Calculus of Several Variables (Serge Lang)
Calculus of Several Variables (Serge Lang)
Abstract
In this paper we prove an asymptotically sharp Bernstein-type inequal-
ity for polynomials on analytic Jordan arcs. Also a general statement on
mapping of a domain bounded by finitely many Jordan curves onto a
complement to a system of the same number of arcs with rational func-
tion is presented here. This fact, as well as, Borwein-Erdélyi inequality
for derivative of rational functions on the unit circle, Gonchar-Grigorjan
estimate of the norm of holomorphic part of meromorphic functions and
Totik’s construction of fast decreasing polynomials play key roles in the
proof of the main result. 1
Classification (MSC 2010): 41A17, 30C20, 30E10
Introduction
Let T := {z ∈ C : |z| = 1} denote the unit circle, D := {z ∈ C : |z| < 1} denote
the unit disk and C∞ := C ∪ {∞} denote the extended complex plane. We also
use D∗ := {z ∈ C : |z| > 1} ∪ {∞} for the exterior of the unit disk and k.kK for
the sup norm over the set K.
First, we recall a Bernstein-type inequality proved by Borwein and Erdélyi
in [BE96] (and in a special case, by Li, Mohapatra and Rodriguez in [LMR95]).
We rephrase their inequality using potential theory (namely, normal derivatives
of Green’s functions) and for the necessary concepts, we refer to [ST97] and
[Ran95]. Then we present one of our main tools, the “open-up” step in Propo-
sition 5, similar step was also discussed by Widom, see [Wid69], p. 205–206
and Lemma 11.1. This way we switch from polynomials and Jordan arcs to
rational functions and Jordan curves. Then we use two conformal mappings,
Φ1 and Φ2 to map the interior of the Jordan domain onto the unit disk and
to map the exterior of the domain onto the exterior of the unit disk respec-
tively. We transform our rational function with Φ1 and “construct” a similar
rational function (approximate with another, suitable rational function) so that
the Borwein-Erdélyi inequality can be applied.
Our main theorem is the following.
Theorem 1. Let K be an analytic Jordan arc, z0 ∈ K not an endpoint. Denote
the two normals to K at z0 by n1 (z0 ) and n2 (z0 ). Then for any polynomial Pn
1 This is author accepted manuscript, including a few typo corrections. The published
1
of degree n we have
+
X |aj |2 − 1 −
X 1 − |aj |2
Bm (u) := , Bm (u) := ,
|aj − u|2 |aj − u|2
j: |aj |>1 j: |aj |<1
+ −
and Bm (u) := maxQ (Bm (u) , Bm (u)). If R is a polynomial with deg(R) ≤ m
m
and f (u) = R (u) / j=1 (u − aj ) is a rational function, then
If all the poles of f are inside or outside of D, then this result was improved
in [LMR95], Theorem 2 and Corollary 2 on page 525 using different approach.
We need to relax the condition on the degree of the numerator and the
denominator.
If we could allow poles at infinity, then the degree of the numerator can be
larger than that of the denominator. More precisely, we can easily obtain the
following
Theorem 3. Using the notations from Borwein-Erdélyi
Qm Theorem, if R is a
polynomial with deg(R) > m and f (u) = R (u) / j=1 (u − aj ) is a rational
function, then
|f 0 (u) | ≤ max Bm
+ −
(u) + deg (R) − m, Bm (u) ||f ||T , w ∈ T. (1)
2
f (u)
Proof. Let d := deg (R) − m > 0, and let f1 (τ ; u) = f1 (u) := (u−τ )d
, where
d d
τ ∈ R, τ > 1. Then (τ − 1) |f1 (u) | ≤ |f (u) | ≤ (τ + 1) |f1 (u) | for |u| = 1, so
1
||f1 ||T ≤ d
||f ||T .
(τ − 1)
1 1
|f10 (u) | ≥ |f 0 (u) | d
− d||f ||T d+1
.
(τ + 1) (τ − 1)
τ2 − 1
|f10 (u) | ≤ max Bm
+
(u) + d , B −
m (u) ||f1 ||T .
|u − τ |2
Letting τ → ∞ and combining the last three displayed estimates, we obtain the
Theorem.
Note that if we let all the poles tend to infinity, then we get back the original
Bernstein (Riesz) inequality for polynomials on the unit disk. Let us also re-
mark that the original proof of Borwein and Erdélyi also proves (1), with little
modifications.
The relation with Green’s functions is as follows. It is well known (see e.g.
