Beruflich Dokumente
Kultur Dokumente
0. Introduction
The KP hierarchy can be completely solved by several methods. The most clas-
sical methods are based on Grassmann manifolds [SS], [SW], fermions and vertex
operators [DJKM] and factorisation of microdifferential operators [Mu]. Unfor-
tunately, those methods are not very suited for a “quasi-classical” (~-dependent,
where ~ is the Planck constant) formulation [TT2] of the KP hierarchy.
The ~-dependent formulation of the KP hierarchy was introduced to study the
dispersionless KP hierarchy [KG], [Kr], [TT1] as a classical limit (i.e., the lowest
order of the ~-expansion) of the KP hierarchy. This point of view turned out to be
very useful for understanding various features of the dispersionless KP hierarchy
such as Lax equations, Hirota equations, infinite dimensional symmetries, etc., in
the light of the KP hierarchy. In this paper, we return to the ~-dependent KP
hierarchy itself, and consider all orders of the ~-expansion.
We first address the issue of solving a Riemann-Hilbert problem for the pair
(L, M ) of Lax and Orlov-Schulman operators [OS]. This is a kind of “quantisation”
of a Riemann-Hilbert problem that solves the dispersionless KP hierarchy [TT1].
Though the Riemann-Hilbert problem for the full KP hierarchy was formulated
in our previous work [TT2], we did not consider the existence of its solution in a
general setting. In this paper, we settle this issue by an ~-expansion of the dressing
operator W , which is assumed to have the exponential form W = exp(X/~) with an
operator X of negative order. Roughly speaking, the coefficients Xn , n = 0, 1, 2, . . .,
of the ~-expansion of X are shown to be determined recursively from the lowest
order term X0 (in other words, from a solution of the dispersionless KP hierarchy).
We next convert this result to the language of the wave function Ψ. This, too,
is to answer a problem overlooked in our previous paper [TT2]. Namely, given
the dressing operator in the exponential form W = exp(X/~), we show that the
associated wave function has the WKB form Ψ = exp(S/~) with a phase function
S expanded into nonnegative powers of ~. This is genuinely a problem of calculus
of microdifferential operators rather than that of the KP hierarchy. A simplest
example such as X = x(~∂)−1 demonstrates that this problem is by no means
trivial. Borrowing an idea from Aoki’s “exponential calculus” of microdifferential
operators [A], we show that dressing operators of the form W = exp(X/~) and wave
functions of the form Ψ = exp(S/~) are determined from each other by a set of
recursion relations for the coefficients of their ~-expansion. More precisely, we need
many auxiliary quantities other than X and S, for which we can derive a large set
of recursion relations. Thus our construction is essentially recursive. Consequently,
the wave function of the solution of the aforementioned Riemann-Hilbert problem,
too, are recursively determined by the ~-expansion.
Having the ~-expansion of the wave function, we can readily derive an ~-expansion
of the tau function as stated in our previous work [TT2]. This ~-expansion is a
generalisation of the “genus expansion” of partition functions in string theories and
random matrices [D], [Kr], [Mo], [dFGZ].
This paper is organised as follows. Section 1 is a review of the ~-dependent formu-
lation of the KP hierarchy. Relevant Riemann-Hilbert problems are also reviewed
here. Section 2 presents the recursive solution of the Riemann-Hilbert problem. A
technical clue is the Campbell-Hausdorff formula, details of which are collected in
Appendix A. The construction of solution is illustrated for the case of the Kont-
sevich model [AvM] in Appendix B. Section 3 deals with the ~-expansion of the
wave function. Aoki’s exponential calculus is also briefly reviewed here. Section 4
mentions the ~-expansion of the tau function. Section 5 is devoted to concluding
remarks.
and “( )≥0 ” stands for the projection onto a differential operator dropping negative
powers of ∂. The coefficients un (~, x, t) of L are assumed to be formally regular
with respect to ~. This means that they have an asymptotic expansion of the form
P∞ (m)
un (~, x, t) = m=0 ~m un (x, t) as ~ → 0.
We need two kinds of “order” of microdifferential operators: one is the ordinary
order,
X
def
(1.3) ord an,m (x, t)~n ∂ m = max{m | an,m (x, t) 6= 0},
and the other is the ~-order defined by
X
def
(1.4) ord~ an,m (x, t)~n ∂ m = max{m − n | an,m (x, t) 6= 0}.
and α(~) is a constant with respect to x and t. (In [TT2] we did not introduce α,
which will be necessary in Section 2.)
The Orlov-Schulman operator M is defined by
∞
!
X
−1 n−1
+ x W −1
(1.11) M = Ad W exp ~ ζ(t, ~∂) x = W ntn (~∂)
n=1
P∞ n
where ζ(t, ~∂) = n=1 tn (~∂) . It is easy to see that M has a form
∞
X ∞
X
(1.12) M= ntn Ln−1 + x + α(~)L−1 + vn (~, x, t)L−n−1 ,
n=1 n=1
and satisfies
4 KANEHISA TAKASAKI AND TAKASHI TAKEBE
• ord~ (M ) = 0;
• the canonical commutation relation: [L, M ] = ~;
• the same Lax equations as L:
∂M
(1.13) ~ = [Bn , M ], n = 1, 2, . . . .
