Skip to main content

On Analysis of Recovering Short Generator Problems via Upper and Lower Bounds of Dirichlet L-Functions: Part 1

  • Chapter
  • First Online:
Book cover Mathematical Modelling for Next-Generation Cryptography

Part of the book series: Mathematics for Industry ((MFI,volume 29))

  • 1464 Accesses

Abstract

This article is a survey on upper and lower bounds of Dirichlet L-functions \(L(s,\chi )\) associated with Dirichlet characters \(\chi \) at \(s=1\). We give proofs of well-known upper and lower bounds of \(L(1, \chi )\) to let the reader know the difficulty of giving (lower) bounds of \(L(1,\chi )\). In the last part, we also review some explicit upper and lower bounds of Dirichlet L-functions which will be applied to security analysis of ideal lattice-based cryptography for cyclotomic fields explained in Part 2 (S. Okumura, On analysis of recovering short generator problems via upper and lower bounds of Dirichlet L-functions : Part 2 [20]).

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

Chapter
USD 29.95
Price excludes VAT (USA)
  • Available as PDF
  • Read on any device
  • Instant download
  • Own it forever
eBook
USD 89.00
Price excludes VAT (USA)
  • Available as EPUB and PDF
  • Read on any device
  • Instant download
  • Own it forever
Softcover Book
USD 119.99
Price excludes VAT (USA)
  • Compact, lightweight edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info
Hardcover Book
USD 169.99
Price excludes VAT (USA)
  • Durable hardcover edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Institutional subscriptions

Notes

  1. 1.

    RSG is an abbreviation for Recovering Short Generator.

References

  1. S. Chowla, A new proof of a theorem of Siegel. Ann. Math. 2(51), 120–122 (1950)

    Article  MATH  MathSciNet  Google Scholar 

  2. R. Cramer, L. Ducas, C. Peikert, O. Regev, Recovering short generators of principal ideals in cyclotomic rings, in Advances in Cryptology - EUROCRYPT 2016, Part II, vol. 9666 (LNCS, 2016), pp. 559–585

    Google Scholar 

  3. H. Davenport, Graduate Texts in Math, vol. 74, 3rd edn., Multiplicative Number Theory (Springer, New York, 2000)

    Google Scholar 

  4. S.S. Eddin, D.J. Platt, Explicit upper bounds for \(|L(1, \chi )|\) when \(\chi (3)=0\). Colloq. Math. 133(1), 23–34 (2013)

    Article  MATH  MathSciNet  Google Scholar 

  5. T. Estermann, On Dirichlet’s \(L\)-functions. J. Lond. Math. Soc. 23, 275–279 (1948)

    Article  MATH  MathSciNet  Google Scholar 

  6. D. Goldfeld, A simple proof of Siegel’s theorem. Proc. Nat. Acad. Sci. USA 71(4), 1055–1055 (1974)

    Article  MATH  MathSciNet  Google Scholar 

  7. H. Heilbronn, On the class number in imaginary quadratic fields. Quarterly J. Math. 5, 150–160 (1934)

    Article  MATH  Google Scholar 

  8. H. Iwaniec, E. Kowalski, American Mathematical Society Colloquium Publications, vol. 53, Analytic number theory (American Mathematical socity, Providence, 2004)

    Google Scholar 

  9. T. Kimura, S. Koyama, N. Kurokawa, Euler products beyond the boundary. Lett. Math. Phys. 104, 1–19 (2014)

    Article  MATH  MathSciNet  Google Scholar 

  10. E. Landau, Über Dirichletsche Reihen mit komplexen Charakteren. J. Reine Angew. Math. 157, 26–32 (1927)

    MATH  MathSciNet  Google Scholar 

  11. J.E. Littlewood, On the class number of the corpus \(P(\sqrt{-k})\). Proc. Lond. Math. Soc. 27, 358–372 (1928)

    Article  MATH  MathSciNet  Google Scholar 

  12. S. Louboutin, Majorations explicites de \(|L(1, \chi )|\) (quatrième partie). C. R. Acad. Sci. Paris 334, 625–628 (2002)

    Article  MATH  Google Scholar 

  13. S. Louboutin, Simple proofs of the Siegel–Tatuzawa and Brauer–Siegel theorems. Colloq. Math. 108, 277–283 (2007)

    Article  MATH  MathSciNet  Google Scholar 

  14. S. Louboutin, An explicit lower bound on moduli of Dirichlet \(L\)-functions at \(s=1\). J. Ramanujan Math Soc. 101–113 (2015)

