Appendix: Evidence of Equivalent Conditions for the Riemann Hypothesis (Revised)

By Ing. Robert Polák, Slovakia

Date: 9.10.2024


Proof 1: For \( n = 2 \times 3 \times 5 \times 7 \times \dots \times p_k \), it holds that \( \log(n) < p_k \)

Claim:

Let \( n \) be the product of the first \( k \) primes:

\( n = p_1 \times p_2 \times \dots \times p_k \),

where \( p_k \) is the \( k \)-th prime. Then, it holds that:

\( \log(n) < p_k \)

Proof:

  1. Expressing \( \log(n) \) as a sum:

    Using the logarithm property for products:

    \( \log(n) = \log(p_1) + \log(p_2) + \dots + \log(p_k) \)
  2. Definition of the Chebyshev function \( \theta(x) \):

    The Chebyshev function is defined as:

    \( \theta(x) = \sum_{\substack{p \leq x \\ p \text{ is prime}}} \log(p) \)

    For \( x = p_k \):

    \( \theta(p_k) = \sum_{i=1}^{k} \log(p_i) \)
  3. Application of Chebyshev's inequality:

    Chebyshev proved that for all \( x \geq 2 \) it holds:

    \( \theta(x) < x \)

    Therefore, for \( x = p_k \):

    \( \theta(p_k) < p_k \)
  4. Conclusion of the inequality:

    Thus:

    \( \boxed{\log(n) < p_k} \)

Proof 2: For Highly Composite Numbers \( n \), it holds that \( \log(n) > p_k \) for large \( n \)

Claim:

Let \( n \) be a highly composite number with the prime factorization:

\( n = p_1^{j_1} \times p_2^{j_2} \times \dots \times p_k^{j_k} \),

where \( p_i \) are primes in ascending order (\( p_1 = 2,\, p_2 = 3,\, \dots \)) and the exponents satisfy \( j_1 \geq j_2 \geq \dots \geq j_k \geq 1 \) with at least one \( j_i > 1 \). Then, it holds that:

\( \log(n) > p_k \)

Proof:

  1. Expressing \( \log(n) \):

    The natural logarithm of \( n \) is given by:

    \( \log(n) = \sum_{i=1}^{k} j_i \ln(p_i) \)
  2. Splitting the sum:

    We split the sum into two parts:

    \( \log(n) = \underbrace{\sum_{i=1}^{k} \ln(p_i)}_{\theta(p_k)} + \underbrace{\sum_{i=1}^{k} (j_i - 1)\ln(p_i)}_{\Delta} \)
  3. Estimate of \( \delta_k \):

    Let \( \delta_k = p_k - \theta(p_k) \). According to historical estimates, it holds that:

    \( |\theta(x) - x| \leq C\,\frac{x}{\ln x} \)

    For \( x = p_k \) therefore:

    \( |\delta_k| \leq C\,\frac{p_k}{\ln(p_k)} \)

    Let us choose the conservative estimate \( C = 0.5 \), which asymptotically gives:

    \( |\delta_k| \lesssim 0.5\, k \)
  4. Estimate of \( \Delta \):

    For those primes with \( j_i > 1 \), we define the sum:

    \( S_r = \sum_{\substack{p_i \text{ with } j_i>1}} \ln(p_i) \)

    Thus, it holds that:

    \( \Delta \ge (j_{\min}-1)\, S_r \)
  5. Condition for \( \log(n) > p_k \):

    In order to achieve \( \log(n) > p_k \), it is sufficient that:

    \( (j_{\min}-1)\, S_r > C\,\frac{p_k}{\ln(p_k)} \)

    We thus define the minimal required sum for the enhanced factors:

    \( S_{\min} = \frac{C\,p_k}{(j_{\min}-1)\,\ln(p_k)} \)

    If \( S_r > S_{\min} \), then we have \( \Delta > \delta_k \) and therefore:

    \( \log(n)=\theta(p_k)+\Delta > \theta(p_k)+\delta_k \ge p_k \)
  6. Example Illustrating the Proof:

    Consider \[ n = 2^5 \times 3^4 \times 5^3 \times 7^2 \times 11^2 \times 13 \times 17 \times 19, \] where:

    • Primes: \(2,\,3,\,5,\,7,\,11,\,13,\,17,\,19\)
    • Exponents: \( j_1=5,\; j_2=4,\; j_3=3,\; j_4=2,\; j_5=2,\; j_6=j_7=j_8=1\)
    • \( k\approx\frac{p_k}{\ln(p_k)}\approx\frac{19}{2.944}\approx6.45 \)
    • Enhanced primes: \(2,\,3,\,5,\,7,\,11\) (i.e. \( r=5 \) and \( j_{\min}=2 \))

    Calculation of \( \theta(p_k) \):

    \( \theta(p_k)=\ln2+\ln3+\ln5+\ln7+\ln11+\ln13+\ln17+\ln19\approx16.09 \)

    Calculation of \( \Delta \):

    \( \Delta=(5-1)\ln2+(4-1)\ln3+(3-1)\ln5+(2-1)\ln7+(2-1)\ln11\approx13.63 \)

    Calculation of \( \delta_k \):

    \( \delta_k=p_k-\theta(p_k)\approx19-16.09\approx2.91 \)

    For the enhanced primes we have: \[ S_r = \ln2+\ln3+\ln5+\ln7+\ln11 \approx 0.6931+1.0986+1.6094+1.9459+2.3979\approx7.745, \] and therefore \[ \Delta \ge (2-1)S_r \approx 7.745 > \delta_k \approx 2.91. \]

    The minimal required sum is \[ S_{\min} = \frac{0.5\times19}{(2-1)\times2.944}\approx\frac{9.5}{2.944}\approx3.225. \] Hence, \( S_r > S_{\min} \) and the inequality \( \log(n) > p_k \) holds.

  7. Conclusion:

    It is important to emphasize that the derivation of the minimal contribution for the enhanced prime factors—specifically the definition \( S_{\min} = \frac{C\,p_k}{(j_{\min}-1)\,\ln(p_k)} \)—assumes that for all such factors the exponent is at least \( j_{\min} = 2 \). However, in real highly composite numbers the distribution of exponents is often much more favorable, especially for the smaller primes.

    In other words, even though the theoretical estimate \( r_{\min} \) is based on the conservative assumption \( j_{\min} = 2 \), real highly composite numbers have exponents for the small primes that are often significantly higher. This increases the overall contribution \(\Delta\) beyond the boundary \(\delta_k\), thereby practically ensuring that the condition \(\Delta > \delta_k\) (i.e., \( \log(n)=\theta(p_k)+\Delta > p_k \)) is met not only in theory but also in practice.


Swap–Argument – A Rigorous Three-Step Proof (Supplement)

Let \( n = \prod_{i=1}^{k} p_i^{\,j_i} \) be a highly composite number and assume that there exists at least one prime \(p_r\) with exponent 2 (\(j_r=2\)). Suppose that this \(p_r\) is the candidate for a swap. Let \(p_1\) be the smallest prime with exponent \(j_1\) (in practice often \(j_1=2\)).

Assume that \( n \) is very large and, for instance, the first 20 larger bases (primes) have exponent 2. For illustration, let us choose specifically \(p_r = p_7 = 17\) with exponent 2. We will show that by decreasing the exponent of 17 (from 2 to 1) and increasing the exponent of \(p_1=2\) by a suitable \(\Delta\), we zlepšíme \(\sigma(n)/n\), while maintaining \(\tilde{n}\le n\).

We know that \(\sigma(n)/n\) is multiplicative, so we can decompose it into a product of functions:

\(\displaystyle \frac{\sigma(n)}{n} \;=\; \prod_{i=1}^{k} f(p_i,j_i), \quad \text{where} \quad f(p,j) \;=\; \frac{p^{\,j+1}-1}{\,p^j\,(p-1)\,}. \)

In the original configuration (before the swap), both \(f(p_1,j_1)\) and \(f(p_r,2)\) are part of \(\sigma(n)/n\).

