Professional Documents
Culture Documents
TUW-??-??
Mauricio Romo∗1
∗Kavli Institute for the Physics and the Mathematics of the Universe
The University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa 277-8583, Japan
Abstract
Here goes the abstract
1
mauricio.romo-at-ipmu.jp
Contents
1 Holomorphic limit and rational functions 2
2 An idea 4
4 Further ambiguities 7
which has the opposite sign as the one used in the original BCOV paper, therefore the speacial
geometry relation reads
mj
∂ j̄ Γjkl = Gj̄k δlj + Gj̄l δkj − Cklm C j̄ (1.2)
2
+ 480w2 z + 840w3 z − 1420w4 z + 372w5 z + 120w6 z + 34z 2 + 60wz 2 − 60w3 z 2
− 34w4 z 2 − 12z 3 − 212wz 3 − 96w2 z 3 − 5z 4 )
fzww = −16(1 + 2w − 22w2 + 34w3 + 20w4 − 106w5 + 118w6 − 58w7 + 11w8 − 2z + 62wz
+ 476w2 z − 564w3 z − 474w4 z + 438w5 z + 64w6 z − 34wz 2 − 94w2 z 2 − 94w3 z 2
− 34w4 z 2 + 2z 3 − 30wz 3 − 40w2 z 3 − z 4 )
fwww = 8(−1 − 16w + 20w2 + 112w3 − 230w4 + 16w5 + 276w6 − 240w7 + 63w8
+ 4z + 64wz − 1380w2 z − 4224w3 z + 2332w4 z + 2944w5 z + 260w6 z − 6z 2 − 80wz 2
− 564w2 z 2 − 688w3 z 2 − 198w4 z 2 + 4z 3 + 32wz 3 − 124w2 z 3 − z 4 ) (1.7)
then, recall
−1
∂t ∂t
= M −1 ∂α M
Γα top = ∂α (1.10)
∂z ∂z
we also need the topological limit of Kl , for this, recall
K = − ln(Π0 )top
(1.11)
so, we have
e i )lj := Kj δil + Ki δjl
(K (1.12)
then, we can write
ez = − 1
K
2∂z Π0 ∂w Π0 ew = − 1
K
∂ w Π0 0
(1.13)
Π0 0 ∂z Π0 Π0 ∂z Π0 2∂w Π0
εda det(ma(b,c) )
e b )dc =
(Γb − K (1.16)
det(m)
3
so, for example
!
1 det(mw
(z,z) ) det(mw
(z,w) )
Γz − K
ez = (1.17)
det(m) − det(mz(z,z) ) − det(mz(w,z) )
SCi = −(Γi − K
e i ) + si (1.18)
must be a matrix of rational functions. This is hard to see since it depends on the periods ~π
and its first and second derivatives. One can use the Picard-Fuchs equations:
p11 zw∂z ∂w ~π + p00~π + w(p01 + p02 )∂w ~π + z(p10 + p20 )∂z ~π + p02 w2 ∂w
2
~π + z 2 p20 ∂z2~π = 0(1.20)
1
det(ma(z,w) ) = ~π · (∂a~π × (∂zw
2
~π )) = − (z 2 p20 det(ma(z,z) ) + w2 p02 det(ma(w,w) ) + fa det(m))
zwp11
fz = −w(p01 + p02 ), fw = z(p10 + p20 ) (1.22)
Using this substitution for ∂z ∂w ~π and replacing the explicit expressions for the Yukawa cou-
plings and the pij ’s is enough to show that the components of (1.19) are indeed rational
functions. There is no need to replace any explicit expression for the periods. Moreover, the
terms involving the functions ∂a2~π all cancel. So one can read the rational functions, in terms
of arbitrary values of Cijk and pij just by setting to zero all the ∂a2~π terms in (1.19).
must also be a matrix of rational functions, but an analogous approach fails for this. It seems
there is some extra relation needed, or can we bypass this somehow?