    Google Scholar 

  15. T. Metsänkylä, Estimations for L-functions and the class numbers of certain imaginary cyclic fields. Ann. Univ. Turku. Ser. A I(140), 1–11 (1970)

    MATH  Google Scholar 

  16. G. Molteni, \(L\)-functions: Siegel-type theorems and structure theorems, Ph.D. thesis, University of Milan, Milan (1999)

    Google Scholar 

  17. H.L. Montgomery, R.C. Vaughan, Multiplicative Number Theory I (Cambridge University Press, Cambridge, 2006)

    Book  MATH  Google Scholar 

  18. W. Narkiewicz, Rational Number Theory in the 20th Century : From PNT to FLT, Springer monographs in mathematics (Springer, Berlin, 2012)

    Book  MATH  Google Scholar 

  19. J. Neukirch, Algebraic Number Theory, Grundlehren Math. Wiss. (Springer, Berlin, 1999)

    Book  MATH  Google Scholar 

  20. S. Okumura, On analysis of recovering short generator problems via upper and lower bounds of Dirichlet \(L\)-functions : Part 2, in this book

    Google Scholar 

  21. S. Okumura, M. Yasuda, T. Takagi, An improvement on the Recovering Short Generator Attack over Ideal Lattices and its Countermeasure, preprint

    Google Scholar 

  22. S. Okumura, S. Sugiyama, M. Yasuda, T. Takagi, Security Analysis of Cryptosystems Using Short Generators Over Ideal Lattices, preprint, https://eprint.iacr.org/2015/1004

  23. D. Ramakrishnan, R.J. Valenza, Fourier Analysis on Number Fields, vol. 186, Graduate texts in math (Springer, New York, 1999)

    MATH  Google Scholar 

  24. O. Ramaré, Approximate formulae for \(L(1, \chi )\). Acta Arith. 100, 245–266 (2001)

    Article  MATH  MathSciNet  Google Scholar 

  25. C.L. Siegel, Über die Classenzahl quadratischer Zahlkörper. Acta Arith. 1, 83–86 (1935)

    Article  MATH  Google Scholar 

  26. T. Tatuzawa, On a theorem of Siegel. Jpn. J. Math. 21, 163–178 (1951)

    Article  MATH  MathSciNet  Google Scholar 

  27. E.C. Titchmarsh, The Theory of Riemann Zeta-Function, 2nd edn. (Oxford science publications, Clarendon Press, Oxford, 1986). revised by H. Brown

    MATH  Google Scholar 

  28. L. Washington, Introduction to Cyclotomic Fields, vol. 83, 2nd edn., Graduate texts in math (Springer, New York, 1997)

    Book  MATH  Google Scholar 

Download references

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Shingo Sugiyama .

Editor information

Editors and Affiliations

Appendices

8 Appendix 1

We prepare several lemmas on complex analysis for Theorem 1.2.

Lemma 8.1

(The Schwartz lemma) Set \(D=\{z \in \mathbb {C} \ | \ |z|<1 \}\). Let \(f : D \rightarrow D\) be a holomorphic map such that \(f(0)=0\). Then, \(|f(z)|\le |z|\) for all \(z \in D\) and \(f'(0)\le 1\).

Moreover, suppose either that there exists \(z_0 \in D\) such that \(|f(z_0)|=|z_0|\), or that \(f'(0)= 1\) holds. Then we have \(f(z)=az\) for some \(a \in \partial D\).

Proof

Set \(g(z)= f(z)/z\) if \(z \in D-\{0\}\) and \(g(z)=f'(0)\) if \(z=0\). Then, g is holomorphic on D. For each \(r\in (0, 1)\) and the closed disc \(D_r = \{z \in \mathbb {C} \ | \ |z|\le r \}\), by the maximum modulus principle, there exists \(z_r \in \partial D_r\) such that \(\max _{z \in D_r}|g(z)| = |g(z_r)|=\frac{|f(z_r)|}{|z_r|} \le 1/r\). As we take the limit \(r\rightarrow 1-0\), we have the first assertion. As for the second assertion, we have only to verify that g is a constant, which is proved by the maximum modulus principle similarly.\(\square \)

Lemma 8.2

(The Borel–Carathéodory theorem) Let f(s) be a holomorphic function on \(|s|\le R\) with \(R>0\). For any \(r \in (0, R)\), we have

$$\max _{|s|\le r} |f(s)| \ (= \max _{|s|=r}|f(s)|) \le \frac{2r}{R-r}\max _{|s|\le R} \mathfrak {R}(f(s)) +\frac{R+r}{R-r}|f(0)|.$$

Proof

First consider the case \(f(0)=0\). Set \(A=\max _{|s|\le R}\mathfrak {R}(f(s))\). Then \(A \ge \mathfrak {R}(f(0))=0\). In what follows, we may assume \(A>0\); indeed, if \(A = 0\), the function \(\mathfrak {R}(f(x+iy))\) in (xy) with \(x+iy \in D_R\) is a harmonic function and attains a maximum at the interior point 0 of \(D_R\). By the maximum modulus principle, \(\mathfrak {R}(f(s))\) must be a constant \(\mathfrak {R}(f(0))=0\) on \(D_R\). Thus f(s) is identically equal to zero on \(D_R\).