[ST97], p.109) that Green’s function of the unit disk D with pole at a ∈ D is
1 − au
gD (u, a) = log
u−a
and Green’s functions of the complement of the unit disk D∗ = {|u| > 1} ∪ {∞}
with pole at a ∈ C, |a| > 1 and with pole at infinity are
1 − au
gD∗ (u, a) = log
and gD∗ (u, ∞) = log |u|.
u−a
3
Theorem 4. Let f (u) = R (u) /Q (u) be an arbitrary rational function with no
poles on the unit circle where R and Q are polynomials. Denote the poles of
f on C∞ by a1 , . . . , am ∈ C∞ \ {|u| = 1} where each pole is repeated as many
times as its order. Then, for u ∈ T,
Note that if deg (R) > deg (Q), then f has a pole at ∞, therefore it is
repeated deg (R) − deg (Q) times and this pole at ∞ is taken into account in
the second term of maximum. Inequality (5) is sharp, the factor on the right
hand side cannot be replaced for smaller constant, see, e.g., [BE95], p. 324.
4
1/F 0 = Q2 /A, that is, the location of the poles are known. Our goal is to find
β0 , β1 , β2 , . . . , β2k0 ∈ C such that
ˆ
1
P2k0 βj du is a rational function.
β0 + j=1 u−ζj
Q
k (u−ζk ) Q
Or equivalently, F1 (u) := β0
Q P must have 0 residue
k (u−ζk )+ j>0 βj k6=j (u−ζk )
everywhere, Res (F1 , u) = 0 for all Q
Q u ∈ C. Since ζk ’s are pairwise different,
k6=j (u − ζk ), j = 1, 2, . . . , 2k0 and k (u − ζk ) are linearly independent, so we
can choose βj ’s so that
Y X Y 2k
β0 (u − ζk ) + βj (u − ζk ) = (u − u∗ ) 0
k j>0 k6=j
5
Since F has no critical values outside K, the F (κj,δ )’s are smooth Jordan
curves. Fix b ∈ C \ K, then there is (at least one) b0 ∈ G with F (b0 ) = b,
because F (G) is open, F (G) ⊂ C \ K and F (∂G) = F (κ1 ∪ . . . ∪ κk0 ) ⊂
K. If δ > 0 is small enough, then b ∈ IntF (κ0,δ ) and b ∈ C \ IntF (κj,δ )
(j = 1, . . . , k0 ), so index (b, F (κ0,δ ) ∪ F (κ1,δ ) ∪ . . . ∪ F (κk0 ,δ )) = 1. Therefore
index (b0 , κ0,δ ∪ κ1,δ ∪ . . . ∪ κk0 ,δ ) = 1, so there is exactly one inverse image, this
shows the univalence of F .
We can give another proof for the univalence as follows. There is a (local)
branch of F −1 such that F −1 [z] = z/c1 + .. as z → ∞, in other words, ∞
is not a branch point of F −1 . Furthermore, the function F has branch points
only at ζj ’s, j = 1, . . . , 2k0 and it behaves as a square root there. Therefore
every analytic continuations along any curve in C \ K give the same function
element. Now we use Lemma 2, p. 175 in [SFS89] with this (local) branch.
Therefore we can choose a (global) regular branch of F −1 such that F −1 [∞] =
∞. Since this branch is regular and F is a rational function, there is no other
inverse image of ∞ by F −1 in G. By the construction of G and applying the
maximum principle, we have gC∞ \K (F (u) , ∞) ≡ gG (u, ∞), u ∈ G. Using the
majorization principle (see [Kal08], Theorem 1 on p. 624) or Theorem 4.4.1
on p. 112 from [Ran95], we obtain that F is conformal bijection from G onto
C∞ \ K.
As for the smoothness assertion (∂G analytic), this follows from standard
considerations as follows. Without loss of generality, we may assume that z =
κ (t) = t + c1 t + c2 t2 + . . ., is a convergent power series for 0 ≤ t ≤ t0 and
z = F (u) is such that F (0) = 0, F 0 (0) = 0 and F 00 (0) 6= 0. It is known,
see e.g. [Sto62], p. 286, that √ the two branches of the inverse of F near z = 0
can be written as G0 (z) ± zG1 (z) where G0 , G1 are holomorphic functions.
Denote them by F1−1 and F2−1 . This way γ1 (t) := F1−1 κ t2 = G0 κ t2 +
p
t 1 + κ1 (t2 )G1 κ t2 is a convergent power series in t ∈ [0, t1 ] and similarly
for γ2 (t) := F2−1 κ t2 and γ10 (0) 6= 0. Considering γ1 (−t) for t ∈ [0, t1 ],
Fix one, C 2 smooth Jordan arc γ with endpoints ζ1 and ζ2 and let z ∈ γ,
z 6= ζ1 , z 6= ζ2 . Denote the two normal vectors of unit length at z to γ by
n1 (z), n2 (z), where n1 (z) = −n2 (z). We may assume that n1 and n2 depend
continuously on z. We use the same letter for normals in different planes and
from the context, it is always clear that which arc we refer to. We use the
rational mapping F and the domain G2 := G from the previous Proposition for
γ. Denote the inward normal vector to ∂G at u ∈ ∂G by n2 (u) and the outward
normal vector to ∂G at u by n1 (u), n2 (u) = −n1 (u). It is easy to see that
6
Figure 1: The γ, z, G1 and G2 with the normal vectors
∂ ∂
gG1 u1 , F1−1 (b) / |F 0 (u1 )|
gC∞ \K (z, b) =
∂n1 (z) ∂n1 (u1 )
∂
gG2 u2 , F2−1 (b) / |F 0 (u2 )|
=
∂n2 (u2 )
and, similarly for the other side,
∂ ∂
gG1 u2 , F1−1 (b) / |F 0 (u2 )|
gC∞ \K (z, b) =
∂n2 (z) ∂n1 (u2 )
∂
gG2 u1 , F2−1 (b) / |F 0 (u1 )| .