∂tn
Remark 1.2. If an operator M of the form (1.12) satisfies the Lax equations (1.13)
and the canonical commutation relation [L, M ] = ~ with the Lax operator L of the
KP hierarchy, then α(~) in the expansion (1.12) does not depend on any tn nor on
x. In fact, expanding the canonical commutation relation, we have
∂α
~ + ~ (~∂)−1 + (lower order terms) = ~,
∂x
which implies ∂α
∂x = 0. Similarly, ∂α
from (1.13) follows ∂tn
= 0 with the help of (1.1)
n n−1
and [L , M ] = n~L .
The following proposition (Proposition 1.7.11 of [TT2]) is a “dispersionful” coun-
terpart of the theorem for the dispersionless KP hierarchy found earlier (Proposition
7 of [TT1]; cf. Proposition 1.4 below).
Proposition 1.3. (i) Suppose that operators f (~, x, ~∂), g(~, x, ~∂), L and M
satisfy the following conditions:
• ord~ f = ord~ g = 0, [f, g] = ~;
• L is of the form (1.2) and M is of the form (1.12), [L, M ] = ~;
• f (~, M, L) and g(~, M, L) are differential operators:
(1.14) (f (~, M, L))<0 = (g(~, M, L))<0 = 0.
Then L is a solution of the KP hierarchy (1.1) and M is the corresponding Orlov-
Schulman operator.
(ii) Conversely, for any solution (L, M ) of the ~-dependent KP hierarchy there
exists a pair (f, g) satisfying the conditions in (i).
The leading term of this system with respect to the ~-order gives the dispersion-
less KP hierarchy. Namely,
∞
X
(1.15) L := σ ~ (L) = ξ + u0,n+1 ξ −n , (u0,n+1 := σ ~ (un+1 ))
n=1
satisfies the dispersionless Lax type equations
∂L
(1.16) = {Bn , L}, Bn = (Ln )≥0 , n = 1, 2, . . . ,
∂tn
where ( )≥0 is the truncation of Laurent series to its polynomial part and {, } is
the Poisson bracket defined by
∂a ∂b ∂a ∂b
(1.17) {a(x, ξ), b(x, ξ)} = − .
∂ξ ∂x ∂x ∂ξ
The dressing operation (1.7) for L becomes the following dressing operation for
L:
L = exp ad{,} X0 exp ad{,} α0 log ξ ξ = exp ad{,} X0 ξ,
(1.18)
X0 := σ ~ (X), α0 := σ ~ (α)
where ad{,} (f )(g) := {f, g}.
~-EXPANSION OF KP 5
which is equal to
(1.20) M = exp ad{,} X0 exp ad{,} α0 log ξ exp ad{,} ζ(t, ξ) x,
P∞ n
where ζ(t, ξ) = n=1 tn ξ . The series M satisfies the canonical commutation
relation with L, {L, M} = 1 and the Lax type equations:
∂M
(1.21) = {Bn , M}, n = 1, 2, . . . .
∂tn
Proposition 1.4. (i) Suppose that functions f0 (x, ξ), g0 (x, ξ), L and M satisfy
the following conditions:
• {f0 , g0 } = 1;
• L is of the form (1.15) and M is of the form (1.19)
• f0 (M, L) and g0 (M, L) do not contain negative powers of ξ
(f, g) and (L, M ) are quantisation of the canonical transformations (f0 , g0 ) and
(L, M) respectively. (See, for example, [S] for quantised canonical transformations.)
are both differential operators (cf. Proposition 1.3). Let us expand X and α with
respect to the ~-order as follows:
X∞ ∞
X
(2.2) X(~, x, t, ~∂) = ~n Xn (x, t, ~∂), Xn (x, t, ~∂) = χn,k (x, t)(~∂)−k ,
n=0 k=1
X∞
(2.3) α(~) = ~n αn ,
n=0
σ ~ (g)(L, M) = exp ad{,} X0 exp ad{,} α0 log ξ exp ad{,} ζ(t, ξ) σ ~ (g)(x, ξ)
• (Step 1) Set
(2.9) P (i−1) := Ad exp ~−1 X (i−1) exp ~−1 α(i−1) log(~∂) ft ,
(2.10) Q(i−1) := Ad exp ~−1 X (i−1) exp ~−1 α(i−1) log(~∂) gt .
~-EXPANSION OF KP 7
The integral constant of the indefinite integral is fixed so that the right
hand side agrees with the left hand side.
• (Step 3) Define a series Xi (x, t, ξ) = ∞ −k
P
k=1 χi,k (x, t)ξ by
∞
1
Xi = X˜i′ − {σ ~ (X0 ), X˜i′ } +
X
K2p (ad{,} (σ ~ (X0 )))2p X̃i′ ,
(2.14) 2 p=1
where ord~ (αi log(~∂) + X̃i (x, ~∂)) = 0, ord~ (X>i ) ≦ 0 and the principal symbol
of αi log(~∂) + X̃i (x, ~∂) is defined by
Note that the only log term in (2.18) is αi log ξ and the rest is sum of negative
powers of ξ. The principal symbol of Xi is recovered from X̃i by the formula
(2.19)
∞
1 ~ X
~
σ (Xi ) = σ ~
(X̃i′ ) ~ ′
− {σ (X0 ), σ (X̃i )} + K2p (ad{,} (σ ~ (X0 )))2p σ ~ (X̃i′ ),
2 p=1
Here coefficients K2p are defined by (2.15). This inversion relation is the origin of
(2.14). (Note that the principal symbol determines the operator Xi , since it is a
homogeneous term in the expansion (2.2).) We prove formulae (2.17) and (2.19) in
Appendix A.
