uniolympiads English

IMC 2013, Ημέρα 1, Θέμα 5

Does there exist a sequence (an)(a_n) of complex numbers such that for every positive integer pp we have that ∑n=1∞anp\sum_{n=1}^{\infty} a_n^p converges if and only if pp is not a prime?

Πρωτότυπο από το επίσημο PDFDoes there exist P a sequence (a n ) of complex numbers such that for every p positive integer p we have that ∞ n=1 a n converges if and only if p is not a prime?

Η μεταγραφή συγκρίθηκε σύμβολο προς σύμβολο με το πρωτότυπο, όχι ακόμα από άνθρωπο.

Χρόνος σε αυτό το θέμα 0:00

Δούλεψε τουλάχιστον 45 λεπτά πριν ανοίξεις τη λύση.

Επίσημη λύση

(Proposed by Tomáš Bárta, Charles University, Prague)

Solution. The answer is YES. We prove a more general statement; suppose that N=C∪DN = C \cup D is an arbitrary decomposition of NN into two disjoint sets. Then there exists a sequence (an)n=1∞(a_n)_{n=1}^\infty such that ∑n=1∞anp\sum_{n=1}^\infty a_n^p is convergent for p∈Cp \in C and divergent for p∈Dp \in D.

Define Ck=C∩[1,k]C_k = C \cap [1,k] and Dk∩[1,k]D_k \cap [1,k].

Lemma. For every positive integer kk there exists a positive integer NkN_k and a sequence Xk=(xk,1,…,xk,Nk)X_k = (x_{k,1}, \ldots, x_{k,N_k}) of complex numbers with the following properties:

(a) For p∈Dkp \in D_k, we have ∣∑j=1Nkxk,jp∣≥1\displaystyle\left|\sum_{j=1}^{N_k} x_{k,j}^p\right| \ge 1.

(b) For p∈Ckp \in C_k, we have ∑j=1Nkxk,jp=0\displaystyle\sum_{j=1}^{N_k} x_{k,j}^p = 0; moreover, ∣∑j=1mxk,jp∣≤1k\displaystyle\left|\sum_{j=1}^{m} x_{k,j}^p\right| \le \frac{1}{k} holds for 1≤m≤Nk1 \le m \le N_k.

Proof. First we find some complex numbers z1…,zkz_1 \ldots, z_k with

∑j=1kzjp={0p∈Ck1p∈Dk(1)\sum_{j=1}^k z_j^p = \begin{cases} 0 & p \in C_k \\ 1 & p \in D_k \end{cases} \tag{1}
As is well-known, this system of equations is equivalent to another system σν(z1,…,zk)=wν\sigma_\nu(z_1, \ldots, z_k) = w_\nu (ν=1,2,…,k)(\nu = 1, 2, \ldots, k) where σν\sigma_\nu is the ν\nuth elementary symmetric polynomial, and the constants wνw_\nu are uniquely determined by the Newton-Waring-Girard formulas. Then the numbers z1,…,zkz_1, \ldots, z_k are the roots of the polynomial zk−w1zk−1+−…+(−1)kwkz^k - w_1 z^{k-1} + - \ldots + (-1)^k w_k in some order.