=
∂n2 (u1 )
For arbitrary polynomial P , let fP (u) = f (u) := P (F (u)). Then kP kγ =
kf k∂G .
Proof. This immediately follows from the conformal invariance of Green’s func-
tions
gC∞ \K (F (u) , b) = gG1 u, F1−1 (b)
and
gC∞ \K (F (u) , b) = gG2 u, F2−1 (b) .
7
This Proposition implies that it is enough to take into account the normal
derivatives at, say, u1 only , i.e. ∂n2∂(u1 ) gG2 u1 , F2−1 (b) and ∂n1∂(u1 ) gG1 u1 , F1−1 (b)
only.
8
Figure 2: The two Riemann mappings and the points
and if a2 = ∞, then
∂ ∂
gG2 u0 , F2−1 [∞] =
gD∗ (1, ∞) = 1.
∂n2 (u0 ) ∂n2 (1)
Proof. The second equalities in all three lines follow from (2), (3) and (4).
We know that Φ1 (1) = u0 and Φ2 (1) = u0 , moreover |Φ01 (1)| = 1, |Φ02 (1)| =
1 imply that nj (1) is mapped to nj (u0 ) by Φj , j = 1, 2 and the mappings Φj ,
−1
j = 1, 2 also preserve the length at 1 (there is no magnifying factor Φ0j (1)
unlike at Proposition 6). Using the conformal mappings Φ1 and Φ2 , and the
conformal invariance of Green’s functions, we obtain the first equalities in all
three lines.
9
Theorem. Let G̃ ⊂ C∞ be a domain, ∞ ∈ G̃ and denote its Green’s function
by gG̃ (u, ∞) with pole at infinity. Let f˜ : G̃ → C∞ be a meromorphic function
which has only one pole at infinity and we denote the order of the pole by ñ.
Assume that f˜ can be extended continuously to the boundary ∂ G̃ of G̃. Then
˜
˜
f (u) ≤
f
exp (ñ gG̃ (u, ∞)) (8)
∂ G̃
δ2,3 < 1)
1p p
δ2,3 ≤ |w − b1 | ≤ 2 δ2,3 . (9)
2
Let n o
∗ (0)
Kw := w : |w| ≤ 1 + δ2,3 \ D3 .
∗
Obviously, this Kw is a compact set and satisfies the touching-outer-disk con-
1−a2
dition at b1 = 1−a2 of Totik’s theorem. See figure 3 later.
Consider
Ku∗ := Φ2 ◦ ψ −1 [Kw
∗
∩ D∗ ] ∪ Φ1 ◦ ψ −1 [Kw
∗
∩ D∗ ] ∪ G1 .
10
Totik’s theorem, there exists a fast decreasing polynomial for Ku∗ at u0 of degree
at most n1 which we denote by Q = Q (n1 ; u). More precisely, Q has the
109/110
following properties: Q (u0 ) = 1, |Q (u)| ≤ 1 on u ∈ Ku∗ , deg Q ≤ n1 ≤ n1
−9/10
and if |u − u0 | > n1 , u ∈ Ku∗ , then
1/110
|Q (u)| ≤ C2 exp −C3 n1 . (10)
√
Let n1 := b nc, n2 := n3/4 , δ2,1 := 1/n and δ2,3 := n−2/3 .
4.2 Proof
In this subsection, we let f (u) := Pn (F (u)) where Pn is a fixed polynomial of
degree n and F is the open-up rational function (see Proposition 5) for K (from
Theorem 1).
Actually, we use only the following facts. f is a rational function such that it
has one pole in G1 and one in G2 . We know that the poles of f are ∞ = F2−1 [∞]
and F1−1 [∞], and the order of the pole in G1 is n.
It is easy to decompose f into sum of rational functions, that is,
f = f1 + f2
kf2 k∂G2 ≤ C1 G+
1 (log n + 1) kf k∂G2 . (11)
Obviously, we have
kf1 k∂G2 ≤ 1 + C1 G+
1 (log n + 1) kf k∂G2 . (12)
Consider
ϕ1 (v) := f1 (Φ1 (v)) .