The factorisation (2.17) implies
P = Ad exp ~i−1 (αi log(~∂) + X̃i ) + ~i X>i P (i−1)
=P (i−1) + ~i−1 [(αi log(~∂) + X̃i ) + ~X>i , P (i−1) ] + (terms of ~-order < −i).
Similar equations for Q are obtained in the same way. The first equations (2.21)
show that the terms of ~-order greater than −i in (2.5) are already fixed by
X0 , . . . , Xi−1 and α0 , . . . , αi−1 , which justifies the inductive procedure. That is
to say, we are assuming that X0 , . . . , Xi−1 and α0 , . . . , αi−1 have been already de-
(i−1) (i−1)
termined so that Pj = Pj and Qj = Qj for j = 0, . . . , i − 1 are differential
operators.
The operator Xi and constant αi should be chosen so that the right hand side
of (2.22) and the corresponding expression for Q are differential operators. Taking
~-EXPANSION OF KP 9
(i−1) (i−1)
equations P0 = P0 and Q0 = Q0 into account, we define
(i) (i−1)
P̃i := Pi + ~−1 [αi log(~∂) + X̃i , P0 ],
(2.23) (i) (i−1)
Q̃i := Qi + ~−1 [αi log(~∂) + X̃i , Q0 ].
Then the condition for Xi and αi is written in the following form of equations for
symbols:
(i) (i)
(2.24) (σ0~ (P̃i ))≤−1 = 0, (σ0~ (Q̃i ))≤−1 = 0.
(The parts of ~-order less than −1 should be determined in the next step of the
(i) (i−1)
induction.) To simplify notations, we denote the symbols σ0~ (P̃i ), σ0~ (Pi ) and
(i) (i−1)
so on by the corresponding calligraphic letters as Pi , Pi etc. By this notation
we can rewrite the equations (2.24) in the following form:
(i) (i) (i−1)
(P̃i )≤−1 = 0, P̃i := Pi + {αi log ξ + X˜i , P0 } = 0,
(2.25) (i) (i) (i−1)
(Q̃i )≤−1 = 0, Q̃i := Qi + {αi log ξ + X̃i , Q0 } = 0.
(i) (i)
The above definitions of P̃i and Q̃i are written in the matrix form:
∂P0 ∂P0 ∂
−
∂ξ (αi log ξ + X̃i )
(i) (i−1)
∂x ∂ξ Pi − Pi
(2.26) = .
∂Q0 ∂Q0 ∂ (i) (i−1)
Q − Q
− (αi log ξ + X˜i ) i i
∂x ∂ξ ∂x
Recall that operators P (i−1) and Q(i−1) are defined by acting adjoint operation to
the canonically commuting pair (f, g) in (2.9), (2.10) and (2.7). Hence they also
satisfy the canonical commutation relation: [P (i−1) , Q(i−1) ] = ~. The principal
symbol of this relation gives
(i−1) (i−1)
{P0 , Q0 } = {P0 , Q0 } = 1,
which means that the determinant of the matrix in the left hand side of (2.26) is
equal to 1. Hence its inverse matrix is easily computed and we have
∂ ∂Q0 ∂P0
(α log ξ + X̃ ) −
i i (i) (i−1)
∂ξ ∂ξ P
∂ξ i
− P i
(2.27) = .
∂ ∂Q ∂P (i) (i−1)
(αi log ξ + X˜i ) −
0 0 Qi − Qi
∂x ∂x ∂x
We are assuming that P0 and Q0 do not contain negative powers of ξ and we are
(i) (i)
searching for αi log ξ + X˜i such that Pi and Qi are series of ξ without negative
powers. Since αi is constant with respect to x, the left hand side of (2.27) contain
only negative powers of ξ. Thus taking the negative power parts of the both hand
sides in (2.27), we have
∂Q0 (i−1) ∂P0 (i−1)
∂
˜ P − Q
∂ξ (αi log ξ + Xi ) ∂ξ i ∂ξ i ≤−1
(2.28) = .
∂ ∂Q ∂P
˜ 0 (i−1) 0 (i−1)
(αi log ξ + Xi ) Pi − Qi
∂x ∂x ∂x ≤−1
The system (2.28) is solvable thanks to Lemma 2.2 below. Hence, integrating
the first element of the right hand side with respect to ξ, we obtain αi log ξ + X˜i .
This is Step 2, (2.13). In the end, the principal symbol of Xi is determined by
(2.19) or (2.14) in Step 3 and thus Xi is defined as in Step 4. This completes the
construction of Xi and αi and the proof of the theorem.
Lemma 2.2. The system (2.28) is compatible.
Proof. We check:
∂ ∂Q0 (i−1) ∂P0 (i−1)
Pi − Qi
∂x ∂ξ ∂ξ ≤−1
(2.29)
∂ ∂Q0 (i−1) ∂P0 (i−1)
= P − Q .
