVATIVE

We employ the definition of the operator derivative proposed in quantum analysis [4–10].
For other applications of quantum analysis, see, for example, Refs. [11–16]
b where f (x) is a smooth function of x. Then the Gâteaux
Let us consider an operator f (A)
differential of this operator is defined by [3]
b b b
b = lim f (A + hB) − f (A) ,
df (A) (2)
h→0 h
b is an operator which is in general not commutative with the
where h is a c-number and B
b The operator derivative with respect to A
operator A. b is then expressed as (df /dA)
b and

then the Gâteaux differential is given by

b = df b
df (A) {B} . (3)
b
dA
b In
One can see that the operator derivative is a hyper-operator which is operated to B.
quantum analysis, this operator derivative is defined by
Z 1
df b b − λδA )B
b,
{B} = dλ f (1) (A (4)
dAb 0

where f (n) (x) = dn f (x)/dxn and

b ].
δA = [A, (5)

When an operator is given by a function of time, the time derivative is expressed as

( )
df (Â(t)) df dÂ(t)
= . (6)
dt dÂ(t) dt

The above definition can be extended to define the operator partial derivatives. Let us
b and B,
consider that a smooth function of operators A b f (A,
b B),
b which can be expanded as
X
b B)
f (A, b = bn B
fnm A bm , (7)
n,m≥0
 4

where n and m are integers and fnm is an expansion coefficient. The partial derivatives with
b and B
respect to A b are then defined by
b B)
b X Z 1 
∂f (A, b = n−1 bm ,
{C} n fnm dλ (Â − λδA ) Ĉ B
∂Ab 0
n,m≥0
Z 1  (8)
b B)
∂f (A, b X
b =
{C} b
m fnm An b − λδB )
dλ (B m−1 b ,
C
∂Bb 0
n,m≥0

respectively. When Ĉ is given by a c-number, say c, the operator derivatives are equivalent
to the standard derivatives with respect to c-numbers because δA c = 0 and δB c = 0,
b B)
∂f (A, b X
{c} = bmc ,
n fnm Ân−1 B (9)
∂A b
n,m≥0
b B)
∂f (A, b X
{c} = bn B
m fnm A b m−1 c . (10)
∂Bb
n,m≥0

Therefore, in the following calculations, we omit the argument { } when the operator deriva-
tive is operated to a c-number,
b B)
∂f (A, b b B)
∂f (A, b
= {1} , (11)
∂Ab ∂Ab
b B)
∂f (A, b b B)
∂f (A, b
= {1} . (12)
∂Bb ∂Bb

III. MATHEMATICAL FORMULAE

In this section, we derive three formulae which are used to show the relation between the
Poisson bracket operator and the commutator.

Formula 1
b and B,
For arbitrary two operators A b and an integer n ≥ 1, there exists the following
relation,
Z 1
b − λδA )n B
b = 1 bn B
b+A
bn−1 B
bAb+···+B
bAbn ) .
dλ (A (A (13)
0 n+1
Proof
This is proved by mathematical induction. It is easy to confirm Eq. (13) is satisfied for
n = 1,
Z 1
dλ (A b = 1 (A
b − λδA )B bBb+B
b A)
b . (14)
0 2
 5

Suppose that Eq. (13) is satisfied for n = L (L ≥ 1). Then the left-hand side of Eq. (13)
for n = L + 1 is calculated as
Z 1
b − λδA )L+1 B
dλ (A b
0
L  
(−1)L+1 L+1 b L+1 X m 1 1 bL+1−m (δA )m B
b
= (δA ) B + L Cm (−1) + A
L+2 L + 2 m=0 m+1 L+1−m
Z L+1
L+1 b 1 b Lb 1 X m bL+1−m b
= A dλ (A − λδA ) B + L+1 Cm (−1) A (δA )m B
L+2 0 L + 2 m=0
1 b bL b bL−1 b b bAbL ) + 1 (Ab − δA )L+1 B
b
= A(A B + A B A + · · · + B
L+2 L+2
1 b bL b bL−1 b b bAbL ) + 1 BbAbL+1 .
= A(A B + A B A + · · · + B (15)
L+2 L+2

The last equality is the right-hand side of Eq. (13) for n = L + 1. Therefore Eq. (13) is
satisfied for any integer n by mathematical induction. 

