Acta Polytechnica


doi:10.14311/AP.2013.53.0462
Acta Polytechnica 53(5):462–469, 2013 © Czech Technical University in Prague, 2013

available online at http://ojs.cvut.cz/ojs/index.php/ap

gln+1 ALGEBRA OF MATRIX DIFFERENTIAL OPERATORS AND
MATRIX QUASI-EXACTLY-SOLVABLE PROBLEMS

Yuri F. Smirnov (deceased), Alexander V. Turbiner∗

Instituto de Ciencias Nucleares, Universidad Nacional Autónoma de México, Apartado Postal 70-543, 04510
México, D.F., Mexico

∗ corresponding author: turbiner@nucleares.unam.mx

Abstract. The generators of the algebra gln+1 in the form of differential operators of the first order
acting on Rn with matrix coefficients are explicitly written. The algebraic Hamiltonians for matrix
generalization of 3−body Calogero and Sutherland models are presented.

Keywords: algebra of differential operators, exactly-solvable problems.

Submitted: 13 May 2013. Accepted: 6 June 2013.

1. Introduction
This work has a certain history related to Miloslav
Havlicek. On the important occasion of Miloslav’s
75th birthday, we think this story should be re-
vealed. About 25 years ago, when quasi-exactly-
solvable Schroedinger equations with the hidden alge-
bra sl2 were discovered [1], one of the present authors
(AVT) approached Israel M. Gelfand and asked about
the existence of the algebra gln+1 of matrix differ-
ential operators. Instead of giving an answer, Israel
Moiseevich said that M. Havlicek knows the answer
and that he must be asked. A set of Dubna preprints
was given (see [2, 3] and reference therein). Then
AVT studied them for many years, at first separately
and then together with the first author (YuFS), who
also happened to have the same set of preprints. The
results of these studies are presented below. While
carrying out these studies, we always kept in mind
that a constructive answer exists and is known to
Miloslav. Thus, we are certain that at least some of
results presented here are known to Miloslav. Hav-
ing difficulty to understand what is written in the
texts we did not know what he really knew, and were
therefore unable to indicate it in our text. Our main
goal is to find a mixed representation of the algebra
gln+1 which contains both matrices and differential
operators in a non-trivial way. Then to generalize it
to a polynomial algebra which we call g(m) (see below,
Section 4). Another goal is to apply the obtained rep-
resentations for a construction of the algebraic forms
of (quasi)-exactly-solvable matrix Hamiltonians.

2. The algebra gln in mixed
representation

Let us take the algebra gln and consider the vector
field representation

Ẽij = xi∂j, i,j = 1, . . .n,x ∈ Rn. (1)

It obeys the canonical commutation relations

[Ẽij, Ẽkl] = δjkẼil − δilẼkj. (2)

On the other hand, let us consider another representa-
tion Mpm, p,m = 1, . . . ,n of the algebra gln in terms
of some operators (matrix, finite-difference, etc) with
the condition that all ‘cross-commutators’ between
these two representations vanish

[Ẽij,Mpm] = 0. (3)

Let us choose Mpm to obey the canonical commutation
relations

[Mij,Mkl] = δjkMil − δilMkj, (4)

(cf. (2)). It is evident that the sum of these two
representations is also the representation,

Eij ≡ Ẽij + Mij ∈ gln. (5)

Now we consider an embedding of gln ⊂ gln+1 try-
ing to complement the representation (1) of the alge-
bra gln up to the representation of the algebra gln+1.
In principle, this can be done due to the existence of
the Weyl-Cartan decomposition,

gln+1 = L⊕ (gln ⊕ I) ⊕U

with the property

gln+1 = L o (gln ⊕ I) n U, (6)

where L(U) is the commutative algebra of the lowering
(raising) generators with the property [L,U] = gln⊕I.
Thus, it realizes a property of the Gauss decomposition
of gln+1. It is worth emphasizing that dim(L) =
dim(U) = n.

Obviously, the lowering generators (of negative grad-
ing) from L can be given by derivations

T −i = ∂i, i = 1, . . . ,n, ∂i ≡
∂

∂xi
, (7)

(see e.g. [5]) when assuming that all commutators

[T −i ,Mpm] = 0, (8)

462

http://dx.doi.org/10.14311/AP.2013.53.0462
http://ojs.cvut.cz/ojs/index.php/ap


vol. 53 no. 5/2013 gln+1 Algebra of Matrix Differential Operators

vanish. This probably implies that the only possible
choice for Mpm exists when they are either given by
matrices or act in a space which is a complement to
x ∈ Rn. It is easy to check that

[Eij,T −k ] = −δikT
−
j .