∂ξ ∂x i ∂x i ≤−1
Since the time variables tn do not play any role in this section, we set them to
zero. As the factor (~∂)α/~ in (1.8) becomes a constant factor z α when it is applied
to exz/~ , we also omitPit here.
Let A(~, x, ~∂) = n an (~, x)(~∂)n be a microdifferential operator. The total
symbol of A is a power series of ξ defined by
X
(3.4) σtot (A)(~, x, ξ) := an (~, x)ξ n .
n
Actually, this is the factor which appears when the operator A is applied to exz/~ :
(3.5) Aexz/~ = σtot (A)(~, x, z)exz/~ .
Using this terminology, what we show in this section is that a operator of the form
eX/~ has a total symbol of the form eS/~ and that an operator with total symbol
eS/~ has a form eX/~ . Exactly speaking, the main results in this section are the
following two propositions.
Proposition 3.1. Let X = X(~, x, ~∂) be a microdifferential operator such that
ord X = −1 and ord~ X = 0. Then the total symbol of eX/~ has such a form as
(3.6) σtot (exp(~−1 X(~, x, ~∂))) = eS(~,x,ξ)/~ ,
where S(~, x, ξ) is a power series of ξ −1 without non-negative powers of ξ and has
an ~-expansion
X∞
(3.7) S(~, x, ξ) = ~n Sn (x, ξ).
n=0
Moreover, P
the coefficient Sn is determined by X0 , . . . , Xn in the ~-expansion
∞
(2.2) of X = n=0 ~n Xn .
Explicitly, Sn is determined as follows:
• (Step 0) Assume that X0 , . . . , Xn are given. Let Xi (x, ξ) be the total symbol
σtot (Xi (x, ~∂)).
(l)
• (Step 1) Define Yk,m (x, y, ξ, η) and S (l) (x, ξ) by the following recursion re-
lations:
(l)
(3.8) Yk,−1 = 0
(0)
(3.9) Sm = 0,
(l)
(3.10) Y0,m (x, y, ξ, η) = δl,0 Xm (x, ξ)
for l ≧ 0, m = 0, . . . , n,
(l)
(3.11) Yk+1,m (x, y, ξ, η)
1
∂ξ ∂y Y (l) (x, y, ξ, η) +
X (l′ ) (l−l′ )
= k,m−1 ∂ξ Yk,m′ (x, y, ξ, η)∂y Sm−m′ (y, η)
k+1
0≤l′ ≤l−1
0≤m′ ≤m
for k ≧ 0, and
l+m
(l+1) 1 X (l)
(3.12) Sm (x, ξ) = Yk,m (x, x, ξ, ξ).
l+1
k=0
12 KANEHISA TAKASAKI AND TAKASHI TAKEBE
(l)
(We shall prove that Yk,m = 0 if k > l + m.) Schematically this procedure
goes as follows:
(l) (l) (l)
Y0,0 = δl,0 X0 Y0,1 = δl,0 X1 Y0,2 = δl,0 X2
+ ց + ց +
(l) (l) (l) (l)
Yk,−1 = 0 → Yk,0 → Yk,1 → Yk,2 ···
↓ ր ↓ ր ↓
(l+1) (l+1) (l+1)
S0 S1 S2
P∞ (l)
• (Step 2) Sn (x, ξ) = l=1 Sn (x, ξ). (The sum makes sense as a power series
of ξ.)
P∞ n −1
Proposition 3.2. Let S = n=0 ~ Sn be a power series of ξ without non-
negative powers of ξ. Then there exists a microdifferential operator X(~, x, ~∂)
such that ord X ≦ −1, ord~ X ≦ 0 and
for m = 0, . . . , n, k ≧ 0, l ≧ 0 and
(l)
(3.16) Y0,m,j = 0
(l)
for m = 0, . . . , n, l > 0, j ≧ 1. For other (l, k, m, j), (l, k) 6= (0, 0), Yk,m,j
are determined by the recursion relation:
(l) 1 (l)
(3.17) Yk+1,m,j (x, y, ξ, η) = ∂ξ ∂y Yk,m−1,j−1 (x, y, ξ, η)+
k+1
!
X 1 (l′ ) (l−l′ −1)
+ ∂ξ Yk,m′ ,j ′ (x, y, ξ, η)∂y Yk′′ ,m−m′ ,j−j ′ −1 (x, x, ξ, ξ) .
l − l′
0≤l′ ≤l−1
0≤j ′ ≤j−1,0≤m′ ≤m
0≤k′′ ≤j−j ′ −l+l′
(0)
The remaining Y0,m,j is determined by:
(0)
X 1 (l)
(3.18) Y0,m,j (x, y, ξ, η) = Sm,j (x, ξ) − Y (x, x, ξ, ξ).
l + 1 k,m,j
(l,k)6=(0,0)
l,k≥0,l+k≤j
~-EXPANSION OF KP 13
(l)
(We shall show that Yk,m,j = 0 for l + k > j.) Schematically this procedure
goes as follows:
(l)
Yk,m,1 = δl,0 δk,0 Sm,1
↓
(l′ ) ′ (l) (0)
Yk′ ,m′ ,1 (m < m) → Yk,m,2 (k, l 6= 0) → Y0,m,2 ← Sm,2
↓ ւ
(l′ ) (l′ ) ′ (l) (0)
Yk′ ,m′ ,1 , Yk′ ,m′ ,2 (m < m) → Yk,m,3 (k, l 6= 0) → Y0,m,3 ← Sm,3
..