The formula (13) is satisfied for any operators. The other two formulae are however
applicable to operators which satisfy a special commutation relation.

Formula 2
Let us consider two operators which satisfies the commutation relation,

b B]
[A, b = c, (16)

where c is a c-number. Then, for arbitrary integers n, m ≥ 1, the following relation is

satisfied,
Z 1
B bm − A
bnA bm B
b n = mn(δB A)
b b − λδB )n−1 A
dλ (B bm−1 . (17)
0

Proof
bnA
The operator B bm is reexpressed as

bnA
B bm = B
b n−1 (δB A
b+A
bB)
b A
bm−1
bB
= (δB A) b n−1 A
bm−1 + B
b n−1 A
bBbA
bm−1
bB
= m(δB A) b n−1 A
bm−1 + B
b n−1 A
bm B
b. (18)
 6

b = −c. Applying this result to the operator B

Here we used δB A b n−1 A
bm which appears in the
second term on the right-hand side of the above equation, we find
n o
bn A
B bm = m(δB A)
bB b n−1 Abm−1 + m(δB A) bB b n−2 A
bm−1 + B b n−2 A
bm B
b B b
n o
b B
= m(δB A) b n−1 A
bm−1 + Bb n−2 A
bm−1 Bb +B b n−2 Abm B
b2
n o
b
= m(δB A) B Ab n−1 bm−1 b
+B A n−2 bm−1 b b
B + · · · + BA bm−1 b n−2
B +Abm−1 b n−1
B bm B
+A bn
Z 1
b
= mn(δB A) dλ(B b − λδB )n−1 A
bm−1 + A bm B bn . (19)
0

In the last line, the formula (13) was used. This is Eq. (17). 

Using the formula (17), we can show a kind of commutativity.

Formula 3
Let us consider two operators which satisfies the commutation relation,

b B]
[A, b = c, (20)

where c is a c-number. Then, for arbitrary integers n, m ≥ 1, the following relation is

satisfied,
Z 1 Z 1
b − λδB )n−1 A
dλ (B bm−1 = b − λδA )m−1 B
dλ (A bn−1 . (21)
0 0

Proof
b and (n, B)
By the exchange between (m, A) b in the formula (17), we obtain
Z 1
bm bn b n bm
A B − B A = mn(δA B) b b − λδA )m−1 B
dλ (A b n−1 . (22)
0

Using this and Eq. (17) itself, we find

Z 1 Z 1
b
mn(δB A) b
dλ (B − λδB ) An−1 bm−1 b
= −mn(δA B) b − λδA )m−1 B
dλ (A b n−1 . (23)
0 0

b = −δA B.
It is easy to obtain Eq. (21) from this because δB A b 
 7

IV. POISSON BRACKET OPERATOR

b B)
We consider a pair of canonical variables (A, b and two smooth functions f (A,
b B)
b and
b B)
g(A, b which can be expanded as shown in Eq. (7). The Poisson bracket operator is then
defined by
( ) ( )
n o b B)
∂f (A, b b B)
∂g(A, b b B)
∂f (A, b b B)
∂g(A, b
b B),
f (A, b g(A,
b B)
b ≡ − . (24)
b B)
(A, b
∂Ab ∂Bb ∂Bb ∂ Ab
From Eq. (4), the operator derivative agrees with the standard derivative of c-number when
b B)
the canonical variables (A, b are commutative. Therefore one can easily see that the
classical limit of the Poisson bracket operator is given by the Poisson bracket,

{ , }(A, b −−→ { , }P B .
b B) (25)
~→0

When the commutation relation for the canonical operators is given by a non-vanishing
c-number,

b B]
[A, b = c, (26)

we can show that the Poisson bracket operator is represented by the commutator,
n o 1
b B),
f (A, b g(A,
b B)
b = [f (A,b B),
b g(A,b B)]
b . (27)
b B)
(A, b c
Proof
b B)
Let us expand f (A, b and g(A,
b B)
b as
X
b B)
f (A, b = bn B
fnm A bm , (28)
n,m≥0
X
b B)
g(A, b = bn B
gnm A bm , (29)
n,m≥0

where n and m are integers and fnm and gnm are expansion coefficients. The commutator
of these operators is calculated by

b B),
[f (A, b g(A,
b B)]
b
X  
= fab gcd Aba B
bb A
bc B
bd − A
bc B
bdA
ba B
bb
a,b,c,d≥0
X  Z 1  
= ba b
fab gcd A bc(δB A) b − λB δB )
dλ (B b−1 bc−1
A +A Bbc bb bd
B
a,b,c,d≥0 0