This is a meromorphic function in D1 . We may assume that ϕ1 has only one
pole in D1 otherwise we can decrease r2 > 1 so that the pole in G2 is not in
Φ1 (D1 ) = G+
1 . We know that
kϕ1e k∂D ≤ C1 (D) (log n + 1) kQ ◦ Φ1 · ϕ1 k∂D ≤ C1 (D) (log n + 1) kϕ1 k∂D (15)
11
As a remark, let us note that we may write log n + 1 ≤ O (log n) for sim-
plicity since we are interested in asymptotics as n → ∞ in the main theorem.
Otherwise, if n = 0 or n = 1, then Pn is a constant or linear polynomial and
the error term o(1) in the main theorem (Theorem 1) can be sufficiently large
(depending on K and z0 ) for these two particular values of n. In this manner,
we write (log n + 1) in general, but we simplify it to O(log n) frequently.
Furthermore, we can estimate ϕ1e (v) on v ∈ D1 \ D as follows
|ϕ1e (v)| = |(Q · f1 ) ◦ Φ1 (v) − ϕ1r (v)| ≤ |(Q · f1 ) ◦ Φ1 (v)| + |ϕ1r (v)| . (16)
∗
We also need to estimate Q outside D (and Kw ) as follows. Using deg Q ≤
109/110
n1 ≤ n1 and Bernstein-Walsh estimate (8), we can write for v ∈ D1 \ D
This way we can continue (16) and we use u = Φ1 (v) here and that ϕ1r is
a rational function with no poles outside D and the maximum principle for ϕ1r
≤ eC6 n1 |f1 (u)| + kϕ1r k∂D ≤ eC6 n1 kf1 k∂G2 + kϕ1 k∂D + kϕ1e k∂D
and here we used that f1 has no pole in G2 and the maximum principle. We
can estimate these three sup norms with the help of (12) and (13), (12) and
(15), (13), (12). Hence we have for v ∈ D1 \ D
Approximate and interpolate ϕ1e as follows with rational function which has
only one pole, namely at a2 = Φ−1
2 [∞]. Consider ϕ1e ◦ψ
−1
(w) on ψ (D1 ). Using
the properties of ψ, we have
12
It is well known that p1,N does not depend on Γ1 . Since b1 is a double pole
0
of qN , therefore p1,N and p01,N coincide there with ϕ1e ◦ ψ −1 and ϕ1e ◦ ψ −1
respectively.
The error of the approximating polynomial p1,N to ϕ1e ◦ ψ −1 is
ˆ
−1 1 ϕ1e ◦ ψ −1 (ω) qN (w)
ϕ1e ◦ ψ (w) − p1,N (w) = dω
2πi Γ1 ω−w qN (ω)
ˆ
1 1 ϕ1e ◦ ψ −1 (ω)
= qN (w) dω, (18)
2πi Γ1 ω−w qN (ω)
here w ∈ D can be arbitrary. It is easy to see that for w ∈ D, |qN (w)| ≤ 4 and
ˆ
1 1
|dω| ≤ 1 + δ1 .
2π Γ1 ω − w
δ1
Therefore, using (17), we can estimate the error (of approximation of p1,N to
ϕ1e ◦ ψ −1 ) as follows
√
which tends to 0 as n → ∞, because n1 = b nc and
eC6 n1 √
N
= exp C6 n − log (1 + δ1 ) n (1 + o (1)) → 0.
(1 + δ1 )
13
and
gG2 Φ1 ◦ ψ −1 (w) , ∞ ≤ C4 (|w| − 1) .
(21)
Furthermore, there exists C5 > 0 which depends on G2 and independent of Pn , n
(0)
and f such that for all 1 ≤ |ζ| ≤ 1 + δ2,3 we have
ψ ◦ Φ−1 ◦ Φ ◦ ψ −1 0 (ζ)
2 1
≤ 1 + C5 |ζ − b1 |
ψ ◦ Φ−1 −1 (ζ)
2 ◦ Φ 1 ◦ ψ
and
gG2 Φ1 ◦ ψ −1 (ζ) , ∞ ≤ (|ζ| − 1) (1 + C5 |ζ − b1 |) .
(22)
Proof. For simplicity, let ζ ∗ := arg ζ where arg ζ = ζ/ |ζ|, if ζ 6= 0 and arg 0 = 0.
We can express Green’s function in the following ways for u ∈ G2 ,
and for w ∈ D∗
The first displayed inequality in the Lemma comes from continuity consider-
ations and the conformal bijection properties. Integrating this inequality along
radial rays, we obtain (21). If we are close to 1, then more is true:
−1 0
ψ ◦ Φ−1
◦ Φ ◦ ψ (b ) = 1.