.
P∞ (0)
• (Step 2) Xn (x, ξ) = j=1 Y0,n,j (x, x, ξ, ξ). (The infinite sum is the homo-
geneous expansion in terms of powers of ξ.)
Combining these propositions with the results in Section 2, we can, in principle,
make a recursion formula for Sn (n = 0, 1, 2, . . . ) of the wave function of the solution
of the KP hierarchy corresponding to the quantised canonical transformation (f, g)
as follows: let S0 , . . . , Si−1 be given.
(1) By Proposition 3.2 we have X0 , . . . , Xi−1 .
(2) We have a recursion formula for Xi by Theorem 2.1.
(3) Proposition 3.1 gives a formula for Si .
If we take the factor (~∂)α/~ into account, this process becomes a little bit compli-
cated, but essentially the same.
The rest of this section is devoted to the proof of Proposition 3.1 and Proposi-
tion 3.2.
These statements might seem obvious, but since the multiplication in the defi-
nition of
∞
X X(~, x, ~∂)n
(3.19) eX/~ =
n=0
~n n!
(See, for example, [S], [A] or [KR].) The order of an operator corresponding to
symbol a(~, x, ξ) is denoted by ordξ a(~, x, ξ), which is the same as the order of
a(~, x, ξ) as a power series of ξ. The ~-order is the same as that of the operator:
ord~ x = ord~ ξ = 0, ord~ ~ = −1.
The main idea of proof of propositions is due to Aoki [A], where exponential cal-
culus of pseudodifferential operators is considered. He considered analytic symbols
of exponential type, while our symbols are formal ones. Therefore we have only to
confirm that those symbols make sense as formal series.
First, we prove the following lemma.
Lemma 3.3. Let a(~, x, ξ) and b(~, x, ξ) be two symbols such that ordξ a(~, x, ξ) =
M , ord~ a(~, x, ξ) = 0, ordξ b(~, x, ξ) = N , ord~ b(~, x, ξ) = 0, ordξ p(~, x, ξ) =
ordξ q(~, x, ξ) = 0, ord~ p(~, x, ξ) = ord~ q(~, x, ξ) = 0.
Then there exists a symbol c (ordξ c = N + M , ord~ c = 0), r (ordξ r = 0,
ord~ r = 0) such that
In the proof of Proposition 3.1 and Proposition 3.2, we use the construction of c
and r in the proof of Lemma 3.3.
π(t) = ψ(t)ew(t)/~ ,
∞
X
ψ(t) = ψ(t; ~, x, y, ξ, η) = ψn tn ,
(3.25) n=0
X∞
w(t) = w(t; ~, x, y, ξ, η) = wk tk .
k=0
~-EXPANSION OF KP 15
Later we set t = 1 and prove that ψ(1) and w(1) is meaningful as a formal power
series of ξ and η. The differential equation (3.24) is rewritten as
∂ψ ∂w
+ ~−1 ψ
(3.26) ∂t ∂t
=~∂ξ ∂y ψ + ∂ξ ψ∂y w + ∂y ψ∂ξ w + ψ ∂ξ ∂y w + ~−1 ∂ξ w∂y w .
Taking the logarithm (as a function, not as an operator) of this, we can define
S(s) = S(s; ~, x, ξ) by
−1
S(s;~,x,ξ)
(3.35) E(s; ~, x, ξ) = e~ .
What we are to prove is that S(s), constructed as a series, makes sense at s = 1
and formally regular with respect to ~.
Differentiating (3.35), we have
∂S S(s;~,x,ξ)/~
(3.36) e
X(~, x, ξ) ◦ E(s; ~, x, ξ) = .
∂s
By Lemma 3.3 (a 7→ X, b 7→ 1, p 7→ 0, q 7→ S) and the technique in its proof, we
can rewrite the left hand side as follows:
(3.37) X(~, x, ξ) ◦ E(s; ~, x, ξ) = Y (s; ~, x, x, ξ, ξ)eS(s;~,x,ξ)/~ ,
P∞
where Y (s; ~, x, y, ξ, η) = k=0 Yk and Yk (s; ~, x, y, ξ, η) are defined by
Yk+1 (s; ~, x, y, ξ, η)
1
(3.38) = (~∂ξ ∂y Yk (s; ~, x, y, ξ, η) + ∂ξ Yk (s; ~, x, y, ξ, η)∂y S(s; ~, y, η)),
k+1
Y0 (s; ~, x, y, ξ, η) = X(~, x, ξ).
Yk (s) corresponds to ψk in the proof of Lemma 3.3, while wk there corresponds to
δk,0 S(s). On the other hand, substituting (3.37) into the left hand side of (3.36),
we have
∂S
(3.39) (s; ~, x, ξ) = Y (s; ~, x, x, ξ, ξ).