Now we have to add the Euler-Cartan generator of
the gln algebra, see (6)

−E0 =
n∑
j=1

xj∂j −k, (9)

where k is arbitrary constant. Raising generators from
U are chosen as

−T +i = −xiE0 +
n∑
j=1

xjMij

= xi
( n∑
j=1

xj∂j −k
)

+
n∑
j=1

xjMij, i = 1, . . . ,n.

(10)

(cf. for instance [5]). Needless to say that one can check
explicitly that T −i , Eij, E0, T

+
i span the algebra gln+1.

In particular,
[E,T +] = T +,

and
[T +i ,T

−
j ] = Eii −δijE0.

If parameter k takes non-negative integer the alge-
bra gln+1 spanned by the generators (5), (7), (9), (10)
appears in a finite-dimensional representation. There
exists a linear finite-dimensional space of polynomials
of finite-order in the space of columns/spinors of finite
length which is a common invariant subspace for all
generators (5), (7), (9), (10). This finite-dimensional
representation is irreducible.

The non-negative integer parameter k has the mean-
ing of the length of the first row of the Young tableau
of gln+1, describing a totally symmetric representa-
tion (see below). All other parameters are coded in
Mij, which corresponds to an arbitrary Young tableau
of gln. Thus, we have some peculiar splitting of the
Young tableau.

Each representation is characterized by the Gelfand-
Tseitlin signature, [m1,n, . . .mnn], where min ≥
mi+1,n and their difference is positive integer. Each
basic vector is characterized by the Gelfand-Tseitlin
scheme. An explicit form of the representation is given
by the Gelfand-Tseitlin formulas [4].

It can be demonstrated that all Casimir operators of
gln+1 in this realization (5), (7), (9), (10) are expressed
in Mij, and thus do not depend on x. They coincide
with the Casimir operators of the gln-subalgebra real-
ized by matrices Mij.

3. Example: the algebra gl3 in
mixed representation

In the case of the algebra gl3, the generators (5), (7),
(9), (10) take the form

E11 = x1∂1 + M11, E22 = x2∂2 + M22,
E12 = x1∂2 + M12, E21 = x2∂1 + M21,
E0 = k −x1∂1 −x2∂2,
T −1 = ∂1, T

−
2 = ∂2,

T +1 = x1(k −x1∂1 −x2∂2) −x1M11 −x2M12,
T +2 = x2(k −x1∂1 −x2∂2) −x1M21 −x2M22. (11)

The Casimir operators of gl3 in this realization are
given by

C1 = E11 + E22 + E0 = k + M11 + M22 = k + C1(M),
C2 = E12E21 + E21E12 + T +1 T

−
1 + T

−
1 T

+
1 + T

+
2 T

−
2

+ T −2 T
+
2 + E

2
11 + E

2
22 + E

2
0 = k(k + 2)

+ M211 + M
2
22 + M12M21 + M21M12

−M11 −M22 = k(k + 2) + C2(M) −C1(M),

and, finally,

C3 = −
1
2
C31 +

3
2
C1C2 + 3C2 − 2C21 − 2C1.

In this realization, the Casimir operator C3 is alge-
braically dependent on C1 and C2. In fact, C1 and
C2 are nothing but the Casimir operators of the gl2
sub-algebra. Therefore, the center of the gl3 universal
enveloping algebra in realization (11) is generated by
the Casimir operators of the gl2 sub-algebra realized
by Mij. Thus, it seems natural that these reps are
irreducible.

Now we consider concrete matrix realizations of the
gl2-subalgebra in our scheme.

3.1. Reps in 1 × 1 matrices
This corresponds to the trivial representation of gl2,

M11 = M12 = M21 = M22 = 0.

This is [k, 0] or, in other words, a symmetric represen-
tation (the Young tableau has two rows of length k
and 0, correspondingly). We also can call it a scalar
representation, since the generators

E11 = x1∂1, E22 = x2∂2,
E12 = x1∂2, E21 = x2∂1,
E0 = k −x1∂1 −x2∂2,
T −1 = ∂1, T

−
2 = ∂2,

T +1 = x1(k −x1∂1 −x2∂2),
T +2 = x2(k −x1∂1 −x2∂2), (12)

act on one-component spinors or, in other words, on
scalar functions (see e.g. [5]). The Casimir operators
are:

C1 = k, C2 = k(k + 2).

463



Yu. F. Smirnov, A. V. Turbiner Acta Polytechnica

If parameter k takes non-negative integer the algebra
gl3 spanned by the generators (12) appears in finite-
dimensional representation. Its finite-dimensional rep-
resentation space is a space of polynomials

Pk,0 =
〈
x1
p1x2

p2
∣∣ 0 ≤ p1 +p2 ≤ k〉, k = 0, 1, 2, . . . .