We define the swap: by decreasing the exponent of \(p_r\) (i.e. 17) from 2 to 1, thereby removing the factor \(p_r^2\). Simultaneously, we increase the exponent of \(p_1=2\) by \(\Delta\). To ensure that \(\tilde{n}\le n\), we require:

\(\displaystyle 2^\Delta \;\le\; 17, \quad \Delta = \left\lfloor \frac{\ln(17)}{\ln(2)} \right\rfloor = 4. \)

Then, the new number is given by \(\tilde{n} = \frac{n}{\,17^2\,}\times 2^\Delta \,\le\, n\).

Before the swap, the contributions in the factorization are:

  • \(f(2,\,j_1)\) (for example, if \(j_1=2\), then \(f(2,2)=\tfrac{2^3-1}{2^2}=1.75\)),
  • \(f(17,2)\) (computed as \(\approx1.0623\)).

After the swap, the exponent of \(17\) is decreased to 1 and the exponent of \(2\) is increased to \((j_1 + \Delta)\). The new contributions become \(f(2,\,j_1+\Delta)\) and \(f(17,\,1)\approx1.0588\).

We define the ratio

\(\displaystyle R \;=\; \frac{\sigma(\tilde{n})/\tilde{n}}{\sigma(n)/n} \;=\; \frac{\,f\bigl(2,\,j_1+\Delta\bigr)\,\times\,f\bigl(17,\,1\bigr)\,} {\,f\bigl(2,j_1\bigr)\,\times\,f\bigl(17,\,2\bigr)\,}. \)

The swap is beneficial if \(R>1\). For \(j_1=2\) and \(\Delta=4\):

  • \(f(2,2)=\tfrac{2^3-1}{2^2}=\tfrac{7}{4}=1.75\),
  • \(f(17,2)\approx1.0623\),
  • \(f(2,6)=\tfrac{2^7-1}{2^6}=\tfrac{127}{64}\approx1.9844\),
  • \(f(17,1)=\tfrac{17^2-1}{17\times16}=\tfrac{288}{272}\approx1.0588\).

Therefore, \(\displaystyle R \;=\; \frac{1.9844\times1.0588}{\,1.75\times1.0623\,} \approx \frac{2.102}{1.859} \approx 1.13 > 1. \) Hence, the swap increased \(\sigma(n)/n\) by more than 13% while maintaining \(\tilde{n}\le n\).

With this three-step approach:

  1. We define \(f(p,j)\) and decompose \(\sigma(n)/n\) as a product,
  2. We choose \(\Delta = \lfloor \ln(p_r)/\ln(2)\rfloor\) such that \(\tilde{n}\le n\),
  3. We show that \(\dfrac{\,f(2,j_1+\Delta)\,}{f(2,j_1)} > \dfrac{\,f(17,2)\,}{f(17,1)}\), i.e. \(R>1\).

The swap argument then states that if in the factorization of a highly composite number any larger prime has an exponent of 2, it is always possible to transfer exponent units to smaller bases (typically \(2\)) and achieve a greater value of \(\sigma(n)/n\), while ensuring \(\tilde{n}\le n\). This refutes the so-called minimal configuration and proves that the optimal configuration must have significantly higher exponents for the smallest primes, which leads to \(\log(n) > p_k\).

Explicit Bridge Between Swap Argument and Highly Composite Numbers

As demonstrated in the swap argument, if for some number \( n \) in the r-minimal configuration it holds that \( \log(n) = p_k \), we could perform an exchange of exponents to obtain a number \( \tilde{n} \leq n \) with a larger value of \( \sigma(\tilde{n})/\tilde{n} \). This implies that the r-minimal configuration is not optimal for maximizing \( \sigma(n)/n \).

However, highly composite numbers are defined as those that maximize the number of divisors for their size, and thus also \( \sigma(n)/n \). If for a highly composite number \( n \) it were true that \( \log(n) \leq p_k \), it would mean that its exponent configuration resembles the r-minimal configuration, which, as shown, is not optimal. Consequently, there would exist a number \( \tilde{n} \leq n \) with a larger \( \sigma(\tilde{n})/\tilde{n} \), contradicting the fact that \( n \) is highly composite.