∂s
We rewrite the system (3.38) and (3.39) in terms of expansion of S(s; ~, x, ξ) and
Yk (s; ~, x, y, ξ, η) in powers of s and ~:
∞
X ∞ X
X ∞
S(s; ~, x, ξ) = S (l) (~, x, ξ)sl = Sn(l) (x, ξ)~n sl ,
l=0 l=0 n=0
(3.40) ∞ ∞ X∞
(l) (l)
X X
l
Yk (s; ~, x, y, ξ, η) = Yk (~, x, y, ξ, η)s = Yk,n (x, y, ξ, η)~n sl ,
l=0 l=0 n=0
n l
The coefficient of ~ s in (3.38) is
(l)
(3.41) Yk+1,n (x, y, ξ, η)
1
∂ξ ∂y Y (l) (x, y, ξ, η) +
X (l′ ) (l′′ )
= k,n−1 ∂ξ Yk,n′ (x, y, ξ, η)∂y Sn′′ (y, η)
,
k+1
l′ +l′′ =l
n′ +n′′ =n
(l)
(Yk,−1 = 0) and
(l)
(3.42) Y0,n (x, y, ξ, η) = δl,0 Xn (x, ξ),
~-EXPANSION OF KP 17
(5) Fix l0 ≧ 1 and assume that for all l = 0, . . . , l0 −1 and for all k = 0, 1, 2, . . . ,
(l)
we have determined Yk,n and that for all l = 0, . . . , l0 we have determined
(l)
Sn . (The steps (3) and (4) are for l0 = 1.)
(0)
Since Sn′′ = 0 by (3.44), the index l′ in the right hand side of the
recursion relation (3.41) (with l = l0 ) runs essentially from 0 to l0 − 1.
(l0 )
Hence this relation determines Yk+1,n from known quantities for all k ≧ 0.
Because of the initial condition Y0 (s; x, y, ξ, η) = X(x, ξ) (cf. (3.38)) Y0
(l )
does not depend on s, which means that its Taylor coefficients Y0,n0 vanish
for all l0 ≧ 1:
(l )
(3.48) Y0,n0 = 0.
(l )
Thus we have determined all Yk,n0 (k = 0, 1, 2, . . . ).
(l +1)
(6) We shall prove below that Yk,n0 = 0 if k > l0 + n + 1. Hence the sum in
(l +1)
(3.43) is finite and Sn 0 can be determined. The induction proceeds by
incrementing l0 by one.
18 KANEHISA TAKASAKI AND TAKASHI TAKEBE
(l)
Let us prove that Yk,n ’s determined above satisfy
(l)
(3.49) Yk,n = 0, if k > l + n,
(l)
(3.50) ordξ Yk,n ≦ −k − l − 1.
(We define that ordξ 0 = −∞.) In particular, the sum in (3.43) is well-defined and
(3.51) ordξ Sn(l+1) ≦ −l − 1.
If n = −1, both (3.49) and (3.50) are obvious. Fix n0 ≧ 0 and assume that we
have proved (3.49) and (3.50) for n < n0 and all (l, k).
When l = 0, (3.49) and (3.50) are true for all k because of (3.46) and (3.45),
respectively. Fix l0 ≧ 0 and assume that we have prove (3.49) and (3.50) for l ≦ l0
and all k. As a result (3.51) is true for l ≦ l0 .
The recursion relation (3.41) implies (3.49) and (3.50) for l = l0 + 1 and all
k > 0. In fact, if k + 1 > (l0 + 1) + n, then k > (l0 + 1) + (n − 1) and k > l′ + n′
(l0 +1) (l′ )
which guarantees that Yk,n−1 = 0 and Yk,n′ = 0 in the recursion relation (3.41)
for l = l0 + 1 by the induction hypothesis. The estimate (3.50) is easy to check for
l0 + 1. (Recall once again that ∂ξ lowers the order by one.)
For k = 0, (3.49) is void and (3.50) is true because of (3.42) and ord Xn ≦ −1.
The step l = l0 + 1 being proved, the induction proceeds with respect to l and
consequently with respect to n.
Here terms with index j are homogeneous terms of degree −j with respect to ξ and
η.
At the end of this proof we shall determine Xn by (3.42),
(0)
(3.52) Xn (x, ξ) = Y0,n (x, y, ξ, η).
~-EXPANSION OF KP 19
(0)
(In particular, Y0,n (x, y, ξ, η) should not depend on y and η.) For this purpose,
(0)
Y0,n should be determined by
(0)
X 1 (l)
(3.53) Y0,n (x, y, ξ, η) = Sn (x, ξ) − Y (x, x, ξ, ξ)
l + 1 k,n
(l,k)6=(0,0)
l,k≥0
P∞ (l)
because of (3.43) and Sn (x, ξ) = l=1 Sn (x, ξ).
(l)
Since ordξ Yk,n should be less than −l − k (cf. (3.50)), we expect
(l)
(3.54) Yk,n,1 = 0
for (l, k) 6= (0, 0). Hence picking up homogeneous terms of degree −1 with respect
to ξ, the following equation should hold:
(0)
(3.55) Y0,n,1 = Sn,1
(l)
All Yk,n,1 are determined by the above two conditions, (3.54) and (3.55). Note also
that
(l)
(3.56) Y0,n,j = 0 for l 6= 0
because Y0 should not depend on s because of (3.42).
(l)
Having determined initial conditions in this way, we shall determine Yk,n,j in-
ductively. To this end we rewrite the recursion relation (3.41) by (3.43) and pick
up homogeneous terms of degree j:
(l) 1 (l)
(3.57) Yk+1,n,j (x, y, ξ, η) = ∂ξ ∂y Yk,n−1,j−1 (x, y, ξ, η)+
k+1
!