(13)
Namely in this representation (12), the algebra gl3

appears as the hidden algebra of the 3-body Calogero
and Sutherland models [5], BC2 rational and trigono-
metric, and G2 rational models [6] and even of the
BC2 elliptic model [7].

3.2. Reps in 2 × 2 matrices
Take gl2 in two-dimensional reps by 2 × 2 matrices,

M11 =
(

1 0
0 0

)
, M22 =

(
0 0
0 1

)
,

M12 =
(

0 1
0 0

)
, M21 =

(
0 0
1 0

)
,

Then the generators (11) of gl3 are:

T −1 =
(
∂1 0
0 ∂1

)
, T −2 =

(
∂2 0
0 ∂2

)
,

E11 =
(
x1∂1+1 0

0 x1∂1

)
, E12 =

(
x1∂2 1

0 x1∂2

)
,

E21 =
(
x2∂1 0

1 x2∂1

)
, E22 =

(
x2∂2 0

0 x2∂2+1

)
,

E0 =
(
A 0
0 A

)
,

T +1 =
(
x1(A− 1) −x2

0 x1A

)
,

T +2 =
(
x2A 0
−x1 x2(A− 1)

)
, (14)

where A = k−x1∂1−x2∂2. This is [k, 1]-representation
(the Young tableau has two rows of length k and 1,
correspondingly), and their Casimir operators are:

C1 = k + 1, C2 = (k + 1)2.

If parameter k takes non-negative integer the algebra
gl3 spanned by the generators (14) appears in finite-
dimensional representation.

Let us consider several different values of k in detail.

The case k = 1. Then three-dimensional represen-
tation space V (2)1 appears to be spanned by:

P− =
[
0
1

]
, P+ =

[
1
0

]
, Y1 =

[
x2
−x1

]
. (15)

This corresponds to antiquark multiplet in standard
(fundamental) representation. The Newton polygon
is a triangle with points P± as vortices at the base.

Figure 1. Newton hexagon for the representation
space V (2)4 of the [4, 1]-representation of dimension
24.

The case k = 2. Then eight-dimensional represen-
tation space V (2)2 appears to be spanned by:

P− =
[
0
1

]
, P+ =

[
1
0

]
, P

(1)
− =

[
0
x2

]
,

Y
(1)

1 =
[

0
x1

]
, Y

(2)
1 =

[
x2
0

]
, P

(1)
+ =

[
x1
0

]
,

Y2 =
[

x22
−x1x2

]
, Y3 =

[
x1x2
−x21

]
. (16)

This corresponds to octet in standard (fundamental)
representation. Space V (2)2 contains V

(2)
1 as a sub-

space, V (2)1 ⊂ V
(2)

2 . It should be mentioned that
Y1 = −Y

(1)
1 + Y

(2)
1 . Now the Newton polygon is a

hexagon where the central point is doubled, being
presented by Y (1,2)1 , and the lower (upper) base has
length two being given by P± (Y2,3).

The case k = 3. The representation space V (2)3 is
15-dimensional. In addition to P±,P

(1)
± and Y

(1,2)
1

(see (15) and (16)), it contains several vectors more,
namely,

P
(2)
− =

[
0
x22

]
, P

(2)
+ =

[
x21
0

]
, (17)

which are situated on the ±-sides of the Newton
hexagon, doubling the points corresponding to Y2,3
(see (16))

Y
(1)

2 =
[

0
x1x2

]
, Y

(2)
2 =

[
x22
0

]
,

Y
(1)

3 =
[

0
x21

]
, Y

(2)
3 =

[
x1x2

0

]
, (18)

plus three extra vectors on the boundary

Y8 =
[

x32
−x1x22

]
, Y9 =

[
x1x

2
2

−x21x2

]
, Y10 =

[
x21x2
−x31

]
.

(19)
It is clear that V (2)1 ⊂ V

(2)
2 ⊂ V

(2)
3 . All internal points

of the Newton hexagon are double points, while the
points on the boundary are single ones.

464



vol. 53 no. 5/2013 gln+1 Algebra of Matrix Differential Operators

The general case. The finite-dimensional repre-
sentation space V (2)k has dimension k(k + 2) and is
presented by the Newton hexagon, which contains
(k + 1) horizontal layers. The lower base has length
two, while the upper base has length k (see Fig. 1 as
an illustration for k = 4). All internal points of the
Newton hexagon are double points, while the points
on the boundary are single ones. Except for k vectors
of the last (highest) layer of the Newton hexagon, the
remaining k(k + 1) vectors span the space of all pos-
sible two-component spinors with components given
by the inhomogeneous polynomials in x1,x2 of degree
not higher than (k − 1). We denote this space as
Ṽ

(2)
k ⊂ V

(2)
k . The non-trivial task is to describe k vec-

tors of the last (highest) layer of the hexagon. After
some analysis one can find that they have the form

Yk(k+1)+i =
[

xk−i2 x
i
1

−xk−i−12 x
i+1
1

]
,

i = 0, 1, 2, . . . , (k − 1), (20)

hence they span a non-trivial k-dimensional subspace
of spinors with components given by specific homoge-
neous polynomials of degree k.