2 An idea
We can explicitly solve for si for i = w, z the eqs (1.19):
e z )(Cz )−1 − (Γw − K
(Γz − K e w )(Cw )−1 − sz (Cz )−1 + sw (Cw )−1 = 0 (2.1)
by noting that
e z )(Cz )−1 + (Γw − K
−(Γz − K e w )(Cw )−1
4
−(p10 + p20 )Czzw w−1 (p10 + p20 )Czzz w−1
−1
= (det(Cz )p11 )
−(p01 + p02 )Czzw z −1 (p01 + p02 )Czzz z −1
−1 −(p + p )C −1
−1 (p10 + p20 )Cwww w 10 20 zww w
− (det(Cw )p11 ) (2.2)
(p01 + p02 )Cwww z −1 −(p01 + p02 )Czww z −1
in the current GN example one can show that these eqs can be solved for example by imposing
constraints on szwz , sw z w w z z w
wz , szz and szz and writing them completely in terms of (szw , szw , sww , sww ).
We write these constraints here, for completeness:
szwz = szzw sw w
wz = szw (2.3)
((−p11 zsw w
zw − p10 − p20 )Czzw − p11 zCzzz sww )Czww
sw
zz =
p11 zdet(Cw )
p11 wCzzw sww + Cwww Czzz (p11 zsw
2 w
zw + p10 + p20 )
+ (2.5)
p11 zdet(Cw )
all these constraints are well defined as long as det(Ca ) 6= 0, etc. Denote the si functions
satisfying these constrains by s̃i .
We can also write (2.2) in a more explicit form for our GN example, we define:
q1 := w2 − 2w − 5z + 1
q2 := 5w2 − 10w − z + 5
q3 := 23w4 − 32w3 − 58w2 z − 2w2 + 56zw + 3z 2 + 8w − 6z + 3 (2.6)
5
Next, we can write the propagators as
e z )(Cz )−1 + s̃z (Cz )−1 = −(Γw − K
Se = −(Γz − K e w )(Cw )−1 + s̃w (Cw )−1 (2.9)
The propagators S,e even though they now don’t depend on the choice of z or w, it does still
suffer from the problem that they are not symmetric, this is because:
T
1 −1
−1
−(Γa − Ka )(Ca ) + (Γa − Ka )(Ca )
e e = −Aa (2.10)
2
!
zwr1
0 8q3
Aw = (2.11)
− zwr
8q3
1
0
where
!
zwr2
0 8q1 q2
Az = zwr2 (2.13)
− 8q1 q2
0
where
Eqs. (2.15) corresponds to only one constraint (instead of 2), since we are using the s̃i ’s. This
means that after solving all the constraints coming from consistency of the S ij propagators
we can write all the 8 functions slij only in terms of 3 of them. In the following I will outline
6
an idea to fix the ambiguities of S ij , assuming we don’t have access to Aa explicitly. Consider
the symmetrized propagators:
1 e eT e z )(Cz )−1 + Az + 1 (s̃z (Cz )−1 + s̃z (Cz )−1 T )
S+S = −(Γz − K
2 2
e w )(Cw )−1 + Aw + 1 (s̃w (Cw )−1 + s̃w (Cw )−1 T )
= −(Γw − K
2
(2.16)
and the new matrices of rational functions s0i by solving
1 T
s0a (Ca )−1 = Aa + (s̃a (Ca )−1 + s̃a (Ca )−1 ) (2.17)
2
0
note that, the solutions for sa always exist. Then define S,
1 e eT e z )(Cz )−1 + s0 (Cz )−1
S := S+S = −(Γz − K z
2
= −(Γw − K e w )(Cw )−1 + s0w (Cw )−1 (2.18)
this is our propagator, we only need to know that the rational functions s0a exist, since S+
e SeT is
w z z w
completely known for us. Moreover we are left with (szw , szw , sww , sww ) as arbitrary functions
we can still choose. In other words we just show that solving (1.19) and then symmetrizing
is a consistent way to obtain propagators for the BCOV eqs.
4 Further ambiguities
Suppose we have solved the ambiguities for S ij . Then, we need to solve for S i . S i must
satisfy the equation
∂ ī S i = Gīj S ij = ∂ī Kj S ij (4.1)
where there is only sum over j.