X  Z 1  
− fab gcd A b
bc ad(δB A) b − λB δB )d−1 A
dλ (B ba−1 ba bd bb
+A B B . (30)
a,b,c,d≥0 0
 8

Here we used the formula (17). Applying the formula (21) to the last line, we find

b B),
[f (A, b g(A,
b B)]
b
Z !
X 1 b b
= b a b
fab A b(δB A) b − λB )b−1 dg(A, B)
dλ (B
0 dAb
a,b,c,d≥0
Z !
X 1
dg( b B)
A, b
− b
fab a(δB A) b − λA δA )a−1
dλ (A Bbb
0 dBb
a,b,c,d≥0
( ) ( )
df ( b B)
A, b dg( b B)
A, b df ( b B)
A, b dg( b B)
A, b
b
= (δB A) b
− (δB A)
dBb dAb dA b dB b
n o
= −(δB A)b f (A, b B),
b g(A,b B)b . (31)
b B)
(A, b

b = −c 6= 0. Therefore Eq. (27) was derived. 

In the present case, δB A
The above result can be generalized to many-body systems. We consider N-pairs of
bi , B
canonical variables (A bi ) which satisfy the following commutation relations for i, j =

1, · · · N,
bi , B
[A bj ] = ci δi,j ,
bi , A
[A bj ] = 0 , (32)
bi , B
[B bj ] = 0 ,

b B})
where ci is a c-number. Let us consider two smooth functions f ({A, b and g({A,
b B})
b
which can be expanded as
X
b B})
f ({A, b = bα1 1 B
fα1 ,β1 ···αN ,βN A b1β1 · · · A
bαN B
b βN , (33)
N N
α1 ,β1 ···αN ,βN ≥0
X
b B})
g({A, b = bα1 B
gα1 ,β1 ···αN ,βN A b β1 · · · A
bαN B
b βN , (34)
1 1 N N
α1 ,β1 ···αN ,βN ≥0

where α1 , β1 · · · αN , βN are integers and fα1 ,β1···αN ,βN and gα1 ,β1···αN ,βN are expansion coeffi-
cients. Then the commutator is expressed as
h i XN n o
b B})
f ({A, b , g({A,
b B})
b = b B})
ci f ({A, b , g({A,
b B})
b . (35)
bi ,B
(A bi )
i=1

Proof
It is sufficient to prove the following equation,
h i
Abα1 B
b β1 · · · A
bαN B
b βN , A
bγ1 B
b δ1 · · · A
bγN B
b δN
1 1 N N 1 1 N N
N
X n o
= ci Abα1 1 B
b1β1 · · · A
bαN B
b βN , A
bγ11 B
b1δ1 · · · A
bγN B
b δN . (36)
N N N N bi ,B
bi )
(A
i=1
 9

The case for N = 1 is already shown in Eq. (27). Suppose that the above equation is
satisfied for N = L (L ≥ 1). Then, for N = L + 1, we find

bα1 B
[A bαL+1 B
b β1 · · · A b βL+1 , A
bγ1 B bγL+1 B
b δ1 · · · A b δL+1 ]
1 1 L+1 L+1 1 1 L+1 L+1
(L) (L) bαL+1 b βL+1 bγL+1 b δL+1 (L) (L) bγL+1 b δL+1 bαL+1 b βL+1
= FAB GAB [A L+1 BL+1 , AL+1 BL+1 ] + [FAB , GAB ]AL+1 BL+1 AL+1 BL+1
n o
(L) (L) bαL+1 Bb βL+1 , A
bγL+1 Bb δL+1
= cL+1 F G AB A AB L+1 L+1 L+1 L+1 bL+1 ,B
(A bL+1 )
L
X n o
(L) (L) bγL+1 B
b δL+1 A
bαL+1 b βL+1
+ ci FAB , GAB AL+1 L+1 L+1 BL+1
bi ,B
(A bi )
i=1
L+1
X n o
= ci bα1 · · · A
A bαL B
b β1 · · · B
b βL , A bγL+1 B
bγ1 · · · A b δL+1
b δ1 · · · B , (37)
1 L 1 L 1 L+1 1 L+1 bi ,B
bi )
(A
i=1

where we introduced

(L) bα1 B
b δ1 · · · A b δL ,
bαL B
FAB = A1 1 L L (38)
(L) bγ11 B
b1δ1 · · · A
bγL B
b δL .
GAB = A L L (39)