3.3. Reps in 3 × 3 matrices
Take gl2 in three-dimensional reps by 3 × 3 matrices,

M11 =


2 0 00 1 0

0 0 0


 , M22 =


0 0 00 1 0

0 0 2


 ,

M12 =


0

√
2 0

0 0
√

2
0 0 0


 , M21 =


 0 0 0√2 0 0

0
√

2 0


 .

Then the generators (11) of gl3 are:

T −1 =


∂1 0 00 ∂1 0

0 0 ∂1


 , T −2 =


∂2 0 00 ∂2 0

0 0 ∂2


 ,

E11 =


x1∂1 + 2 0 00 x1∂1 + 1 0

0 0 x1∂1


 ,

E12 =


x1∂2

√
2 0

0 x1∂2
√

2
0 0 x1∂2


 ,

E21 =


x2∂1 0 0√2 x2∂1 0

0
√

2 x2∂1


 ,

E22 =


x2∂2 0 00 x2∂2 + 1 0

0 0 x2∂2 + 2


 ,

E0 =


A 0 00 A 0

0 0 A


 ,

T +1 =


x1(A− 2) −

√
2x2 0

0 x1(A− 1) −
√

2x2
0 0 x1A


 ,

T +2 =


 x2A 0 0−√2x1 x2(A− 1) 0

0 −
√

2x1 x2(A− 2)


 , (21)

where A = k−x1∂1−x2∂2. This is [k, 2]-representation
(the Young tableau has two rows of length k and 2,
correspondingly) and their Casimir operators are:

C1 = k + 2, C2 = (k + 1)2 + 3.

As an illustration let us explicitly show finite-
dimensional representation spaces for k = 2, 3.

The case k = 2. Then the six-dimensional repre-
sentation space V (3)2 appears to be spanned by:

P− =


00

1


 , P0 =


01

0


 , P+ =


10

0


 ,

Y1 =


 0x2
−
√

2x1


 , Y2 =


−
√

2x2
x1
0


 ,

Y3 =


 x22−√2x1x2

x21


 . (22)

This corresponds to ‘di-antiquark’ multiplet.

The case k = 3. Then 15-dimensional representa-
tion space V (3)3 appears to be spanned by:

P− =


00

1


 , P0 =


01

0


 , P+ =


10

0


 ,

Y
(1)

1 =


 0x2

0


 , Y (2)1 =


 00
x1


 , Y (1)2 =


x20

0


 ,

Y
(2)

2 =


 0x1

0


 , P (1)− =


 00
x2


 , P (1)+ =


x10

0


 ,

Y
(1)

3 =


−
√

2x22
x1x2

0


 , Y (2)3 =


 0x1x2
−
√

2x21


 ,

Y4 =


 0−√2x22

2x1x2


 , Y5 =


 2x1x2−√2x21

0


 ,

Y6 =


 x32−√2x1x22

x21x2


 , Y7 =


 x1x22−√2x21x2

x31


 . (23)

It is worth mentioning that as a consequence of
a particular realization of the generators (11) of the
gl3 algebra there exist a certain relations between
generators other than those given by the Casimir
operators. The first observation is that there are no
linear relations between generators of such a type.
Some time ago nine quadratic relations were found

465



Yu. F. Smirnov, A. V. Turbiner Acta Polytechnica

m s0

1/2

1

3/2

2

5/2

3

l/2

Figure 2. Verma module with the lowest weight (057) for s = 5/2.

between gl3 generators taken in scalar representation
(12) other than Casimir operators [8]. Surprisingly,
certain modifications of these relations also exist for
[kn] mixed representations (11),

−T +1 E22 + T
+
2 E12 = x1

[
M22x1∂1

+ M11x2∂2 + (M11 −k)M22 −M21M12
]

−x2(x1∂1 −k − 1)M12 −M21x21∂2 ≡−T̃
+
1 ,
(24)

−T +2 E11 + T
+
1 E21 = x2

[
M22x1∂1

+ M11x2∂2 + (M22 −k)M11 −M12M21
]

−x1(x2∂2 −k − 1)M21 −M12x22∂1 ≡−T̃
+
2 ,
(25)

−E12(E0 + 1) + T +1 T
−
2

= M12(x1∂1 −k − 1) −M11x1∂2 ≡−Ẽ12, (26)
−E21(E0 + 1) + T +2 T

−
1

= M21(x2∂2 −k − 1) −M22x2∂1 ≡−Ẽ21, (27)
T +1 T

−
1 −E11(1 + E0) = M11x2∂2
−M12x2∂1 − (k + 1)M11 ≡−Ẽ11, (28)