Set \(H_a=\{z \in \mathbb {C} \ | \ \mathfrak {R}(z)\le a\}\) for \(a\in \mathbb {R}\). Let us define the functions g and h by \(g(z)=z/A-1\) and \(h(z)= \frac{z+1}{z-1}\), respectively. By applying the Schwartz lemma for the composite

$$D_R \xrightarrow {f} H_A \xrightarrow {g} H_0 \xrightarrow {h} D_R,$$

we have \(|h\circ g\circ f(s)| \le |s/R|\) for all \(s \in D_R\), that is,

$$\frac{|R \, f(s)|}{|f(s)-2A|} \le |s|,$$

by which we have \(R |f(s)| \le r|f(s)-2A| \le r|f(s)|+2Ar\) on \(D_r\). Hence we obtain \(|f(s)|\le \frac{2r}{R-r}A\), which proves the assertion for f such that \(f(0)=0\). For the case \(f(0)\ne 0\), by applying the estimate above to \(f(s)-f(0)\), we obtain

$$|f(s)| \le |f(s)-f(0)|+|f(0)| \le \frac{2r}{R-r}\max _{|s|\le R}\mathfrak {R}(f(s)-f(0)) + |f(0)|.$$

for \(s\in D_r\). This completes the proof.\(\square \)

Lemma 8.3

([27, p. 56]) Suppose \(r>0\) and let f(s) be a holomorphic function on \(|s-s_0|\le r\) with \(f(s_0)\ne 0\). For a constant \(M>0\), suppose that \(\left| \frac{f(s)}{f(s_0)}\right| <e^M\) on \(|s-s_0|\le r\). Then, there exists an absolute constant \(A>0\) such that

$$\left| \frac{f'(s)}{f(s)}-\sum _{\rho }\frac{1}{s-\rho }\right| <\frac{AM}{r},$$

where \(\rho \) runs over the zeros of f(s) such that \(|\rho -s_0|\le r/2\) (the set of zeros is regarded as a multi-set).

Furthermore, if we suppose \(f(s)\ne 0\) for all \(s\in \mathbb {C}\) such that \(|s-s_0|\le r\) and \(\mathfrak {R}(s_0)<\mathfrak {R}(s)\), then we have

$$-\mathfrak {R}\left( \frac{f'(s_0)}{f(s_0)}\right) <\frac{AM}{r}.$$

Proof

Set \(g(s)=f(s)\prod _{\rho }(s-\rho )^{-1}\) and \(h(s)=\log \frac{g(s)}{g(s_0)}\), where \(\rho \) runs over the zeros of f(s) such that \(|\rho -s_0|\le r/2\). Then, g(s) is holomorphic on \(|s-s_0|\le r\) and non-zero for \(|s-s_0| \le r/2\). For \(s \in \mathbb {C}\) such that \(|s-s_0|=r\), we have

$$\left| \frac{g(s)}{g(s_0)}\right| = \left| \frac{f(s)}{f(s_0)}\prod _{\rho }\frac{s_0-\rho }{s-\rho }\right| \le \left| \frac{f(s)}{f(s_0)}\right| <e^M$$

by \(|s_0-\rho | \le r/2 \le |s-\rho |\). This inequality as above is still valid for \(|s-s_0|\le r\) by the maximum modulus principle. From this, we have \(\mathfrak {R}(h(s))<M\) for \(|s-s_0|\le r\). Then, by the Borel–Carathéodory theorem, the estimate

$$|h(s)|< \frac{2\times 3r/8}{r/2-3r/8}M \le A_0 M$$

holds for \(|s-s_0| \le 3r/8\) with \(A_0>0\) being an absolute constant (\(A_0=6\) is sufficient). Hence, for \(|s-s_0| \le r/4\) we obtain

$$\left| \frac{f'(s)}{f(s)}-\sum _{\rho }\frac{1}{s-\rho }\right| = |h'(s)|= \left| \frac{1}{2\pi i}\oint _{|z-s|=r/4} \frac{h(z)}{(z-s)^2}dz \right| < \frac{4A_0M}{r}.$$