From the first to the second equality, we used Eq. (36) by mathematical induction. The
last equality is the right-hands side of Eq. (36) for N = L + 1 and thus Eq. (36) is hold for
arbitrary integer N ≥ 1. From this, it is easy to show the formula (35). 

V. QUANTUM CANONICAL EQUATION

In this section, we derive the differential equation to describe the time evolution of canon-
ical variables.
Let us consider the system which is described by N-pairs of canonical variables
bi (t), B
(A bi (t)), (i = 1, · · · N). Then the time derivative of f ({A(t),
b b
B(t)}), defined by Eq.
(33), is given by
N
( ) N
( )
d X b
∂f ({A(t), b
B(t)}) bi (t)
dA X b
∂f ({A(t), b
B(t)}) bi (t)
dB
b b
f ({A(t), B(t)}) = + .
dt ∂Abi (t) dt bi (t)
∂B dt
i=1 i=1
(40)

Here we used the properties of the operator derivatives given by Eqs. (6) and (8). Suppose
that a Hamiltonian operator Hb is the time generator of the canonical variables and thus the
 10

bi (t) and B
time evolutions of the canonical variables A bi (t) are given by

bi (t) b X N n o
dA ∂H b b
= = Ai (t), H ,
dt bi (t)
∂B bi (t),B
(A bi (t))
i=1
N n
(41)
bi (t)
dB ∂H b X o
=− = Bbi (t), H
b ,
dt ∂Abi (t) (A bi (t),B
bi (t))
i=1

respectively. Then Eq. (40) can be expressed in terms of the Poisson bracket operator,

XN n o
d b b b b b
f ({A(t), B(t)}) = f ({A(t), B(t)}), H
dt i=1
bi (t),B
(A bi (t))
n o
b
≡ f ({A(t), b
B(t)}), Hb . (42)
b B(t)})
({A(t), b

We call this equation the quantum canonical equation.

The Heisenberg equation can be obtained from the quantum canonical equation. In
the standard application of quantum mechanics, the commutation relations of all pairs of
canonical variables are characterized by the same constant, the Planck constant ~,

bi (t), B
[A bj (t)] = i~δij ,
bi (t), A
[A bj (t)] = 0 , (43)
bi (t), B
[B bj (t)] = 0 .

In this case, the formula (35) is applicable and the quantum canonical equation is shown to
be equivalent to the Heisenberg equation,

d n o
b b b b b
f ({A(t), B(t)}) = f ({A(t), B(t)}), H
dt b B(t)})
({A(t), b

i b b b .
= − [f ({A(t), B(t)}), H] (44)
~

The correspondence between classical and quantum behaviors is clear in the quantum
canonical equation. In the derivation of Eq. (42), we did not use the property of the com-
bi (t), B
mutation relation [A bj (t)] and thus the quantum canonical equation is applicable to the
bi (t), B
commutative case, [A bj (t)] = 0. Because the Poisson bracket operator behaves as the

Poisson bracket for c-number variables, the quantum canonical equation reproduces the clas-
sical canonical equation in the application to c-number canonical variables. In other words,
the quantum canonical equation enables us to describe classical and quantum behaviors in
a unified way.
 11

VI. APPLICATION TO NON-STANDARD SYSTEM

As was shown in the previous section, the Poisson bracket operator is equivalent to
the commutator when canonical variables satisfy the standard commutation relations (43).
The quantum canonical equation is obtained independently of the behaviors of commutation
relations and thus is applicable to describe a system where the Heisenberg equation is difficult
to apply. As an example, we consider a system where a c-number and a q-number particles
coexist.