Therefore, for all highly composite numbers, it must hold that \( \log(n) > p_k \), as only then is their exponent configuration truly optimal and cannot be improved by any exchange of exponents.


References

  • Rosser, J. B. & Schoenfeld, L. (1975). "Sharper Bounds for the Chebyshev Functions θ(x) and ψ(x)". Mathematics of Computation, 29, 243–269.
  • Dusart, P. (1999). "The kth prime is greater than k (ln k + ln ln k − 1)". Mathematics of Computation.
  • Chebyshev, P. L. (1852). First works on functions now known as Chebyshev functions.
  • Rosser, J. B. (1938). "The nth Prime is Greater than n ln n". Proceedings of the London Mathematical Society.
  • Broadbent, S., Kadiri, H., Lumley, A. R., Ng, N., & Wilk, K. (2021). "Sharper bounds for the Chebyshev function θ(x)". Mathematics of Computation.

Dôkaz 1: Pre \( n = 2 \times 3 \times 5 \times 7 \times \dots \times p_k \) platí \( \log(n) < p_k \)

Tvrdenie:

Nech \( n \) je súčin prvých \( k \) prvočísel:

\( n = p_1 \times p_2 \times \dots \times p_k \),

kde \( p_k \) je \( k \)-té prvočíslo. Potom platí:

\( \log(n) < p_k \)

Dôkaz:

  1. Vyjadrenie \( \log(n) \) ako súčet:

    Využitím vlastnosti logaritmu pre súčin:

    \( \log(n) = \log(p_1) + \log(p_2) + \dots + \log(p_k) \)
  2. Definovanie Chebyshevovej funkcie \( \theta(x) \):

    Chebyshevova funkcia je definovaná ako:

    \( \theta(x) = \sum_{\substack{p \leq x \\ p \text{ je prvočíslo}}} \log(p) \)

    Pre \( x = p_k \):

    \( \theta(p_k) = \sum_{i=1}^{k} \log(p_i) \)
  3. Uplatnenie Chebyshevovej nerovnosti:

    Chebyshev dokázal, že pre všetky \( x \geq 2 \) platí:

    \( \theta(x) < x \)

    Preto pre \( x = p_k \):

    \( \theta(p_k) < p_k \)
  4. Záver nerovnosti:

    Teda:

    \( \boxed{\log(n) < p_k} \)

Dôkaz 2: Pre vysoko-zložené čísla \( n \) platí \( \log(n) > p_k \) pre veľké \( n \)

Tvrdenie:

Nech \( n \) je vysoko-zložené číslo s prvočíselným rozkladom:

\( n = p_1^{j_1} \times p_2^{j_2} \times \dots \times p_k^{j_k} \),

kde \( p_i \) sú prvočíselá v rastúcom poradí (\( p_1 = 2,\, p_2 = 3,\, \dots \)) a exponenty spĺňajú \( j_1 \geq j_2 \geq \dots \geq j_k \geq 1 \) s aspoň jedným \( j_i > 1 \). Potom platí:

\( \log(n) > p_k \)

Dôkaz:

  1. Vyjadrenie \( \log(n) \):

    Prirodzený logaritmus čísla \( n \) je:

    \( \log(n) = \sum_{i=1}^{k} j_i \ln(p_i) \)
  2. Rozdelenie súčtu:

    Rozdelíme súčet na dve časti:

    \( \log(n) = \underbrace{\sum_{i=1}^{k} \ln(p_i)}_{\theta(p_k)} + \underbrace{\sum_{i=1}^{k} (j_i - 1)\ln(p_i)}_{\Delta} \)
  3. Odhad \( \delta_k \):

    Nech \( \delta_k = p_k - \theta(p_k) \). Podľa historických odhadov platí:

    \( |\theta(x) - x| \leq C\,\frac{x}{\ln x} \)

    Pre \( x = p_k \) teda:

    \( |\delta_k| \leq C\,\frac{p_k}{\ln(p_k)} \)

    Zvoľme konzervatívny odhad \( C = 0.5 \), čo dáva asymptoticky:

    \( |\delta_k| \lesssim 0.5\, k \)
  4. Odhad \( \Delta \):

    Pre tie prvočísla, kde \( j_i > 1 \), definujeme sumu:

    \( S_r = \sum_{\substack{p_i \text{ s } j_i>1}} \ln(p_i) \)

    Takže platí:

    \( \Delta \ge (j_{\min}-1)\, S_r \)
  5. Podmienka pre \( \log(n) > p_k \):

    Aby sme dosiahli \( \log(n) > p_k \), postačí, aby bolo:

    \( (j_{\min}-1)\, S_r > C\,\frac{p_k}{\ln(p_k)} \)

    Definujeme teda minimálnu požadovanú hodnotu pre súčet vylepšených faktorov:

    \( S_{\min} = \frac{C\,p_k}{(j_{\min}-1)\,\ln(p_k)} \)

    Ak \( S_r > S_{\min} \), potom máme \( \Delta > \delta_k \) a teda:

    \( \log(n)=\theta(p_k)+\Delta > \theta(p_k)+\delta_k \ge p_k \)
  6. Príklad na ilustráciu dôkazu:

    Uvažujme \[ n = 2^5 \times 3^4 \times 5^3 \times 7^2 \times 11^2 \times 13 \times 17 \times 19 \] kde:

    • Prvočísla: \(2,\,3,\,5,\,7,\,11,\,13,\,17,\,19\)
    • Exponenty: \( j_1=5,\; j_2=4,\; j_3=3,\; j_4=2,\; j_5=2,\; j_6=j_7=j_8=1\)
    • \( k\approx\frac{p_k}{\ln(p_k)}\approx\frac{19}{2.944}\approx6.45 \)
    • Vylepšené prvočísla: \(2,\,3,\,5,\,7,\,11\) (t.j. \( r=5 \) a \( j_{\min}=2 \))

    Výpočet \( \theta(p_k) \):

    \( \theta(p_k)=\ln2+\ln3+\ln5+\ln7+\ln11+\ln13+\ln17+\ln19\approx16.09 \)

    Výpočet \( \Delta \):

    \( \Delta=(5-1)\ln2+(4-1)\ln3+(3-1)\ln5+(2-1)\ln7+(2-1)\ln11\approx13.63 \)

    Výpočet \( \delta_k \):

    \( \delta_k=p_k-\theta(p_k)\approx19-16.09\approx2.91 \)

    Pre vylepšené prvočísla máme: \[ S_r = \ln2+\ln3+\ln5+\ln7+\ln11 \approx 0.6931+1.0986+1.6094+1.9459+2.3979\approx7.745, \] a teda \[ \Delta \ge (2-1)S_r \approx 7.745 > \delta_k \approx 2.91. \]

    Minimálna požadovaná suma je \[ S_{\min} = \frac{0.5\times19}{(2-1)\times2.944}\approx\frac{9.5}{2.944}\approx3.225. \] Preto \( S_r > S_{\min} \) a nerovnosť \( \log(n) > p_k \) platí.

  7. Záver:

    Je dôležité zdôrazniť, že odvodzovanie minimálneho príspevku pre vylepšené prvočíselné faktory, konkrétne definícia \( S_{\min} = \frac{C\,p_k}{(j_{\min}-1)\,\ln(p_k)} \), vychádza z predpokladu, že pre všetky takéto faktory je exponent aspoň \( j_{\min} = 2 \). Avšak v reálnych vysoko-zložených číslach sa stretávame s oveľa lepším rozložením exponentov, najmä pre malé prvočísla v rozklade.

    Pre takéto čísla sú exponenty často omnoho väčšie než 2, čo znamená, že príspevok jednotlivých faktorov do sumy \( S_r = \sum_{\substack{p_i \text{ s } j_i>1}} \ln(p_i) \) je podstatne vyšší, než by vyplývalo iba z minimálnej hodnoty \((j_{\min}-1)\ln(p_i)\). Tento lepší rozklad zabezpečuje, že celková suma \( S_r \) dosahuje hodnoty, ktoré výrazne prekračujú minimálny odhad \( S_{\min} \).