2 1 1
Using continuity, we see that there exists C5 > 0 such that for all ζ, 1 ≤
(0)
|ζ| ≤ 1 + δ2,3 , we have
ψ ◦ Φ−1 ◦ Φ ◦ ψ −1 0 (ζ)
2 1
≤ 1 + C5 |ζ − b1 | .
ψ ◦ Φ−1 −1 (ζ)
2 ◦ Φ 1 ◦ ψ
= (|ζ| − 1) (1 + C5 |ζ − b1 |) .
14
∗
Figure 3: Kw and the arcs that make up Γ2
will make up Γ2 . Let Γ2,3 be the (shorter, circular) arc between w1∗ (δ2,3 ) and
1+δ 1+δ2,1
w2∗ (δ2,3 ), Γ2,1 be the longer circular arc between w1∗ (δ2,3 ) 1+δ2,1
2,3
and w2∗ (δ2,3 ) 1+δ2,3 ,
∗
Γ2,2 := {w : 1 + δ2,1 ≤ |w| ≤ 1 + δ2,3 , arg w = arg (w1 (δ2,3 ))} and similarly Γ2,4 :=
{w : 1 + δ2,1 ≤ |w| ≤ 1 + δ2,3 , arg w = arg (w2∗ (δ2,3 ))} be the two segments con-
necting Γ2,1 and Γ2,3 . Finally let Γ2 be the union of Γ2,1 , Γ2,2 , Γ2,3 and Γ2,4 .
∗
Figure 3 depicts these arcs and Kw defined above.
We estimate the error of p2,N to (Q · f2 )◦Φ1 ◦ψ −1 on each integral separately:
ˆ
−1 1 (Q · f2 ) ◦ Φ1 ◦ ψ −1 (ω) qN (w)
(Q · f2 ) ◦ Φ1 ◦ ψ (w) − p2,N (w) = dω
2πi Γ2 ω−w qN (ω)
ˆ ˆ ˆ ˆ !
1
= + + + .
2πi Γ2,1 Γ2,2 Γ2,3 Γ2,4
For the first term, we use Bernstein-Walsh estimate (8) for the polynomial f2
polynomial Q as follows. If w ∈ Γ2,1 , then with
on G2 and the fast decreasing
(21), gG2 Φ1 ◦ ψ −1 (w) , ∞ ≤ C4 δ2,1 = C4 /n, therefore
f2 Φ1 ◦ ψ −1 (w) ≤ kf2 k C4
≤ kf k∂G2 C1 G+ C4
∂G2 exp n 1 (log n + 1) e
n
= O (log (n)) kf k∂G2
15
where we used (11). Now we use the fast decreasing property of Q as fol-
∗ (0)
lows. We know that Γ2,1 ⊂ Kw (if n ≥ 1/δ2,3 ) and with the elementary
p −9/10
geometric considerations (9) we have δ2,3 /2 ≥ n1 which is equivalent
−1/3
to nn /2 ≥ n−9/20 (this is true oif n is large). It is also important that
0
sup Φ1 ◦ ψ −1 (w) : w ∈ ψ (D1 ) < ∞ and Kw ∗
⊂ ψ (D1 ) therefore the
1 ˆ
(Q · f2 ) ◦ Φ1 ◦ ψ −1 (ω) qN (w)
dω
2πi Γ2,1 ω−w qN (ω)
ˆ !
1 1 log (n) 1
≤ O kf k∂G2 4 N 2
|dω|
2π Γ2,1 |ω − w| exp C3 n 1/220
(1 + δ2,1 ) δ2,1
! !
2 2π (1 + δ2,1 ) log (n) n3 log (n)
≤ O kf k∂G2 = O kf k∂G2
π (1 + δ2,1 )N δ2,1
3 exp C3 n1/220 exp C3 n1/220
1 ˆ
(Q · f2 ) ◦ Φ1 ◦ ψ −1 (ω) qN (w)
dω
2πi Γ2,3 ω−w qN (ω)
ˆ
1 1 (Q · f2 ) Φ1 ◦ ψ −1 (ω)
1
≤ 4 |dω| . (23)
2π Γ2,3 |ω − w| |qN (ω)|
2 N
Here, |ω| = 1 + δ2,3 , |w − ω| ≥ δ2,3 , |qN (ω)| ≥ δ2,3 (1 + δ2,3 ) . Roughly speak-
−1
ing, f2 grows and this time Q grows too (the bad guys) and only |qN (ω)|
decreases (the good guy). We estimate their growth using Bernstein-Walsh es-
timate (8) for f2 on G2 and Lemma 9 (and estimate (11) as well) in the following
way. Here, as earlier, ω ∈ Γ2,3
f2 Φ1 ◦ ψ −1 (ω) ≤ kf2 k −1
∂G2 exp ngG2 Φ1 ◦ ψ (ω) , ∞
≤ C1 G+
1 (log n + 1) kf k∂G2 exp (n (|ω| − 1) (1 + C5 |ω − b1 |))
≤ C1 G+
p
1 (log n + 1) kf k∂G2 exp nδ2,3 + C5 nδ2,3 2 δ2,3
= C1 G+ 2C5
1 (log n + 1) kf k∂G2 exp (nδ2,3 ) e
16
As for qN ,
1 1 1 1
≤ 2 = 2 exp (− (n + n1 + n2 ) log (1 + δ2,3 ))
|qN (ω)| δ2,3 (1 + δ2,3 )N δ2,3
!