A. model and application of quantum canonical equation

As an example of the non-standard application, we consider a system where two pairs of

canonical variables satisfy different commutation relations (i, j = 1, 2),

bi (t), B
[A bj (t)] = ci (t)δij , (45)

where c1 (t) and c2 (t) are c-numbers and c1 (t) 6= c2 (t). Suppose that the system 1 described
b1 (t), B
by (A b1 (t)) is separated from the system 2 by (A b2 (t), B
b2 (t)), and these systems do not
interact each other. The Heisenberg equations for the system 1 and for the system 2 will
be, respectively, characterized by c1 (t) and c2 (t),

d b b1 (t)) = 1 [f (A
b1 (t), B
b1 (t)), H1 ] ,
f (A1 (t), B c1 (t)
(46)
dt
d b b2 (t)) = 1 [f (A
b2 (t), B
b2 (t)), H2 ] .
f (A2 (t), B c2 (t)
(47)
dt

Here H1 and H2 are the Hamiltonian operators for each system. When an interaction
between the two systems is introduced, we have to generalize the Heisenberg equations so
that the generalized equation reproduces Eqs. (46) and (47) in the vanishing limit of the
interaction. Such a generalization is not trivial.
By contrast, the quantum canonical equation is applicable to such a system systematically.
As an extreme case of the above system, we consider a toy model where two particles
coexist: one is described by a c-number and the other by a q-number when there is no
interaction between the particles. The canonical variables for the former particle are denoted
b
by (x(t), p(t)) and those for the latter by (X(t), Pb(t)). The commutation relations are given
 12

by

[x(t), p(t)] = 0 ,
(48)
b
[X(t), Pb(t)] = i~ ,

respectively. For the sake of simplicity, we assume free particles. Then the c-number canon-
ical variables satisfy the classical canonical equations,
d p(t)
x(t) = {x(t), Hc }P B = ,
dt m (49)
d
p(t) = {p(t), Hc }P B = 0 ,
dt
where the c-number Hamiltonian with mass m is defined by
p2 (t)
Hc = . (50)
2m
The q-number canonical variables satisfy the Heisenberg equations,

d b i b Pb(t)
X(t) = − [X(t), Hq ] = ,
dt ~ M (51)
d b i
P (t) = − [Pb(t), Hq ] = 0 ,
dt ~
where the q-number Hamiltonian with mass M is

Pb 2(t)
Hq = . (52)
2M
The above two equations are described by the common quantum canonical equation,
d b
f (x(t), p(t), X(t), Pb(t))
dt
b
= {f (x(t), p(t), X(t), Pb(t)), H}(x(t),p(t)) + {f (x(t), p(t), X(t),
b Pb(t)), H}(X(t),
b Pb(t))

b
≡ {f (x(t), p(t), X(t), Pb(t)), H}(x(t),p(t);X(t),
b Pb(t)) , (53)

b
where H is the total Hamiltonian defined by H = Hc + Hq , and f (x(t), p(t), X(t), Pb(t))
is an arbitrary smooth function of the canonical variables which can be expanded as Eq.
(33). As pointed out earlier, the Poisson bracket operator behaves as the Poisson bracket
for c-number variables. In the following calculation, we suppose that the quantum canonical
equation is applicable to any Hamiltonian.
Let us consider the interaction Hamiltonian defined by
α b b
HI = (x(t) − X(t))(x(t) − X(t)) , (54)
2
 13

where α is a coupling constant. It should be noted that the canonical variables (x(t), p(t))
becomes non-commutative and behave as operators by the influence of this interaction.
Therefore the commutation relations (48) are modified by the interaction. The modifica-
tions of the commutation relations are obtained only after solving the quantum canonical
equations as shown later.
Using total Hamiltonian defined by

H = Hc + Hq + HI , (55)

the quantum canonical equations are given by

dx(t) p(t)
= {x(t), H}(x(t),p(t);X(t),
b Pb(t)) = ,
dt m
dp(t) b
= {p(t), H}(x(t),p(t);X(t),
b Pb(t)) = −α(x(t) − X(t)) ,
dt
b (56)
dX(t) b Pb(t)
= {X(t), H}(x(t),p(t);X(t),
b b
P (t)) = ,
dt M
dPb(t)
= {Pb(t), H}(x(t),p(t);X(t),
b
b
Pb(t)) = α(x(t) − X(t)) .
dt
To solve these equations, suppose that the two particles start to interact each other at
an initial time t = 0. Then the initial canonical variables (x(0) = x0 and p(0) = p0 ) behave
b
as c-numbers and (X(0) =X b0 and Pb(0) = Pb0 ) as q-numbers satisfying

