• Non ci sono risultati.

where the parentheses are nested to depthn

N/A
N/A
Protected

Academic year: 2021

Condividi "where the parentheses are nested to depthn"

Copied!
2
0
0

Testo completo

(1)

Problem 11149

(American Mathematical Monthly, Vol.112, April 2005) Proposed by M. Ivan and I. Rasa (Romania).

Leta > 0. Find

n→∞lim n log(1 + log(1 + (· · · log(1 + a/n) · · · ))) where the parentheses are nested to depthn.

Solution proposed by Roberto Tauraso, Dipartimento di Matematica, Universit`a di Roma “Tor Vergata”, via della Ricerca Scientifica, 00133 Roma, Italy.

Let f (z) be the principal complex logarithm log(1 + z). We denote by fn(z) the nth-iterate of f in the right half plane H = {z ∈ C : Re(z) > 0} (note that f (H) ⊂ H). We will prove that

n→∞lim nfn(z/n) = 2z

z + 2 for z ∈ H (which is the required limit when z is a real positive number).

Since |f (z)| ≤ |z| for z ∈ H (see the final remark) then

|nfn(z/n)| = n|f (fn−1(z/n))| ≤ n|fn−1(z/n)| ≤ n|z/n| = |z|

and therefore the sequence of analytic functions {nfn(z/n)} is locally bounded on H. If we show that this sequence converges to the analytic function g(z) = 2z/(z + 2) in (0, 2) (which has an accumulation point in H), then by Vitali’s theorem the sequence converges uniformly on compact sets of H to g.

LetP

k=1an,kzkbe the expansion of fn(z) at zero. Then a1,k= (−1)k+1/k and, since fn= fn−1◦ f and f(k)(0) = (−1)k+1(k − 1)!, by the Fa`a di Bruno’s theorem we are able find a recursion formula for the coefficients for n ≥ 2:

an,k = fn(k)(0)

k! =X

π(k)

f(1)(0)r1f(2)(0)r2

· · ·f(k)(0)rk

r1!r2! · · · rk! · (1!)r1(2!)r2· · · (k!)rk · fn−1(r) (0)

= X

π(k)

 r

r1, r2, . . . , rk



· (−1)k+r

1r12r2· · · krk · an−1,r

where the sum runs over all partitions π(k) of the positive integer k: k = 1r1+ 2r2+ · · · + krk, ri

denotes the number of parts of size i, and r = r1+ r2+ · · · + rk is the total number of parts.

It is easy to determine an explicit formula for the first coefficients:

an,1 = an−1,1 = a1,1 = 1

an,2 = an−1,2− an−1,1/2 = an−1,2− 1/2 = −n/2

an,3 = an−1,3− an−1,2+ an−1,1/3 = an−1,3− (n − 1)/2 + 1/3 = n2/4 + n/12.

Now we show by induction with respect to k that

an,k =

−n 2

k−1

+ Qk(n)

where Qk is a polynomial of degree k − 2. The inductive step: by the recursion formula

an,k= an−1,k− k − 1 2



an−1,k−1+

k−2

X

r=1

ck,ran−1,r

(2)

where the first two terms are associated to the partitions 1 · k and 1 · (k − 2) + 2 · 1. Hence

an,k = a1,k+

n−1

X

j=1

− k − 1 2



aj,k−1+

k−2

X

r=1

ck,raj,r

!

= a1,k+

n−1

X

j=1

− k − 1 2

 

−j 2

k−2

+ Qk−1(j)

! +

k−2

X

r=1

ck,raj,r

!

= (k − 1)



−1 2

k−1 n−1

X

j=1

jk−2+ O

n−1

X

j=1

jk−3

=



−1 2

k−1

nk−1+ O(nk−2).

Finally

nfn(z/n) =

X

k=1

an,k

nk−1 · zk=

X

k=1



−1 2

k−1

+Qk(n) nk−1

! zk

and for z ∈ (0, 2) taking the limit as n goes to infinity we have that

n→∞lim nfn(z/n) = z

X

k=1

−z 2

k−1

= z

1 − (−z/2) = 2z 2 + z.



Remark. | log(1 + z)| ≤ |z| for all z in the right half-plane H.

The function h(z) = log(1 + z)/z is analytic in a neighborhood of H then by the maximum modulus theorem it suffices to show that |h(z)| ≤ 1 on the boundary of the domain H ∩ {|z| < r} for r greater than some r0. On the imaginary axes z = iy:

| log(1 + z)|2= |(1/2) log(1 + y2) + i arctan y|2= (1/4)(log(1 + y2))2+ (arctan y)2≤ y2= |z|2. Moreover, on the semicircumference z = re with θ ∈ [−π/2, π/2]:

| log(1 + z)| ≤ log |1 + z| + π/2 ≤ log(1 + r) + π/2 ≤ (1/2)(r − 1) + log 2 + π/2 ≤ r = |z|

for r ≥ r0= 2 log 2 + π − 1.

Riferimenti

Documenti correlati

The seismic performance of the statues is comparatively assessed for the base isolated floor and a conventional fixed base configuration, by referring to three limit states

Valutando i dati su base varietale, dalla Tab. 5 relativa ai parametri di base si puÒ osservare come i campioni delle varieta Rebo e Teroldego manifestassero un più elevato

The method, tested on ill-posed problems with this feature, appears to be more reliable than the other consid- ered nonnegatively constrained regularization methods from the point

This, together with the high serum prolactin levels compared to those reported by other authors for healthy domestic dogs, may indicate that prolactin might be a good biomarker

Similarly to what has been done for Stacksort (more recently by Defant and less recently by Bousquet-Mélou), we study preimages of permutations under Queue- sort. Characterization

We used a multiple regression analysis to test pre- dictions derived from social cognition theory stat- ing that raters who had higher levels of social cognitive intelligence

Architecture must be able to respond to emergency, wherever a natural disaster or an unexpected political event involves the established order of a community, where populations

In this context, this paper proposes an integrated methodology for PSS addressing both technical and business aspects; it adopts a QFD-based approach to structure PSS