2
1 δ2,3
≤ 2 exp −nδ2,3 − n1 δ2,3 − n2 δ2,3 + (n + n1 + n2 )
δ2,3 2
1
≤ 2 exp (−nδ2,3 − n1 δ2,3 − n2 δ2,3 ) exp 3n n−4/3
δ2,3
exp (−nδ2,3 − n1 δ2,3 − n2 δ2,3 ) 3
≤ 2 e
δ2,3
As for Q (this time it is a bad guy), we use Bernstein-Walsh estimate (8) for
Q on G1 ∪ ∂G1 and that G1 ∪ ∂G1 ⊂ Ku∗ . Therefore, kQk∂G2 = 1 and we know
109/110
that deg Q ≤ n1 ≤ n109/220 , hence
Q Φ1 ◦ ψ −1 (ω) ≤ kQk −1
∂G2 exp n1 gG2 Φ1 ◦ ψ (ω) , ∞
≤ exp (n1 (|ω| − 1) (1 + C5 |ω − b1 |)) ≤ exp n109/220 δ2,3 1 + C5 2 δ2,3
p
= exp n109/220 δ2,3 + 2C5 n109/220 n−1 ≤ exp n109/220 δ2,3 e2C5 .
where we used several estimates: length of Γ2,3 is at most 4π, the definitions
of
n1 , n2 and δ2,3 and that n1 > n109/220 , therefore exp n109/220 − n1 δ2,3 ≤ 1.
For Γ2,2 and Γ2,4 , we apply the same estimate which we detail for Γ2,2 only.
We again start with the integral for w ∈ D
1 ˆ
(Q · f2 ) ◦ Φ1 ◦ ψ −1 (ω) qN (w)
dω
2πi Γ2,2 w−ω qN (ω)
ˆ
1 1 (Q · f2 ) Φ1 ◦ ψ −1 (ω)
1
≤ 4 |dω| . (24)
2π Γ2,2 |w − ω| |qN (ω)|
17
Since ω ∈ Γ2,2 , we can rewrite it in the form ω = (1 + δ) w1∗ / |w1∗ | where δ2,1 ≤
δ ≤ δ2,3 (with w1∗ = w1∗ (δ2,3 ) ). We use essentially the same steps to estimate
f2 (the only one bad guy this time) and qN and Q (this time it is a good guy).
In estimating f2 , the only difference is that |ω| − 1 = δ, so
f2 Φ1 ◦ ψ −1 (ω) ≤ kf2 k −1
∂G2 exp ngG2 Φ1 ◦ ψ (ω) , ∞
≤ C1 G+
1 (log n + 1) kf k∂G2 exp (n (|ω| − 1) (1 + C5 |ω − b1 |))
≤ C1 G+
p
1 (log n + 1) kf k∂G2 exp nδ + C5 nδ2,3 2 δ2,3
= C1 G+ 2C5
1 (log n + 1) kf k∂G2 exp (nδ) e .
1 1 1 1
≤ 2 = 2 exp (− (n + n1 + n2 ) log (1 + δ))
|qN (ω)| δ2,1 (1 + δ)N δ2,1
!
2
1 δ2,3
≤ 2 exp −nδ − n1 δ − n2 δ + (n + n1 + n2 )
δ2,1 2
1
≤ 2 exp (−nδ − n1 δ − n2 δ) exp 3n n−4/3
δ2,1
exp (−nδ − n1 δ − n2 δ) 3 exp (−nδ) 3
≤ 2 e ≤ 2 e .
δ2,1 δ2,1
Putting these all together, we see that exp (nδ) cancels and actually Q make
the integrand small. So we can continue the estimate (24)
ˆ
2 1 2C5 exp (−nδ) 3
C1 G+
≤ 1 (log n + 1) kf k∂G2 exp (nδ) e 2 e
π δ
Γ2,2 2,1 δ2,1
ˆ
2e2C5 +3 C2 C1 G+
1
· C2 exp −C3 n1/220 |dω| = kf k∂G2 |dω|
π Γ2,2
!
log n + 1
1/220
n3 log (n)
· 3 exp −C3 n ≤ kf k∂G2 O
δ2,1 exp C3 n1/220
(0)
where we used that the length of Γ2,2 is at most 1 (since δ2,3 < 1) and δ2,1 = 1/n.