[x0 , p0 ] = 0 , b0 , Pb0 ] = i~ ,
[X
(57)
b0 ] = 0 ,
[x0 , X [p0 , Pb0 ] = 0 .

b0 and Pb0 operate to an initial wave function |Ψi normalized by

The canonical operators X

hΨ|Ψi = 1 . (58)

To solve the differential equations, we further introduce new canonical variables associated
with the center of mass and relative motions,
!
b
mx(t) + MX(t)
bC (t), PbC (t)) =
(X , p(t) + Pb(t) , (59)
m+M
!
b Mp(t) − mPb(t)
(b
q(t), pbq (t)) = x(t) − X(t), . (60)
m+M
 14

The quantum canonical equations are simplified for these new canonical variables. Solving
the equations, the motions for the center of mass coordinate are described by
b
bC (t) = mx0 + MX0 +
X
1
(p0 + Pb0 )t ,
m+M m+M (61)
PbC (t) = p0 + Pb0 ,
and those for the relative coordinate are
r  r ! r 
α µ p 0 Pb0 α
b
qb(t) = (x0 − X0 ) cos t + − sin t ,
µ α m M µ
r  ! r  (62)
√ α Mp 0 − m Pb0 α
b
pbq (t) = − αµ(x0 − X0 ) sin t + cos t .
µ m+M µ
Here the reduced mass is defined by
mM
µ= . (63)
m+M

B. Ehrenfest theorem and non-interacting limit

The quantum canonical equations (56) satisfy the Ehrenfest theorem. Indeed we obtain
classical canonical equations from a classical Hamiltonian obtained by replacing q-numbers
with the corresponding c-numbers in Eq. (55). The structures of these classical canonical
equations coincide with our canonical equations (56) when we ignore the non-commutativity
of q-numbers.
Below Eq. (47), we discussed the desirable property which should be satisfied in the
vanishing limit of the interaction α → 0. This property is hold in our model. In this limit,
the solution of Eqs. (61) and (62) are reduced to
bC (t) + M p0
x(t) = X qb(t) = x0 + t ,
m+M m
m b
p(t) = pbq (t) + PC (t) = p0 ,
m+M
(64)
b b M b Pb0
X(t) = XC (t) − qb(t) = X0 + t ,
m+M M
M
Pb(t) = −b pq (t) + PbC (t) = Pb(t) .
m+M
It is easy to confirm that these are the solutions of the classical canonical equations (49)
and the Heisenberg equations (51). Moreover, the commutation relations of these canonical
variables satisfy Eq. (48) and thus the canonical variables (b
x(t), pb(t)) recover commutativity
as we expected in the vanishing limit of the interaction.
 15

C. conservation law

The total momentum conservation is easily seen from the behavior of PbC (t) shown in Eq.
(61).
The total energy of this system is defined by the expectation value of the total Hamil-
tonian operator, hΨ|H|Ψi, where Ψ is the initial wave function. The time evolution of the
Hamiltonian operator (55) is determined by
d
H = {H, H}(x(t),p(t);X(t),
b Pb(t)) . (65)
dt
Thus, to conserve the energy, {H, H}(x(t),p(t);Xb (t),Pb(t)) should vanish. This is however not
trivial because the Poisson bracket operator is not equivalent to the commutator in the
present toy model as will be shown later in Sec. VI D.
To show the energy conservation, we need to calculate directly the Poisson bracket oper-
ators using the Hamiltonian (55). We then find

{H, H}(x(t),p(t)) = {H, H}(X(t),

b Pb(t)) = 0 . (66)

In this calculation, we used

  Z 1 
∂H ∂H α p(t) b p(t) p(t) b
= dλ2(x(t) − λδx(t) ) − X(t) − X(t)
∂x(t) ∂p(t) 2 0 m m m
α n b b
o
= p(t)(x(t) − X(t)) + (x(t) − X(t))p(t) . (67)
2m
Therefore the total energy of this model is conserved.