X 1 (l′ ) (l′′ −1)
+ ∂ξ Yk,n′ ,j ′ (x, y, ξ, η)∂y Yk′′ ,n′′ ,j ′′ (x, x, ξ, ξ)
l′′
l′ +l′′ =l,l′′ ≧1
j ′ +j ′′ =j−1,n′ +n′′ =n
0≦k′′
(l)
(As before, terms like Yk,−1,j−1 appearing the above equation for n = 0 can be
ignored.)
(l) (l)
Fix n0 ≧ 0 and assume that Yk,0,j , . . . , Yk,n0 −1,j are determined for all (l, k, j).
(l)
(1) First we determine Yk,n0 ,1 for all (l, k) by (3.55) and (3.54).
(l)
(2) Fix j0 ≧ 2 and assume that Yk,n0 ,j are determined for j = 1, . . . , j0 − 1 and
all (l, k). (The above step is for j0 = 2.)
Since all the quantities in the right hand side of the recursion relation
(3.57) with j = j0 are known by the induction hypothesis, we can determine
(l)
Yk,n0 ,j0 for l = 0, 1, 2, . . . and k = 1, 2, . . . .
(l)
(3) Together with (3.56), Y0,n0 ,j0 = 0 for l = 1, 2, . . . , we have determined all
(l)
Yk,n0 ,j0 except for the case (l, k) = (0, 0).
(l)
(4) It follows from (3.57), (3.55) and (3.56) by induction that for all Yk,n0 ,j
determined in (1), (2) and (3),
(l)
(3.58) Yk,n,j = 0 for l + k + 1 > j.
20 KANEHISA TAKASAKI AND TAKASHI TAKEBE
(l)
This corresponds to ordξ Yk,n0 ≦ −l − k − 1 (3.50) in the proof of Proposi-
tion 3.1.
(0)
(5) We determine Y0,n0 ,j0 by
(0)
X 1 (l)
(3.59) Y0,n0 ,j0 = Sn0 ,j0 − Y (x, x, ξ, ξ),
l + 1 k,n0 ,j0
(l,k)6=(0,0)
l,k≥0
which is the homogeneous part of degree −j0 in (3.53). The sum in the
right hand side is finite thanks to (3.58).
(6) The induction with respect to j proceeds by incrementing j0 .
(l) (0)
Thus all Yk,n0 ,j are determined and Xn0 is determined by Xn0 (x, ξ) = ∞
P
j=1 Y0,n0 ,j
(cf. (3.52)), which completes the proof of Proposition 3.2.
of the tau function from the ~ expansion (3.7) of the S-function. Note that we have
suppressed the variable x, which is understood to be absorbed in t1 .
Let us recall the fundamental relation [DJKM]
τ (t − ~[z −1 ]) ~−1 ζ(t,z)
(4.2) Ψ(t; z) = e ,
τ (t)
P∞
where [z −1 ] = (1/z, 1/2z 2, 1/3z 3, . . . ) and ζ(t, z) = n=1 tn z n . This implies that
(4.3) ~−1 Ŝ(t; z) = e−~D(z) − 1 log τ (t)
P∞ −j
where Ŝ(t; z) = S(t; z) − ζ(t, z) and D(z) = j=1 z j ∂t∂j . Differentiating this with
respect to z, we have
∂
~−1 Ŝ(t; z) = − ~D′ (z)e−~D(z) log τ (t)
(4.4) ∂z
= − ~D′ (z)(log ŵ(t; z) + log τ (t)),
∂
P∞
where D′ (z) := ∂z D(z) = − j=1 z −j−1 ∂t∂j . Hence
∂
(4.5) − ~D′ (z) log τ (t) = ~−1 + ~D′ (z) Ŝ(t; z)
∂z
Multiplying z n to this equation and taking the residue, we obtain a system of
differential equations
∂ ∂
(4.6) ~ log τ (t) = ~−1 Res z n + ~D′ (z) Ŝ(t; z) dz, n = 1, 2, . . .
∂tn ∂z
which is known to be integrable [DJKM]. We can thus define the tau function τ (t),
up to multiplication τ (t) → cτ (t) by a nonzero constant c, as a solution of (4.5).
By substituting the ~-expansions
∞
X ∞
X
(4.7) log τ (t) = ~n−2 Fn (t), Ŝ(t; z) = ~n Sn (t; z),
n=0 n=0
~-EXPANSION OF KP 21
5. Concluding remarks
We have presented a recursive construction of solutions of the ~-dependent KP
hierarchy. The input of this construction is the pair (f, g) of quantised canonical
transformation. The main outputs are the dressing operator W in the exponential
form (1.8), the wave function Ψ in the WKB form (3.2) and the tau function with
the quasi-classical expansion (4.1). Thus the ~-dependent KP hierarchy introduced
in our previous work [TT2] is no longer a heuristic framework for deriving the
dispersionless KP hierarchy, but has its own raison d’être.
A serious problem of our construction is that the recursion relations are extremely
complicated. In Appendix B, calculations are illustrated for the Kontsevich model
[Ko], [D], [AvM]. As this example shows, this is by no means a practical way to
construct a solution. We believe that one cannot avoid this difficulty as far as
general solutions are considered.
