Fact-checked by Grok 2 weeks ago

Dirichlet's approximation theorem

Dirichlet's approximation theorem is a foundational result in the field of Diophantine approximation, asserting that for any real number \alpha and any integer Q \geq 1, there exist integers p and q with $1 \leq q \leq Q such that |\alpha - p/q| \leq 1/(q(Q+1)).^[1] When \alpha is irrational, the theorem implies the existence of infinitely many coprime integers p and q > 0 satisfying |\alpha - p/q| < 1/q^2.^[2] This bound demonstrates that irrational numbers can be approximated by rationals to a precision inversely proportional to the square of the denominator, highlighting the density of rationals in the reals. Proved by the German mathematician Peter Gustav Lejeune Dirichlet in 1842, the theorem relies on a simple yet elegant application of the pigeonhole principle (also known as Dirichlet's box principle).^[3] The proof examines the Q+1 fractional parts \{k\alpha\} for k = 0, 1, \dots, Q, which lie in the interval [0, 1), and divides this interval into Q subintervals of length $1/Q; by the pigeonhole principle, at least two fractional parts fall into the same subinterval, yielding integers p and q (with q \leq Q) such that |q\alpha - p| < 1/Q.^[2] For the infinite case with irrational \alpha, repeated application with increasing Q ensures infinitely many distinct approximations, as any finite set would contradict the theorem for larger Q.^[1] The theorem serves as a cornerstone for more advanced results in number theory, including continued fraction expansions, which construct explicit sequences of optimal approximations, and extensions like Hurwitz's theorem, which improves the constant $1/q^2 to $1/(\sqrt{5} q^2) for quadratic irrationals.^[2] It also underpins studies of transcendental numbers, such as Liouville's theorem, which bounds approximations for algebraic numbers and characterizes certain transcendentals via exceptionally good rational approximations.^[4] Beyond pure mathematics, the result has implications in dynamical systems, ergodic theory, and computational number theory for analyzing the distribution of integer points near linear forms.^[5]

Statement

Basic version

Dirichlet's approximation theorem asserts that for any real number \alpha and any integer N \geq 1, there exist integers p and q with $1 \leq q \leq N such that |q \alpha - p| < 1/N.^[2] This finite approximation result guarantees the existence of rational approximations to \alpha whose quality improves as N increases. An equivalent formulation applies specifically to irrational numbers: if \alpha is irrational, then there are infinitely many rational numbers p/q (with p, q integers, q > 0, and \gcd(p, q) = 1) such that |\alpha - p/q| < 1/q^2.^[6] This follows from the basic statement by dividing the inequality |q \alpha - p| < 1/N by q, yielding |\alpha - p/q| < 1/(q N) \leq 1/q^2 since q \leq N. To obtain the infinitude of such approximations from the finite case, apply the theorem repeatedly with successively larger values of N. Suppose only finitely many rationals p/q satisfy |\alpha - p/q| < 1/q^2; let Q exceed all such denominators q. The theorem then produces a new pair p', q' with $1 \leq q' \leq Q and |q' \alpha - p'| < 1/Q, yielding |\alpha - p'/q'| < 1/(q' Q) \leq 1/(q')^2. Since \alpha is irrational, the fractional parts are distinct, ensuring the strict inequality < 1/Q from the pigeonhole principle, which implies < 1/(q')^2 overall and contradicts the finiteness assumption.^[1] The theorem implies that the irrationality exponent (or measure) \mu(\alpha) of any irrational \alpha satisfies \mu(\alpha) \geq 2, where \mu(\alpha) is the supremum of all \mu > 0 such that |\alpha - p/q| < 1/q^\mu holds for infinitely many rationals p/q.^[7] This lower bound establishes that every irrational number admits rational approximations at least as good as order $1/q^2 infinitely often.

Simultaneous version

The simultaneous version of Dirichlet's approximation theorem provides a multidimensional generalization, enabling the joint rational approximation of several real numbers by a common denominator. Specifically, for any real numbers \alpha_1, \dots, \alpha_d and any integer N \geq 1, there exist integers p_1, \dots, p_d and q with $1 \leq q \leq N such that

|q \alpha_i - p_i| \leq N^{-1/d}

for each i = 1, \dots, d.^[8] This bound is uniform across all coordinates and holds for arbitrary real \boldsymbol{\alpha} = (\alpha_1, \dots, \alpha_d), with the case d=1 reducing to the basic version of the theorem.^[9] Geometrically, the theorem relates to the distribution of points in the d-dimensional torus [0,1)^d, identified with \mathbb{R}^d / \mathbb{Z}^d. The multiples q \boldsymbol{\alpha} \mod 1 for q = 1, \dots, N lie in this compact space, and the result guarantees that one such point (or a suitable difference) falls into a small box of side length approximately N^{-1/d}, ensuring proximity to the integer lattice in all directions simultaneously.^[10] This formulation has key implications for simultaneous Diophantine approximation, assuring the existence of "good" joint rational approximations for any system of real numbers, including irrationals. For irrational \boldsymbol{\alpha} (in the sense that $1, \alpha_1, \dots, \alpha_d are linearly independent over \mathbb{Q}), iterating over increasing N yields infinitely many such approximations, with the exponent $1/d being optimal in the worst case.^[8]