T +2 T
−
2 −E22(1 + E0) = M22x1∂1
−M21x1∂2 − (k + 1)M22 ≡−Ẽ22, (29)

E12E21 −E11E22 −E11 = M12x2∂1 + M21x1∂2
−M22x1∂1 −M11x2∂2 + M12M21

−M11M22 −M11 ≡−Ê11, (30)
E22T

−
1 −E21T

−
2 = M22∂1 −M21∂2 ≡ T̃

−
1 , (31)

E12T
−
1 −E11T

−
2 = M12∂1 −M11∂2 ≡−T̃

−
2 . (32)

Not all these relations are independent. It can be
shown that one relation is linearly dependent, since
the sum of (28) + (29) + (30) gives the second Casimir
operator C2.

In scalar case, at least, we can assign a natural (vec-
torial) grading to the generators. The above relations
also reflect a certain decomposition of the gradings,

(1, 0)(0, 0) = (0, 1)(1,−1)
(0, 1)(0, 0) = (1, 0)(−1, 0)

for the first two relations,

(1,−1)(0, 0) = (1, 0)(0,−1)
(−1, 1)(0, 0) = (0, 1)(−1, 0)

for the second two,

(1, 0)(−1, 0) = (0, 0)(0, 0)
(0, 1)(0,−1) = (0, 0)(0, 0)

(1,−1)(−1, 1) = (0, 0)(0, 0)

for three before the last two, and

(0, 0)(−1, 0) = (−1, 1)(0,−1)
(0, 0)(0,−1) = (1,−1)(−1, 1)

for the last two.

4. Algebra g(m) in mixed
representation

The basic property which was used to construct the
mixed representation of the algebra gln+1 is the ex-
istence of the Weyl-Cartan decomposition gln+1 =
L⊕ (gln ⊕ I) ⊕U with property (6). One can pose a
question about the existence of other algebras than
gln+1 for which the Weyl-Cartan decomposition with
property (6) holds. The answer is affirmative. Let us
consider the important particular case of the Cartan

466



vol. 53 no. 5/2013 gln+1 Algebra of Matrix Differential Operators

algebra gl2 ⊕ I, and construct a realization of a new
algebra denoted g(m) with the property

g(m) = Lm+1 o (gl2 ⊕ I) n Um+1, (33)

where Lm(Um) is the commutative algebra of the
lowering (raising) generators with the property
[Lm,Um] = Pm−1(gl2 ⊕I) with Pm−1 as a polynomial
of degree (m − 1) in generators of gl2 ⊕ I. Thus, it
realizes a property of the generalized Gauss decompo-
sition. The emerging algebra is a polynomial algebra.
It is worth emphasizing that the realization we are go-
ing to construct appears at dim(Lk) = dim(Um) = m.
For m = 1 the algebra g(1) = gl3, see (6). Our final
goal is to build the realization of (33) in terms of finite
order differential operators acting on the plane R2.
The simplest realization of the algebra gl2 by dif-

ferential operators in two variables is the vector field
representation, see (1) at n = 2. Exactly this rep-
resentation was used to construct the representation
of the gl3 algebra acting of R2, see (11), (12). In
this case dim(Lm) = dim(Um) = 2. We are unable
to find other algebras with dim(Lm) = dim(Um) > 2.
However, there exists another representation of the
algebra gl2 by the first order differential operators in
two variables,

J̃12 = ∂x,

J̃
(k)
11 = −x∂x +

k

3
,

J̃
(k)
22 = −x∂x + sy∂y,

J̃
(k)
21 = x

2∂x + sxy∂y −kx, (34)

(see S. Lie, [9] at k = 0 and A. González-Lopéz et
al, [10] at k 6= 0 (Case 24)), where s,k are arbitrary
numbers. These generators obey the standard commu-
tation relations (2) of the algebra gl2 in the vector field
representation (1). It is evident that the sum of the
two representations, J̃ij and the matrix representation
Mij, is also a representation,

Jij ≡ J̃ij + Mij ∈ gl2. (35)

(cf. (5)). It is worth mentioning that the gl2 algebra
commutation relations for Mpm are taken in a canon-
ical form (4). The unity generator I in (33) is written
in the form of a generalized Euler-Cartan operator

J
(k)
0 = x∂x + sy∂y −k. (36)

Now let us assume that s is non-negative integer,
s = m, m = 0, 1, 2, . . .. Evidently, the lowering gener-
ators (of negative grading) from Lm+1 can be given
by

T −i = x
i∂y, i = 0, 1, . . . ,m, (37)

forming commutative algebra

[T −i ,T
−
j ] = 0. (38)

(cf. [9, 10]). Eventually, the generators of the algebra
(gl2 ⊕ I) n Lm+1 take the form

J12 = ∂x + M12,

J
(k)
11 = −x∂x +

k

3
+ M11,

J
(k)
22 = −x∂x + my∂y + M22,

J
(k)
21 = x

2∂x + mxy∂y −kx + M21, (39)

with J(k)0 and T
−
i given by (36) and (37), respectively.