D. commutation relations

At the initial time t = 0, the c-number and q-number particles behave as the independent
particles, satisfying the commutation relations (57). Because of the interaction, however,
these commutation relations are modified.
The modified commutation relations are obtained from the solutions of the quantum
canonical equations, (61) and (62). The commutation relations for the pairs of the canonical
variables are given by

bG (t), PbG (t)] = M

[X i~ , (68)
m+M
m
[b
q (t), pbq (t)] = i~ . (69)
m+M
 16

The right-hand sides of these equations are characterized by different constants.

The commutation relations for the positions and for the momenta are
 r  r r 
bG (t)] = −i~ α µ α
[b
q (t), X t cos t − sin t , (70)
m+M µ α µ
r 
b √ α
pq (t), PG (t)] = i~ αµ sin
[b t , (71)
µ
respectively. Because of these commutation relations, the condition (32) is not satisfied.
Therefore the Poisson bracket operator cannot be represented by the commutator and thus
the quantum canonical equation does not coincide with the Heisenberg equation. That
is, the time evolution of operators cannot be represented by using the unitary operator.
Therefore this toy model is described only in the Heisenberg picture and the corresponding
Schrödinger picture is not defined.
We require attention to the interpretation of these commutation relations. In quantum
mechanics, simultaneous observables are represented by commutative operators. If this
interpretation is applied to our model, Eq. (70) means that the center of mass and relative
coordinates are not simultaneously observable. This interpretation is difficult to understand
because both coordinates are known to be simultaneous observables both in classical and
quantum mechanics. It should be however noted that the quantum-mechanical relation
between observables and commutativity is not directly applicable to the present toy model,
because wave functions and simultaneous eigenstates are not defined at t > 0. Thus we need
further study to define the simultaneous measurement in this model.

VII. CONCLUDING REMARKS

We introduced the Poisson bracket operator using the operator derivative defined in quan-
tum analysis. This operator is an alternative quantum counterpart of the Poisson bracket
in classical mechanics and there are at least three advantages compared to the commutator.
The operator derivative behaves as the standard derivative in the application to c-numbers
and thus the Poisson bracket operator coincides with the Poisson bracket in the classical
limit. We further showed that the time differential equation of operators is represented by
using the Poisson bracket operator when there exists the time generator of canonical vari-
ables. This quantum canonical equation agrees with the classical canonical equation in the
classical limit. This clear correspondence to classical mechanics is the first advantage of
 17

the introduction of the Poisson bracket operator. In the standard applications of quantum
mechanics, the Poisson bracket operator is expressed in terms of the commutator and then
the quantum canonical equation is equivalent to the Heisenberg equation. Moreover, the
quantum canonical equation is applicable to c-number canonical variables and then coin-
cides with the classical canonical equation. The second advantage is that the Poisson bracket
operator enables us to describe classical and quantum behaviors in a unified way.
The third advantage is that the quantum canonical equation is applicable to the system
where the application of the Heisenberg equation is not clear. As an example, we considered
a toy model where a c-number and a q-number particles start to interact at an initial time.
The behaviors of the particles in this model are reasonable. Indeed the differential equations
for the two particles satisfy the Ehrenfest theorem and are decomposed into the classical
canonical equation for the c-number particle and the Heisenberg equation for the q-number
particle in the vanishing interaction limit. Moreover the conserved energy of the system is
defined by the expectation value of the Hamiltonian operator.
If we identify the c-number and q-number particles of our toy model with, respectively,
the classical and quantum particles, this model may be regarded as one of quantum-classical
hybrids [17–20]. The description of our model is however incomplete and its consistency
is still controversial. For example, our model is described in the Heisenberg picture but
the corresponding Schrödinger picture is not defined. It is because the quantum canonical
equations do not agree with the Heisenberg equations in this model and the time evolution
of operators is not represented by the unitary operator. As a result, the conservation of
probability is not confirmed and then simultaneous observables are not defined.
The invariance for the canonical transformation is known to be another important prop-
erty of the Poisson bracket in classical mechanics, but the corresponding property in the
Poisson bracket operator is not yet known. As a matter of fact, the property of the canoni-
cal transformation in quantum mechanics is not well understood. See Ref. [2] and references
therein. One of the reasons of this difficulty is attributed to the fact that quantum mechan-
ics is not necessarily defined for arbitrary generalized coordinate systems. For example, in
polar coordinates, we need the operator representation of angle to describe the position of a
particle. However it is difficult to define the angle operator because there is no self-adjoint
multiplicative operator which satisfies the periodicity and the canonical commutation rela-
tion simultaneously. This is the origin of the famous paradox in the angular uncertainty
 18

relation. See Ref. [21] and references therein.