Proofs

Pigeonhole principle proof

The proof of Dirichlet's approximation theorem relies on an elementary application of the pigeonhole principle to the fractional parts of multiples of a real number \alpha. Consider any real number \alpha and any positive integer N. Form the N+1 fractional parts \{k\alpha\} for k = 0, 1, \dots, N, where \{x\} = x - \lfloor x \rfloor denotes the fractional part of x, so each \{k\alpha\} lies in the interval

[0, 1)&#36;. Divide this interval into

Nsubintervals of equal length1/N: [0, 1/N), [1/N, 2/N), \dots, [(N-1)/N, 1)$.^[11] Since there are N+1 fractional parts and only N subintervals, the pigeonhole principle guarantees that at least one subinterval contains two distinct points, say \{j\alpha\} and \{i\alpha\} with $0 \leq i < j \leq N. The difference between these fractional parts satisfies |\{j\alpha\} - \{i\alpha\}| < 1/N. This difference equals the fractional part of (j - i)\alpha up to an integer adjustment, so there exists an integer m such that |(j - i)\alpha - m| < 1/N. Setting q = j - i (so $1 \leq q \leq N) and p = m, we obtain |q\alpha - p| < 1/N, or equivalently, \left|\alpha - \frac{p}{q}\right| < \frac{1}{qN}. Note that q \neq 0 since i < j.^[1]^[11] This establishes the basic version of the theorem for any real \alpha and N \geq 1. For irrational \alpha, the result implies infinitely many such approximations. Suppose there are only finitely many pairs (p, q) satisfying \left|\alpha - \frac{p}{q}\right| < \frac{1}{q^2}; let Q be larger than all such q. Applying the theorem with this larger Q yields a new q \leq Q and p with \left|\alpha - \frac{p}{q}\right| < \frac{1}{qQ} \leq \frac{1}{q^2}, contradicting the finiteness assumption. Iterating over increasing values of N (e.g., powers of 2) produces infinitely many distinct rationals p/q with \left|\alpha - \frac{p}{q}\right| < \frac{1}{q^2}.^[1]^[12]

Minkowski's theorem proof

Minkowski's convex body theorem, a fundamental result in the geometry of numbers, asserts that if S \subset \mathbb{R}^n is a compact convex set symmetric about the origin with \mathrm{Vol}(S) > 2^n \det(\Lambda), where \Lambda is a lattice in \mathbb{R}^n, then S contains a nonzero lattice point from \Lambda.^[13] This theorem, originally proved by Hermann Minkowski in 1896, provides a geometric framework for establishing the existence of good rational approximations to real numbers.^[14] To apply this to the basic version of Dirichlet's approximation theorem, consider a real number \alpha and a positive integer N. Define the set S = \{(x, y) \in \mathbb{R}^2 : |y| < N + 1, \, |\alpha y - x| < 1/N \}. This set is a bounded open parallelogram, convex and symmetric about the origin, with area $4(N + 1)/N > 4 = 2^2 \det(\mathbb{Z}^2).^[15] By Minkowski's theorem, S contains a nonzero point (p, q) \in \mathbb{Z}^2. Since |q| < N + 1 and q is an integer, it follows that $1 \leq |q| \leq N. Moreover, |\alpha q - p| < 1/N, so |\alpha - p/q| < 1/(q N). Without loss of generality, take q > 0 by symmetry.^[15] The same approach extends naturally to the simultaneous version in d dimensions. For real numbers \alpha_1, \dots, \alpha_d and positive integer N, consider the linear forms L_0(x_0, x_1, \dots, x_d) = x_0 and L_i(x_0, x_1, \dots, x_d) = \alpha_i x_0 - x_i for i = 1, \dots, d, with coefficient matrix B satisfying |\det B| = 1. Define the set S = \{ \mathbf{v} \in \mathbb{R}^{d+1} : |L_0(\mathbf{v})| < N + 1, \, |L_i(\mathbf{v})| < N^{-1/d} \ \forall i = 1, \dots, d \}. The volume of S is $2^{d+1} (N + 1) > 2^{d+1} = 2^{d+1} \det(\mathbb{Z}^{d+1}).^[16] Thus, Minkowski's theorem guarantees a nonzero \mathbf{v} = (q, p_1, \dots, p_d) \in \mathbb{Z}^{d+1} \cap S, yielding $1 \leq |q| \leq N and |q \alpha_i - p_i| < N^{-1/d} for each i, or equivalently, |\alpha_i - p_i / q| < 1/(q N^{1/d}).^[16] This geometric method offers a unified treatment of both versions, leveraging the analytic properties of convex bodies to handle multidimensional approximations seamlessly, in contrast to the pigeonhole principle, which provides a more elementary proof for the one-dimensional case.^[14]