Let us consider two particular cases of the general
construction of the raising generators for the commu-
tative algebra U.

Case 1. For the first case we take the trivial matrix
representation of the gl2,

M11 = M12 = M21 = M22 = 0.

One can check that one of the raising generators is
given by

U0 = y∂mx , (40)

while all other raising generators are multiple commu-
tators of J(k)21 with U0,

Ui ≡ [J
(k)
21 , [J

(k)
21 , [· · ·J

(k)
21 ,T0] · · · ] ]︸ ︷︷ ︸

i

= y∂m−ix J
(k)
0 (J

(k)
0 + 1) . . . (J

(k)
0 + i− 1) , (41)

at i = 1, . . .m. All of them are differential operators
of fixed degree m. The procedure for construction of
the operators Ui has the property of nilpotency:

Ui = 0, i > m.

In particular, for m = 1,

U0 = y∂x, U1 = yJ
(k)
0 = y(x∂x + y∂y −k).

Inspecting the generators T −0,1, Jij, J
(n), U0,1 one can

see that they span the algebra gl3, see (12). Hence,
the algebra g(1) ≡ gl3.

If parameter k takes non-negative integer the alge-
bra g(m) spanned by the generators (39), (40), (41)
appears in finite-dimensional representation. Its finite-
dimensional representation space is a triangular space
of polynomials

Pk,0 =
〈
xp1yp2

∣∣ 0 ≤ p1 + mp2 ≤ k〉,
k = 0, 1, 2, . . . . (42)

Namely in this representation, the algebra g(m) ap-
pears as a hidden algebra of the 3-body G2 trigono-
metric model [6] at m = 2 and of the so-called TTW
model at integer m, in particular, of the dihedral
I2(m) rational model [11].

467



Yu. F. Smirnov, A. V. Turbiner Acta Polytechnica

Case 2. The second case is a certain evident exten-
sion when generators Mij are of an arbitrary matrix
representation of the algebra gl2. Raising generators
(40), (41) remain raising generators even if Cartan
generators are given by (39) with arbitrary Mij ∈ gl2.
However, the algebra is not closed: [T,U] 6= P (gl2⊕I).
It can be fixed, at least, for the case m = 1. If Mij are
generators of gl2 subalgebra of gl3. By adding to T,U
generators (38), (40), (41) the appropriate matrix gen-
erators from gl3, the algebra gets closed. We end up
with the gl3 algebra of matrix differential operators
other than (11). We are not aware of a solution to
this problem for the case of m 6= 1 except for the case
of trivial matrix representation, see Case 1.

5. Extension of the 3-body
Calogero Model

The first algebraic form for the 3-body Calogero Hamil-
tonian [12] appears after gauge rotation with the
ground state function, separation of the center-of-
mass and changing the variables to elementary sym-
metric polynomials of the translationally-symmetric
coordinates [5],

hCal = −2τ2∂2τ2τ2 − 6τ3∂
2
τ2τ3

+
2
3
τ22 ∂

2
τ3τ3

−
[
4ωτ2 + 2(1 + 3ν)

]
∂τ2 − 6ωτ3∂τ3. (43)

These new coordinates are polynomial invariants of
the A2 Weyl group. Its eigenvalues are

− �p = 2ω(2p1 + 3p2), p1,2 = 0, 1, . . . . (44)

As is shown in Ruhl and Turbiner [5], the operator
(43) can be rewritten in a Lie-algebraic form in terms
of gl(3)-algebra generators of the representation [k, 0].
The corresponding expression is

hCal = −2E11T −1 − 6E22T
−
1 +

2
3
E12E12

− 4ωE11 − 2(1 + 3ν)T −1 − 6ωE22 . (45)

Now we can substitute the generators of the represen-
tation [k,n] in the form (11)

h̃Cal = −2τ2∂2τ2τ2 − 6τ3∂
2
τ2τ3

+
2
3
τ22 ∂

2
τ3τ3

− 2
[
2ωτ2 + (1 + 3ν) + (n− 2M22)

]
∂τ2

−
(

6ωτ3 −
4
3
M12τ2

)
∂τ3 +

2
3
M12M12

− 4ωn− 2ωM22. (46)

This is an n×n matrix differential operator. It con-
tains infinitely many finite-dimensional invariant sub-
spaces which are nothing but finite-dimensional repre-
sentation spaces of the algebra gl(3). This operator
remains exactly-solvable with the same spectra as the
scalar Calogero operator.
This operator probably remains completely inte-

grable. A higher-than-second-order integral is the

differential operator of the sixth order (ω 6= 0) or of
the third order (ω = 0), which takes an algebraic form
after gauging away the ground state function in τ
coordinates. It can be rewritten in terms of the gl(3)-
algebra generators of the representation [k, 0], which
then can be replaced by the generators of the repre-
sentation [k,n]. Under such a replacement the spectra
of the integral remain unchanged and algebraic.