Summarizing these estimates on Γ2,1 , Γ2,3 and Γ2,2 (and also on Γ2,4 ), we
have uniformly for |w| ≤ 1,
p2,N (w) − (Q · f2 ) ◦ Φ1 ◦ ψ −1 (w) = o (1) kf k
∂G2
18
a2 (of p2,N ◦ ψ ) is deg p2,N = N = n + n1 + n2 = (1 + o (1)) n and using the
properties of w = ψ (v), we uniformly have for |v| ≤ 1
that is,
kp2,N ◦ ψ − (Q · f2 ) ◦ Φ1 k∂D = o (1) kf k∂G2 . (25)
0
Since b1 is double zero of q, p02,N (b1 ) = (Q · f2 ) ◦ Φ1 ◦ ψ −1 (b1 ), and dividing
0
both sides with ψ −1 (b1 ), we obtain
0 0
(p2,N ◦ ψ) (1) = ((Q · f2 ) ◦ Φ1 ) (1) . (26)
This function h has a pole at a1 (because of ϕ1,r ) and the order of its pole at
a1 is at most n, and h has a pole at a2 (because of p1,N ◦ ψ and p2,N ◦ ψ) and
the order of its pole at a2 is at most N = n (1 + o (1)). We use the identity
f ◦ Φ1 = (Q · f + (1 − Q) · f ) ◦ Φ1
where the second term is zero because of the fast decreasing polynomial (Q (u1 ) =
1) and for the first term we can apply Theorem 1.3 from [NT05] in the following
way (k1 − Qk∂G2 ≤ 2):
√ ∂
(1 − Q)0 · f (u1 ) · Φ01 (1) ≤ kf k
∂G2 n2 (1 + o (1)) gG (u1 , ∞)
∂n2 (u1 ) 2
√ ∂
= kf k∂G2 O n gG (u1 , ∞)
∂n2 (u1 ) 2
∂ ∂
≤ o (1) n kf k∂G2 max gG2 (u1 , ∞) , gG1 (u1 , a1 ) . (27)
∂n2 (u1 ) ∂n1 (u1 )
19
Now we apply the Borwein-Erdélyi inequality (5) for h as follows:
!
X ∂ X ∂
|h0 (1)| ≤ khk∂D max gD (1, α) , gD∗ (1, α) (30)
α
∂n1 (1) α
∂n2 (1)
where the summation is taken over all poles in D and in D∗ respectively, count-
ing multiplicities. We will continue this estimate later after simplifying these
expressions. Using Propositions 8 and 7, we can write
X ∂ ∂ ∂
gG1 u0 , F1−1 [∞]
gD (1, α) ≤ n gD (1, a1 ) = n
α
∂n1 (1) ∂n1 (1) ∂n1 (u0 )
∂
=n gC \K (z0 , ∞) |F 0 (u0 )|
∂n2 (z0 ) ∞
where in the last step we used Proposition 7 with z0 = F (u0 ) and identifying
u0 = u1 . Similarly, we can simplify the second term in the maximum in (30)
X ∂ ∂
gD∗ (1, α) = deg (p1,N + p2,N ) gD∗ (1, a2 )
α
∂n2 (1) ∂n2 (1)
∂ ∂
gG u0 , F2−1 [∞]
≤N gD∗ (1, a2 ) = (1 + o (1)) n
∂n2 (1) ∂n2 (u0 ) 2
∂
= (1 + o (1)) n gC \K (z0 , ∞) |F 0 (u0 )|
∂n1 (z0 ) ∞
where o (1) here does not depend on anything. Note that we “used a slightly
bit more the pole at a2 ”, but it does not cause problem. So we can continue the
main estimate (30)
∂
gG u0 , F1−1 [∞] ,
≤ khk∂D max n
∂n1 (u0 ) 1
∂
(1 + o (1)) n gG2 u0 , F2−1 [∞]
∂n2 (u0 )
≤ khk∂D (1 + o (1)) n
∂ ∂
gG1 u0 , F1−1 [∞] , gG2 u0 , F2−1 [∞]
· max .
∂n1 (u0 ) ∂n2 (u0 )
Summarizing these estimates, we have for h
(Q · f )0 (u1 ) ≤ kQ · f k
∂G2 (1 + o (1)) n
∂ −1
∂ −1
· max gG u0 , F1 [∞] , gG u0 , F2 [∞]
∂n1 (u0 ) 1 ∂n2 (u0 ) 2
∂ −1
∂ −1
+o (1) n kf k∂G2 ·max gG u0 , F1 [∞] , gG u0 , F2 [∞] .