22 KANEHISA TAKASAKI AND TAKASHI TAKEBE
Special solutions stemming from string theory and random matrices [Mo], [dFGZ]
(e.g. the Kontsevich model) can admit a more efficient approach such as the method
of Eynard and Orantin [EO]. Those methods are based on a quite different principle.
In the method of Eynard and Orantin, it is the so called “loop equation” for correla-
tion functions of random matrices. The loop equations amount to “constraints” on
the tau function. Eynard and Orantin’s “topological recursion relations” determine
a solution of those constraints rather than of an underlying integrable hierarchy; it
is somewhat surprising that a solution of those constraints gives a tau function.
Lastly, let us mention that the results of this paper can be extended to the Toda
hierarchy. That case will be treated in a forthcoming paper.
The coefficients K2p are defined by (2.15). See, for example, [B].
First we prove
(A.3) exp(~−1 X(x, t, ~∂))
= exp ~i−1 X̃i′ + (terms of ~-order < −i + 1) exp ~−1 X (i−1) ,
which implies that for any operator Z ord~ [Cn , Z] is less than ord~ Z by more than
one. Hence the term of the highest ~-order in the recursive definition (A.2) is the
first term. More precisely, it is decomposed as
1 1 1 1
· [(A + B) − (−A), CN ] = [A, CN ] + [B, CN ],
N +1 2 N +1 2(N + 1)
and the first term in the right hand side has the highest ~-order. By the induction
hypothesis and ord~ A ≦ 1, we have
(ad A)N −1
1 1
[A, CN ] = A, (B) + (terms of ~ order < −i + 1)
N +1 N +1 N!
(A.7)
(ad A)N
= (B) + (terms of ~-order < −i + 1).
(N + 1)!
This proves (A.6) for all n. Taking its symbol of order −i + 1, we have
(ad{,} σ ~ (A))n−1 ~
(A.8) σ ~ (cn (A + B, −A)) = σ (B),
n!
which gives the terms of (A.4). Substituting this into the Campbell-Hausdorff
formula (A.1), we have (A.3).
By factorisation (A.3), we can factorise W = exp(X/~)(~∂)α as follows (α(i−1) :=
Pi−1 j
j=0 ~ αj ):
In order to recover Xi from X̃i (or X̃i′ ), we have only to invert the definition
(A.4). In the definition (A.4) we substitute ad{,} (σ ~ (X0 )) in the equation
∞
et − 1 X tn−1
= .
t n=1
n!
gives the inverse map Xi 7→ X̃i . Here the coefficients K2p are defined in (2.15).
Hence equation (2.19):
∞
1 X
σ ~ (Xi ) = σ ~ (X̃i′ ) − {σ ~ (X0 ), σ ~ (X̃i′ )} + K2p (ad{,} (σ ~ (X0 )))2p σ ~ (X̃i′ )
2 p=1
To begin with, let us determine the leading terms of X and α, which are the
initial data for our procedure. The dispersionless limit of (L, M ) satisfies
2 1 −1
(B.3) (L )≤−1 = 0, ML − L = 0.
2 ≤−1
References
[A] Aoki, T., Calcul exponentiel des opérateurs microdifférentiels d’ordre infini. II. Ann. Inst.
Fourier (Grenoble) 36 (1986), 143–165.
[AvM] Adler, M., and van Moerbeke, P., A matrix integral solution to two-dimensional Wp -
gravity. Comm. Math. Phys. 147 (1992), 25–56.
[B] Bourbaki, N., Elements of Mathematics, Lie Groups and Lie Algebras, Chapters 1-3,
Springer, (1989).
[D] Dijkgraaf, R., Intersection Theory, Integrable Hierarchies and Topological Field Theory,
in New symmetry principles in quantum field theory (Cargèse, 1991), NATO Adv. Sci.
Inst. Ser. B Phys., 295 (1992), 95–158.
[dFGZ] Di Francesco, P., Ginsparg, P., and Zinn-Justin, J., 2D gravity and random matrices.
Phys. Rep. 254 (1995), 133 pp.
[DJKM] Date, E., Kashiwara, M., Jimbo, M., and Miwa, T.: Transformation groups for soliton
equations, in: Nonlinear Integrable Systems — Classical Theory and Quantum Theory
(World Scientific, Singapore), (1983), 39–119.
[EO] Eynard, B., and Orantin, N., Invariants of algebraic curves and topological expansion,
Commun. Number Theory Phys. 1 (2007), 347–452.
[KG] Kodama, Y., A method for solving the dispersionless KP equation and its exact solutions,
Phys. Lett. 129A (1988), 223–226; Kodama, Y., and Gibbons, J., A method for solving
the dispersionless KP hierarchy and its exact solutions, II, Phys. Lett. 135A (1989),
167–170.
[Ko] Kontsevich, M., Intersection theory on the moduli space of curves and the matrix Airy
function, Comm. Math. Phys. 147 (1992), 1–23.
[Kr] Krichever, I.M., The dispersionless Lax equations and topological minimal models, Com-
mun. Math. Phys. 143 (1991), 415–426.
[KR] Kashiwara, M. and Rouquier, R., Microlocalization of rational Cherednik algebras.
arXiv:math.RT/0705.1245
[Mo] Morozov, A., Integrability and Matrix Models Phys. Usp. 37 (1994), 1–55.
28 KANEHISA TAKASAKI AND TAKASHI TAKEBE