6. Extension of the 3-body
Sutherland Model

The first algebraic form for the 3-body Sutherland
Hamiltonian [13] appears after gauge rotation with the
ground state function, separation of the center-of-mass
and changing the variables to elementary symmetric
polynomials of the exponentials of translationally-
symmetric coordinates [5],

hSuth = −
(

2η2 +
α2

2
η22 −

α4

24
η23

)
∂2η2η2

−
(

6 +
4α2

3
η2

)
η3∂

2
η2η3

+
(2

3
η22 −

α2

2
η23

)
∂2η3η3

−
[
2(1+3ν)+2

(
ν+

1
3

)
α2η2

]
∂η2−2

(
ν+

1
3

)
α2η3∂η3,

(47)

where α is the inverse radius of the circle on which
the bodies are situated. These new coordinates are
fundamental trigonometric invariants of the A2 Weyl
group.
As shown in [5], operator (47) can be rewritten

in a Lie-algebraic form in terms of the gl(3)-algebra
generators of the representation [k, 0],

hSuth = −2E11T −1 − 6E22T
−
1 +

2
3
E12E12

−2(1 + 3ν)T −1 +
α4

24
E21E21−

α2

6

[
3E11E11 + 8E11E22

+ 3E22E22 + (1 + 12ν)(E11 + E22)
]
. (48)

Now we can substitute the generators of the represen-
tation [k,n] in the form (11)

h̃Suth = −
(

2η2 +
α2

2
η22 −

α4

24
η23

)
∂2η2η2

−
(

6 +
4α2

3
η2

)
η3∂

2
η2η3

+
(2

3
η22 −

α2

2
η23

)
∂2η3η3

− 2
[
(1 + 3ν) +

(
ν +

1
3

)
α2η2 + (n− 2M22)

]
∂η2

+
α4

24
M21η3∂η2 +

[
2
(
ν +

1
3

)
α2η3 −

4
3
M12η2

]
∂η3

−
α2

3

[
3n(η2∂η2 + η3∂η3 ) + M11η3∂η3 + M22η2∂η2

]
+

2
3
M12M12 +

α4

24
M21M21

−
α2

6

[
2M11M22 + (1 + 12ν + 3n)n

]
. (49)

468



vol. 53 no. 5/2013 gln+1 Algebra of Matrix Differential Operators

This is an n×n matrix differential operator. It con-
tains infinitely-many finite-dimensional invariant sub-
spaces which are nothing but finite-dimensional repre-
sentation spaces of the algebra gl(3). This operator
remains exactly-solvable with the same spectra as the
scalar Sutherland operator.

The operator (49) probably remains completely in-
tegrable. A non-trivial integral is the differential op-
erator of the third order, it takes the algebraic form
after gauging away the ground state function in η
coordinates. It can be rewritten in terms of the gl(3)-
algebra generators of the representation [k, 0], which
then can be replaced by the generators of the repre-
sentation [k,n]. Under such a replacement the spectra
of the integral remain unchanged and algebraic.

7. Conclusions
The algebra gln of differential operators plays the
role of a hidden algebra for all An,Bn,Cn,Dn,BCn
Calogero-Moser Hamiltonians, both rational and
trigonometric, with the Weyl symmetry of classical
root spaces (see [14] and references therein). We have
described a procedure which, in our opinion, should
carry the name of the Havlicek procedure, to construct
the algebra gln of the matrix differential operators.
The procedure is based on a mixed, matrix-differential
operator realization of the Gauss decomposition dia-
gram.
As for Hamiltonian reduction models with the ex-

ceptional Weyl symmetry group G2,F4,E6,7,8, both
rational and trigonometric, there exist hidden alge-
bras of differential operators (see [14] and references
therein). All these algebras are infinite-dimensional
but finitely-generated. For generating elements of
these algebras an analogue of the Weyl-Cartan de-
composition exists but in the Gauss decomposition
diagram, a commutator of the lowering and raising
generators is a polynomial of the higher-than-one or-
der in the Cartan generators. Matrix realizations of
these algebras surely exist. Thus, the above mentioned
procedure for building the mixed representations can
be realized. It may lead to a new class of matrix
exactly-solvable models with exceptional Weyl sym-
metry.