This completes the proof of the first assertion. The inequality as above gives us

$$-\mathfrak {R}\left( \frac{f'(s_0)}{f(s_0)}\right) < \frac{AM}{r} -\sum _{\rho }\mathfrak {R}\left( \frac{1}{s_0-\rho }\right) .$$

This completes the proof of the second assertion since \(\mathfrak {R}\left( \frac{1}{s_0-\rho }\right) \ge 0\) holds under the assumption.\(\square \)

Proposition 8.4

([27, p. 57] Let f(s) be a holomorphic function on \(|s-s_0|\le r\) such that \(f(s_0)\ne 0\)

$$\left| \frac{f(s)}{f(s_0)}\right|< e^M \quad \text {and} \quad \left| \frac{f'(s_0)}{f(s_0)}\right| < \frac{M}{r} \quad \text {on} \quad |s-s_0| \le r $$

for a constant \(M>0\). Furthermore, suppose that \(f(s)\ne 0\) for all \(s=\sigma +it\) such that \(\sigma \ge \sigma _0-2r'\) and \(|s-s_0|\le r\), where we set \(\sigma _0 = \mathfrak {R}(s_0)\) and \(r'\) is some real number such that \(0<r'<r/4\). Then, there exists an absolute constant \(A'>0\) such that

$$\left| \frac{f'(s)}{f(s)}\right| < \frac{A'M}{r} \qquad |s-s_0|\le r'.$$

Proof

As in Lemma 8.3, we have \(-\mathfrak {R}\left( \frac{f'(s)}{f(s)}\right) < \frac{AM}{r}\) for all \(s=\sigma +it\) such that \(|s-s_0|\le r/4\) and \(\sigma \ge \sigma _0 -2r'\). Therefore, the Borel–Carathéodory theorem yields the inequalities

$$\max _{|s-s_0|\le r'}\left| -\frac{f'(s)}{f(s)}\right| \le \frac{2r'}{2r'-r'}\max _{|s-s_0|\le 2r'}\mathfrak {R}\left( \frac{-f'(s)}{f(s)}\right) + \frac{2r'+r'}{2r'-r'} \left| \frac{-f'(s_0)}{f(s_0)} \right| < \frac{(2A+3)M}{r}.$$

Thus we are done.\(\square \)

9 Appendix 2

Let \(\chi \) be a primitive Dirichlet character modulo q. With the aid of a functional equation of \(L(s,\chi )\) in Proposition 3.1 and the Hadamard factorization theorem of entire functions of finite order (cf. [3, Sects. 11 and 12] or [17, Sect. 10.2]), and by taking the logarithmic derivative, there exists a constant \(B(\chi ) \in \mathbb {C}\) such that

$$\frac{L'(s,\chi )}{L(s,\chi )} = -\frac{a_\chi }{s-1}-\frac{1}{2}\log \frac{q}{\pi } -\frac{1}{2}\frac{\varGamma '(s/2+\delta _\chi /2)}{\varGamma (s/2+\delta _\chi /2)} +B(\chi ) + \sum _{\rho }\left( \frac{1}{s-\rho }-\frac{1}{\rho }\right) ,$$

where \(\rho \) runs over all zeros of \(L(s,\chi )\) in the region \(0<\mathfrak {R}(s)<1\), and we put \(a_\chi =1\) if \(\chi \) is trivial and \(a_\chi =0\) otherwise. Furthermore, taking the real part yields \(-\mathfrak {R}\left( \frac{L'(s,\chi )}{L(s,\chi )}\right) = \frac{1}{2}\log \frac{q}{\pi } +\frac{1}{2}\mathfrak {R}\left( \frac{\varGamma '(s/2+\delta _\chi /2)}{\varGamma (s/2+\delta _\chi /2)}\right) -\mathfrak {R}(B(\chi )) - \mathfrak {R}\sum _{\rho }\left( \frac{1}{s-\rho }-\frac{1}{\rho }\right) .\) With the aid of \(\mathfrak {R}(B(\chi )) = -\sum _{\rho }\mathfrak {R}(\frac{1}{\rho })\) (cf. [3, p. 83, (18)]) and \(\mathfrak {R}\left( \frac{\varGamma '(s/2+\delta _\chi /2)}{\varGamma (s/2+\delta _\chi /2)}\right) =\mathscr {O}(\log (3+|t|))\) from Stirling’s formula, we have the following.