∂n1 (u0 ) 1 ∂n2 (u0 ) 2
20
Figure 4: The sets K and K ∗
5 Sharpness
In this section we show that the result is asymptotically sharp, that is, we prove
Theorem 2. The idea is similar to that of [NT13]. Note that we assume C 2
smoothness only.
Proof. We may assume that
∂ ∂
gC∞ \K (z0 , ∞) ≤ gC \K (z0 , ∞) .
∂n1 (z0 ) ∂n2 (z0 ) ∞
Furthermore, we assume that n1 (.) and n2 (.) are defined on the component of
K containing z0 and they are continuous there except for the endpoints.
It is easy to see that for every ε > 0 there exists a compact set K ∗ = K ∗ (ε)
such that ∂K ∗ is finite union of disjoint, C 2 smooth Jordan curves, K ⊂ K ∗ ,
z0 ∈ ∂K ∗ and the normal vector n (K ∗ , z0 ) to K ∗ (pointing outward) at z0 is
equal to n2 (z0 ) and
∂ ∂
gC \K (z0 , ∞) (1 − ε) ≤ gC \K ∗ (z0 , ∞)
∂n2 (z0 ) ∞ ∂n (K ∗ , z0 ) ∞
∂
≤ gC \K (z0 , ∞) .
∂n2 (z0 ) ∞
21
that
0
Pε,n (z0 ) ≥ n (1 − oε (1)) kPε,n k ∂
K ∗ (ε) gC \K ∗ (z0 , ∞)
∂n (K ∗ , z0 ) ∞
∂
≥ n (1 − oε (1)) (1 − ε) kPε,n kK gC \K (z0 , ∞)
∂n2 (z0 ) ∞
where oε (1) depends on K ∗ (ε) and z0 and tends to 0 as deg Pε,n → ∞. Since
ε was arbitrary, we see that (1 − oε (1)) (1 − ε) = 1 − o (1), that is, choosing a
suitable subsequence of {Pε,n } we obtain the assertion.
Acknowledgement
The first author was supported by the European Research Council Advanced
grant No. 267055, while he had a postdoctoral position at the Bolyai Institute,
University of Szeged, Aradi v. tere 1, Szeged 6720, Hungary.
The second author, Béla Nagy was supported by Magyary scholarship: This
research was realized in the frames of TÁMOP 4.2.4. A/2-11-1-2012-0001 „Na-
tional Excellence Program - Elaborating and operating an inland student and
researcher personal support system.” The project was subsidized by the Euro-
pean Union and co-financed by the European Social Fund.
The authors are deeply indebted to Professor Vilmos Totik for several helpful
suggestions and comments which enabled us to improve the presentation of our
result and for drawing our attention to Widom’s open-up result.
References
References
[AKML98] Dmitri Alekseevsky, Andreas Kriegl, Peter W. Michor, and Mark
Losik, Choosing roots of polynomials smoothly, Israel J. Math. 105
(1998), 203–233. MR 1639759 (2000c:58017)
[BE95] Peter Borwein and Tamás Erdélyi, Polynomials and polynomial
inequalities, Graduate Texts in Mathematics, vol. 161, Springer-
Verlag, New York, 1995. MR 1367960 (97e:41001)
[BE96] , Sharp extensions of Bernstein’s inequality to rational
spaces, Mathematika 43 (1996), no. 2, 413–423 (1997). MR 1433285
(97k:26014)
[Con95] John B. Conway, Functions of one complex variable. II, Graduate
Texts in Mathematics, vol. 159, Springer-Verlag, New York, 1995.
MR 1344449 (96i:30001)
22
[GG76] A. A. Gončar and L. D. Grigorjan, Estimations of the norm of the
holomorphic component of a meromorphic function, Mat. Sb. (N.S.)
28 (1976), no. 4, 571–575.
[ST97] Edward B. Saff and Vilmos Totik, Logarithmic potentials with exter-
nal fields, Grundlehren der Mathematischen Wissenschaften [Fun-
damental Principles of Mathematical Sciences], vol. 316, Springer-
Verlag, Berlin, 1997, Appendix B by Thomas Bloom. MR 1485778
(99h:31001)
23
[Wid69] Harold Widom, Extremal polynomials associated with a system of
curves in the complex plane, Advances in Math. 3 (1969), 127–232
(1969). MR 0239059 (39 #418)
Sergei Kalmykov
Institute of Applied Mathematics, FEBRAS, 7 Radio Street, Vladivostok, 690041,
Russia,
Far Eastern Federal University, 8 Sukhanova Street, Vladivostok, 690950, Rus-
sia and
Bolyai Institute, University of Szeged, Szeged, Aradi v. tere 1, 6720, Hungary
email address: [email protected]
Béla Nagy
MTA-SZTE Analysis and Stochastics Research Group, Bolyai Institute, Univer-
sity of Szeged, Szeged, Aradi v. tere 1, 6720, Hungary
email address: [email protected]
24