    Inými slovami, aj keď teoretický odhad \( r_{\min} \) vychádza z konzervatívneho predpokladu \( j_{\min} = 2 \), reálne vysoko-zložené čísla majú exponenty pre malé prvočísla oveľa vyššie, čo zvyšuje celkový príspevok \(\Delta\) nad hranicu \(\delta_k\). Tento fakt prakticky garantuje, že podmienka \(\Delta > \delta_k\) (teda \( \log(n)=\theta(p_k)+\Delta > p_k \)) je splnená nielen teoreticky, ale aj v reálnych prípadoch.


Swap–argument – Rigorózny trojkrokový dôkaz (Supplement)

Nech je \( n = \prod_{i=1}^{k} p_i^{\,j_i} \) vysoko-zložené číslo a nech existuje aspoň jedno prvočíslo \(p_r\) s exponentom 2 (\(j_r=2\)). Predpokladajme, že práve toto \(p_r\) je kandidátom na swap. Nech \(p_1\) je najmenšie prvočíslo s exponentom \(j_1\) (v praxi často \(j_1=2\)).

Uvažujme, že \(n\) je veľmi veľké a napríklad prvých 20 väčších báz (prvočísel) má exponent 2. Pre ilustráciu vyberme konkrétne \(p_r = p_7 = 17\) s exponentom 2. Ukážeme, že znížením exponentu 17 (z 2 na 1) a zvýšením exponentu \(p_1=2\) o vhodné \(\Delta\), zlepšíme \(\sigma(n)/n\), pričom \(\tilde{n}\le n\).

Vieme, že \(\sigma(n)/n\) je multiplikatívna, takže ju môžeme rozložiť na súčin funkcií

\(\displaystyle \frac{\sigma(n)}{n} \;=\; \prod_{i=1}^{k} f(p_i,j_i), \quad \text{kde} \quad f(p,j) \;=\; \frac{p^{\,j+1}-1}{\,p^j\,(p-1)\,}. \)

V pôvodnej konfigurácii (pred swapom) sa teda do \(\sigma(n)/n\) zahŕňajú aj \(f(p_1,j_1)\) a \(f(p_r,2)\).

Definujeme swap: exponent \(p_r\) (teda 17) znížime z 2 na 1, čím odstránime faktor \(p_r^2\). Zároveň exponent \(p_1=2\) zvýšime o \(\Delta\). Aby sa zachovalo \(\tilde{n}\le n\), požadujeme:

\(\displaystyle 2^\Delta \;\le\; 17, \quad \Delta = \left\lfloor \frac{\ln(17)}{\ln(2)} \right\rfloor = 4. \)

Potom vzniká \(\tilde{n} = \frac{n}{\,17^2\,}\times 2^\Delta \,\le\, n\).

Pred swapom sa v rozklade nachádzajú príspevky:

  • \(f(2,\,j_1)\) (napr. ak \(j_1=2\), tak \(f(2,2)=\tfrac{2^3-1}{2^2}=1.75\)),
  • \(f(17,2)\) (vypočítame \(\approx1.0623\)).

Po swape je exponent \(17\) znížený na 1 a exponent \(2\) zvýšený na \((j_1 + \Delta)\). Nové príspevky sú \(f(2,\,j_1+\Delta)\) a \(f(17,\,1)\approx1.0588\).

Definujeme pomer

\(\displaystyle R \;=\; \frac{\sigma(\tilde{n})/\tilde{n}}{\sigma(n)/n} \;=\; \frac{\,f\bigl(2,\,j_1+\Delta\bigr)\,\times\,f\bigl(17,\,1\bigr)\,} {\,f\bigl(2,j_1\bigr)\,\times\,f\bigl(17,\,2\bigr)\,}. \)

Swap je prospešný, ak \(R>1\). Pre \(j_1=2\) a \(\Delta=4\):

  • \(f(2,2)=\tfrac{2^3-1}{2^2}=\tfrac{7}{4}=1.75\),
  • \(f(17,2)\approx1.0623\),
  • \(f(2,6)=\tfrac{2^7-1}{2^6}=\tfrac{127}{64}\approx1.9844\),
  • \(f(17,1)=\tfrac{17^2-1}{17\times16}=\tfrac{288}{272}\approx1.0588\).