Lemma 9.1

There exists an absolute constant \(C>0\) such that

$$\begin{aligned} -\mathfrak {R}\left( \frac{L'(s,\chi )}{L(s,\chi )}\right) < \mathfrak {R}\left( \frac{a_\chi }{s-1}\right) + C \log \{q(3+|t|)\} -\sum _{\rho }\mathfrak {R}\left( \frac{1}{s-\rho }\right) \quad \mathfrak {R}(s)>1. \end{aligned}$$

This inequality is often used in the proof of Proposition 6.1 on zero-free regions. Here we notice that the inequality in Proposition 8.4 holds on a closed disc but Lemma 9.1 holds on a half plane.

9 Appendix 3

For application to security analysis of ideal lattice-based cryptography for cyclotomic fields, we review explicit upper and lower bounds of \(L(1,\chi )\) (cf. [20, 22]). Such bounds of Dirichlet L-functions are related to the size of a dual basis of a log-unit lattice for a cyclotomic field, and moreover the size of such a dual basis is concerned with the RSG attackFootnote 1 as in [2].

Let \(\delta _{a,b}\) denote the Kronecker delta.

Theorem 10.1

([24, Corollaries 1 and 3] (see also [12, Corollary 1.2]) For any non-trivial primitive Dirichlet character modulo q with \(\chi (-1)=(-1)^\varepsilon \) for \(\varepsilon \in \{0,1\}\), we have

$$| L(1,\chi ) |\le \frac{1}{2}\log q+\delta _{\varepsilon , 1} \left( \frac{5}{2}-\log 6\right) .$$

Furthermore, if 2|q, then we have

$$| L(1,\chi ) | \le \frac{1}{4}\log q +\delta _{\varepsilon , 0}\left( \frac{1}{2}\log 2\right) + \delta _{\varepsilon ,1}\left( \frac{5}{4}-\frac{1}{2}\log 3\right) .$$

Theorem 10.2

([4, Theorem 1.1]) For any primitive Dirichlet character modulo q such that 3|q and \(\chi (-1)=(-1)^\varepsilon \) for \(\varepsilon \in \{0,1\}\), we have

$$ | L(1,\chi ) |\le \frac{1}{3}\log q + \delta _{\varepsilon ,0} \, 0.368296+\delta _{\varepsilon , 1} \, 0.838374. $$

By combining Theorems 10.1 and 10.2, we have the following estimates.

Corollary 10.1

Let \(\chi \) be a non-trivial even primitive Dirichlet character modulo \(q=p^k\), where p is prime and \(k \in \mathbb {N}\). Let \(k_\chi \) be the integer such that \(f_\chi =p^{k_\chi }\). Then, we have

$$\displaystyle |L(1,\chi )|\le {\left\{ \begin{array}{ll} \frac{1}{4}(\log f_\chi + 2\log 2) &{} (p=2),\\ \delta _{k_\chi ,2}\left( \frac{1}{2}\log f_\chi \right) +(\sum _{m=3}^{\infty }\delta _{k_\chi ,m})\,\frac{1}{3}(\log f_\chi +1.104888) &{} (p=3),\\ \frac{1}{2}\log f_\chi &{} (p\ge 5). \end{array}\right. }$$

Here we note \(k_\chi \ge 2 \) if \(p=3\) by the fact that there exists no even non-trivial Dirichlet characters modulo 3.

Theorem 10.3

([14, Corollary 2]) For any non-quadratic primitive Dirichlet character modulo \(q>1\), we have

$$|L(1, \chi )|\ge \frac{1}{10 \log (q/\pi )}.$$

Rights and permissions

Reprints and permissions

Copyright information

© 2018 Springer Nature Singapore Pte Ltd.

About this chapter

Cite this chapter

Sugiyama, S. (2018). On Analysis of Recovering Short Generator Problems via Upper and Lower Bounds of Dirichlet L-Functions: Part 1. In: Takagi, T., Wakayama, M., Tanaka, K., Kunihiro, N., Kimoto, K., Duong, D. (eds) Mathematical Modelling for Next-Generation Cryptography. Mathematics for Industry, vol 29. Springer, Singapore. https://doi.org/10.1007/978-981-10-5065-7_14

Download citation

  • DOI: https://doi.org/10.1007/978-981-10-5065-7_14

  • Published:

  • Publisher Name: Springer, Singapore

  • Print ISBN: 978-981-10-5064-0

  • Online ISBN: 978-981-10-5065-7

  • eBook Packages: EngineeringEngineering (R0)

Publish with us

Policies and ethics