Acknowledgements
It was planned long ago to dedicate this text to Miloslav
Havlicek who has always been deeply respected by both
authors as a scientist and also as a citizen.

The text is based mainly on notes jointly prepared by
two authors. It does not include results of the authors
obtained separately (except for Section 4) and which the
authors had no chance to discuss. Thus, the text will
appear somehow incomplete. When the first author (YuFS)
passed away, it took years for the second author (AVT)
to return to the subject due to sad memories. Even now,
almost a decade after the death of Yura Smirnov, the
preparation of this text was quite difficult for AVT.

AVT thanks CRM, Montreal for their kind hospitality
extended to him. A part of this work was done there

during his numerous visits. AVT is grateful to J C Lopez
Vieyra for taking the interest in the work and for technical
assistance. This work was supported in part by the Uni-
versity Program FENOMEC, by PAPIIT grant IN109512,
and by CONACyT grant 166189 (Mexico).

References
[1] A.V. Turbiner. Quasi-Exactly-Solvable Problems and
the SL(2, R) Group, Comm.Math.Phys. 118: 467-474,
1988.

[2] C. Burdik. Realisations of the real semisimple Lie
algebras: a method of construction, J. Phys. A18:
3101-3111, 1985.

[3] C. Burdik, M. Havlicek. Boson realization of the
semi-simple Lie algebras, in Symmetry in physics: in
memory of Robert T. Sharp, CRM Proceeding 2004,
vol.34, p.87-98, edited by R.T. Sharp, P. Winternitz.

[4] I.M. Gelfand and M.L. Tsetlin. Finite-dimensional
representations of groups of orthogonal matrices, Dokl.
Akad. Nauk SSSR 71: 1017–1020, 1950 (in Russian)
English transl. in: I.M. Gelfand, Collected papers. Vol
II, Berlin: Springer-Verlag 1988, pp. 657–661.

[5] W. Rühl and A. V. Turbiner. Exact solvability of the
Calogero and Sutherland models, Mod. Phys. Lett. A10:
2213–2222, 1995 arXiv:hep-th/9506105.

[6] A.V. Turbiner. Hidden Algebra of Three-Body Integrable
Systems, Mod.Phys.Lett. A13: 1473-1483, 1998

[7] A.V. Turbiner. BC2 Lame polynomials, Talks presented
at 1085 Special Session of the American Mathematical
Society, Tucson AZ, USA (October 2012) and at the
Annual Meeting of the Canadian Mathematical Society,
Montreal, Canada (December 2012)

[8] A. V. Turbiner, Lie algebras and linear operators with
invariant subspace, in Lie algebras, cohomologies and
new findings in quantum mechanics (N. Kamran and
P. J. Olver, eds.), AMS Contemporary Mathematics,
vol. 160, pp. 263–310, 1994; arXiv:funct-an/9301001

[9] S. Lie. Gruppenregister, vol. 5, B.G. Teubner, Leipzig,
1924, 767-773

[10] A. González-Lopéz, N. Kamran and P.J. Olver.
Quasi-Exactly-Solvable Lie Algebras of the first order
differential operators in Two Complex Variables, J.Phys.
A24: 3995–4008, 1991; Lie algebras of differential
operators in two complex variables, American J. Math.
114: 1163–1185, 1992;

[11] F. Tremblay, A.V. Turbiner and P. Winternitz. An
infinite family of solvable and integrable quantum
systems on a plane, Journal of Phys. A42, 242001,
2009; 10 pp arXiv:0904.0738

[12] F. Calogero. Solution of a three-body problem in one
dimension, J. Math. Phys. 10: 2191–2196, 1969

[13] B. Sutherland. Exact results for a quantum
many-body problem in one dimension I, Phys. Rev. A4:
2019-2021, 1971

[14] A.V. Turbiner. From quantum AN (Calogero) to H4
(rational) model, SIGMA 7: 071, 2011; 20 pp From
quantum AN (Sutherland) to E8 trigonometric model:
space-of-orbits view, SIGMA 9: 003, 2013; 25 pp

469

http://arxiv.org/abs/hep-th/9506105
http://arxiv.org/abs/funct-an/9301001
http://arxiv.org/abs/0904.0738

	Acta Polytechnica 53(5):462–469, 2013
	1 Introduction
	2 The algebra gl_n in mixed representation
	3 Example: the algebra gl_3 in mixed representation
	3.1 Reps in 1x1 matrices
	3.2 Reps in 2x2 matrices
	3.3 Reps in 3x3 matrices

	4 Algebra g^(m) in mixed representation
	5 Extension of the 3-body Calogero Model
	6 Extension of the 3-body Sutherland Model
	7 Conclusions
	Acknowledgements
	References