Now let

M=⌈max⁡1≤m≤k, p∈Ck∣∑j=1mzjp∣⌉M = \left\lceil \max_{1\le m\le k,\ p\in C_k} \left|\sum_{j=1}^m z_j^p\right| \right\rceil
and let Nk=k⋅(kM)kN_k = k\cdot(kM)^k. We define the numbers xk,1…,xk,Nkx_{k,1} \ldots, x_{k,N_k} by repeating the sequence (z1kM,z2kM,…,zkkM)(\frac{z_1}{kM}, \frac{z_2}{kM}, \ldots, \frac{z_k}{kM}) (kM)k(kM)^k times, i.e. xk,ℓ=zjkMx_{k,\ell} = \frac{z_j}{kM} if ℓ≡j(modk)\ell \equiv j \pmod{k}. Then we have
∑j=1Nkxk,jp=(kM)k∑j=1k(zjkM)p=(kM)k−p∑j=1kzjp;\sum_{j=1}^{N_k} x_{k,j}^p = (kM)^k \sum_{j=1}^k \big(\tfrac{z_j}{kM}\big)^p = (kM)^{k-p} \sum_{j=1}^k z_j^p;
then from (1) the properties (a) and the first part of (b) follows immediately. For the second part of (b), suppose that p∈Ckp \in C_k and 1≤m≤Nk1 \le m \le N_k; then m=kr+sm = kr+s with some integers rr and 1≤s≤k1 \le s \le k and hence
∣∑j=1mxk,jp∣=∣∑j=1kr+∑j=kr+1kr+s∣=∣∑j=1s(zjkM)p∣≤M(kM)p≤1k.\left|\sum_{j=1}^m x_{k,j}^p\right| = \left|\sum_{j=1}^{kr} + \sum_{j=kr+1}^{kr+s}\right| = \left|\sum_{j=1}^s \left(\frac{z_j}{kM}\right)^p\right| \le \frac{M}{(kM)^p} \le \frac{1}{k}.
The lemma is proved.

Now let Sk=N1…,NkS_k = N_1 \ldots, N_k (we also define S0=0S_0 = 0). Define the sequence (a)(a) by simply concatenating the sequences X1,X2,…X_1, X_2, \ldots:

(a1,a2,…)=(x1,1,…,x1,N1,x2,1,…,x2,N2,…,xk,1,…,xk,Nk,…);(1)(a_1, a_2, \ldots) = (x_{1,1}, \ldots, x_{1,N_1}, x_{2,1}, \ldots, x_{2,N_2}, \ldots, x_{k,1}, \ldots, x_{k,N_k}, \ldots); \tag{1}
aSk+j=xk+1,j(1≤j≤Nk+1).(2)a_{S_k+j} = x_{k+1,j} \quad (1 \le j \le N_{k+1}). \tag{2}

If p∈Dp \in D and k≥pk \ge p then

∣∑j=Sk+1Sk+1ajp∣=∣∑j=1Nk+1xk+1,jp∣≥1;\left|\sum_{j=S_k+1}^{S_{k+1}} a_j^p\right| = \left|\sum_{j=1}^{N_{k+1}} x_{k+1,j}^p\right| \ge 1;
By Cauchy's convergence criterion it follows that ∑anp\sum a_n^p is divergent.

If p∈Cp \in C and Su<n≤Su+1S_u < n \le S_{u+1} with some u≥pu \ge p then

∣∑j=Sp+1nanp∣=∣∑k=p+1u−1∑j=1Nkxk,jp+∑j=1n−Su−1xu,jp∣=∣∑j=1n−Su−1xu,jp∣≤1u.\left|\sum_{j=S_p+1}^n a_n^p\right| = \left|\sum_{k=p+1}^{u-1}\sum_{j=1}^{N_k} x_{k,j}^p + \sum_{j=1}^{n-S_{u-1}} x_{u,j}^p\right| = \left|\sum_{j=1}^{n-S_{u-1}} x_{u,j}^p\right| \le \frac{1}{u}.

Then it follows that ∑n=Sp+1∞anp=0\displaystyle\sum_{n=S_p+1}^\infty a_n^p = 0, and thus ∑n=1∞anp=0\displaystyle\sum_{n=1}^\infty a_n^p = 0 is convergent.

Μεταγραφή από το επίσημο κείμενο λύσεων, σελίδες 2–3. Η μεταγραφή δεν έχει ελεγχθεί δεύτερη φορά· αν έχεις αμφιβολία, κοίτα το πρωτότυπο.

Θέμα αρχείου: η εκφώνηση είναι το επίσημο κείμενο (μεταγραμμένο, με το πρωτότυπο παραπάνω). Η ταξινόμηση ανά αντικείμενο και η δυσκολία εκτιμήθηκαν αυτόματα· υποδείξεις και λύση στα ελληνικά προστίθενται σταδιακά.

Πώς πήγε;