These problems are left as future works.

The author thanks T. Kodama for useful comments and acknowledges the financial sup-
port by CNPq (303468/2018-1). A part of the work was developed under the project INCT-
FNA Proc. No. 464898/2014-5.

[1] As a review paper of the deformation quantization, see G. Dito, D. Sternheimer, in: G.
Halbout (Ed.), Deformation Quantization, in: IRMA lectures in mathematics and theoretical
physics, 1, (Walter de Gruyter, Berlin, New York, 2002) pp. 9–54.
[2] M. Blaszak and Z. Domański, “Canonical transformations in quantum mechanics”, Ann. Phys.
331, 70 (2013).
[3] E. Hille and R. S. Phillips, Functional Analysis and Semi-Groups 31 (USA, AMS ,1957).
[4] M. Suzuki, “- Quantum Analysis — Non-Commutative Differential and Integral Calculi”,
Commun. Math. Phys. 183 339 (1997).
[5] M. Suzuki, “Quantum Analysis, Nonequilibrium Density Matrix and Entropy Operator”, Int.
J. Mod. Phys. B 10 1637 (1996).
[6] M. Suzuki, “Compact exponential product formulas and operator functional derivative”, J.
Math. Phys. 3 1183, (1997).
[7] M. Suzuki, “– Algebraic formulation of quantum analysis and its applications to Laurent series
operator functions”, Phys. Lett. A 224 337 (1997).
[8] M. Suzuki, “Quantum Analysis and Nonequilibrium Responses”, Prog. Theor. Phys. 100 475
(1998).
[9] M. Suzuki, “General Formulation of Quantum Analysis”, Rev. Math. Phys. 11 243 (1999).
[10] M. Suzuki, “Refined formulation of quantum analysis, q-derivative and exponential splitting”,
J. Phys. A: Math. Gen. 39, 5617 (2006).
[11] W. Majewski and M. Marciniak, “On quantum Lyapunov exponents”, J. Phys. A: Math. Gen.
39, L523 (2006).
[12] T. Koide “Memory effect in the upper bound of the heat flux induced by quantum fluctua-
 19

tions”, Phys. Rev. E94, 042140 (2016).

[13] T. Koide, “Quantum analysis and thermodynamic operator relations in stochastic energetics”,
J. Phys. : Conf. Ser. 880, 012037 (2017).
[14] M. Ishihara, “Derivation of density operators for generalized entropies with quantum analysis”,
Physica A 543, 123419 (2020).
[15] F. M. Ciaglia, “Quantum states, groups and monotone metric tensors”, Eur. Phys. J. Plus
135, 530 (2020).
[16] E. Castro-Ruis, F. Giacomini, A. Belenchia et al., “Quantum clocks and the temporal local-
isability of events in the presence of gravitating quantum systems”, Nat. Commun. 11, 2672
(2020).
[17] M. J. Hall and M. Reginatto, “Interacting classical and quantum ensembles”, PhysṘev. A72,
06210 (2005).;
[18] H.-T. Elze, “Linear dynamics of quantum-classical hybrids”, Phys. Rev. A85, 052109 (2012).
[19] N. Burić, I. Mendaš, D. B. Popović and et al., “Statistical ensembles in the Hamiltonian
formulation of hybrid quantum-classical systems”, Phys. Rev. A86, 034104 (2012).
[20] T. Koide, “Classicalization of quantum variables and quantum–classical hybrids”, Phys. Lett.
A379, 2007 (2015).
[21] J.-P. Gazeau and T. Koide, “Uncertainty relation for angle from a quantum-hydrodynamical
perspective”, Ann. Phys. 416, 168159 (2020).