Preto \(\displaystyle R \;=\; \frac{1.9844\times1.0588}{\,1.75\times1.0623\,} \approx \frac{2.102}{1.859} \approx 1.13 > 1. \) Swap teda zvýšil \(\sigma(n)/n\) o viac než 13% a pritom \(\tilde{n}\le n\).

Týmto trojkrokovým prístupom:

  1. Definujeme \(f(p,j)\) a rozložíme \(\sigma(n)/n\) na súčin,
  2. Volíme \(\Delta = \lfloor \ln(p_r)/\ln(2)\rfloor\) tak, aby \(\tilde{n}\le n\),
  3. Ukážeme, že \(\dfrac{\,f(2,j_1+\Delta)\,}{f(2,j_1)} > \dfrac{\,f(17,2)\,}{f(17,1)}\), teda \(R>1\).

Swap–argument následne hovorí, že ak by v rozklade vysoko-zloženého čísla akékoľvek väčšie prvočíslo malo exponent 2, vždy je možné presunúť exponentové jednotky k menším bázam (typicky \(2\)) a docieliť väčšiu hodnotu \(\sigma(n)/n\), pričom \(\tilde{n}\le n\). Tým sa vyvracia tzv. minimálna konfigurácia a dokazuje sa, že optimálna musí mať omnoho vyššie exponenty pri najmenších prvočíslach, čo vedie k \(\log(n) > p_k\).

Explicitné premostenie swap argumentu a vysoko-zložených čísel

Ako sme videli v swap argumente, ak by pre nejaké číslo \( n \) v r-minimálnej konfigurácii platilo \( \log(n) = p_k \), mohli by sme vykonať výmenu exponentov a získať číslo \( \tilde{n} \leq n \) s väčšou hodnotou \( \sigma(\tilde{n})/\tilde{n} \). To znamená, že r-minimálna konfigurácia nie je optimálna pre maximalizáciu \( \sigma(n)/n \).

Vysoko-zložené čísla sú však definované ako tie, ktoré maximalizujú počet deliteľov pre danú veľkosť, a teda aj \( \sigma(n)/n \). Ak by pre vysoko-zložené číslo \( n \) platilo \( \log(n) \leq p_k \), znamenalo by to, že jeho konfigurácia exponentov pripomína r-minimálnu konfiguráciu, ktorá, ako sme ukázali, nie je optimálna. Tým by existovalo číslo \( \tilde{n} \leq n \) s väčším \( \sigma(\tilde{n})/\tilde{n} \), čo je v rozpore s tým, že \( n \) je vysoko-zložené.

Z toho vyplýva, že pre všetky vysoko-zložené čísla musí platiť \( \log(n) > p_k \), pretože len tak je ich konfigurácia exponentov skutočne optimálna a nemôže byť zlepšená žiadnou výmenou exponentov.


Referencie

  • Rosser, J. B. & Schoenfeld, L. (1975). "Sharper Bounds for the Chebyshev Functions θ(x) and ψ(x)". Mathematics of Computation, 29, 243–269.
  • Dusart, P. (1999). "The kth prime is greater than k (ln k + ln ln k − 1)". Mathematics of Computation.
  • Chebyshev, P. L. (1852). Prvé práce o funkciách, ktoré dnes nesieme názov Chebyshevových funkcií.
  • Rosser, J. B. (1938). "The nth Prime is Greater than n ln n". Proceedings of the London Mathematical Society.
  • Broadbent, S., Kadiri, H., Lumley, A. R., Ng, N., & Wilk, K. (2021). "Sharper bounds for the Chebyshev function θ(x)". Mathematics of Computation.