Historical context

Dirichlet's original work

Peter Gustav Lejeune Dirichlet introduced his approximation theorem in the 1842 paper titled Verallgemeinerung eines Satzes aus der Lehre von den Kettenbrüchen nebst einigen Anwendungen auf die Theorie der Zahlen, published in the Sitzungsberichte der Preußischen Akademie der Wissenschaften zu Berlin (volume for 1842, pages 633–638). In this seminal work, Dirichlet employed the pigeonhole principle to establish the basic version of the theorem, demonstrating that for any real number \alpha and positive integer Q, there exist integers p and q with $1 \leq q \leq Q such that |q \alpha - p| < 1/Q.^[3] This result highlights approximations where the error is bounded better than $1/q, specifically yielding |\alpha - p/q| < 1/q^2 upon normalization.^[5] Dirichlet's motivation stemmed from his extensive research in Diophantine equations and the theory of continued fractions, where he sought to generalize results on rational approximations to broader number-theoretic contexts.^[17] The paper extended a specific theorem from continued fraction theory, applying it to provide rigorous bounds on how closely irrationals can be approximated by rationals, which connected directly to problems in transcendental number theory by quantifying the "irrationality" of real numbers through their approximability.^[18] This approach marked a shift toward using geometric principles like the pigeonhole method to solve analytic and algebraic issues in number theory. The 1842 proof established the $1/q^2 bound as a cornerstone of Diophantine approximation, proving its sharpness for quadratic irrationals while opening avenues for studying the distribution of approximations.^[19] Its influence extended to metric number theory, where subsequent works, such as those by Khintchine, built upon Dirichlet's framework to analyze the measure-theoretic properties of well-approximable numbers.^[20]

Pre-Dirichlet developments

The foundations of Diophantine approximation were laid in the 18th century through the study of continued fractions, beginning with Leonhard Euler's seminal work. In his 1737 dissertation De fractionibus continuis, Euler introduced a systematic theory of continued fractions, demonstrating their utility in generating rational approximations to irrational numbers. He observed patterns in the continued fraction expansions of quadratic irrationals, such as periodicity, leading to recurrent patterns that yield particularly effective approximations satisfying |\alpha - p/q| < 1/q^2, arising from the properties of continued fraction convergents.^[21] Building on Euler's insights, Joseph-Louis Lagrange advanced the theory in the late 18th century, particularly in his 1770 memoir Additions au mémoire sur la résolution des équations numériques. Lagrange proved that every irrational number possesses an infinite continued fraction expansion, ensuring infinitely many rational approximations p/q such that |\alpha - p/q| < 1/q^2. Unlike later uniform bounds, Lagrange's result emphasized the role of continued fractions in providing these approximations, focusing instead on the iterative nature of the expansions to guarantee their infinitude. He also established the converse: purely periodic continued fractions correspond precisely to quadratic irrationals, solidifying the connection between approximation quality and algebraic structure. For quadratic irrationals, the bounded partial quotients allow for improved constants in the approximation bound.^[22]^[23] Carl Friedrich Gauss contributed indirectly through his 1801 Disquisitiones Arithmeticae, where he developed the theory of binary quadratic forms and their reduction. This work explored Diophantine equations and the representation of integers, revealing connections between quadratic forms and rational approximations via lattice reduction techniques. Although not formulating a general approximation theorem, Gauss's insights into the geometry of quadratic forms provided early tools for quantifying how well irrationals could be approximated by rationals in specific algebraic settings.^[24] These 18th-century advancements in continued fractions and quadratic forms highlighted the potential for systematic rational approximations to irrationals, setting the stage for a universal bound applicable to all real numbers beyond quadratic cases.

Applications

In Diophantine approximation

Dirichlet's approximation theorem provides the foundational result in Diophantine approximation, asserting that for any irrational number \alpha, there exist infinitely many rational numbers p/q (with q > 0) such that |\alpha - p/q| < 1/q^2. This bound, known as the Dirichlet constant, establishes the baseline quality of rational approximations to irrational numbers, guaranteeing that every irrational can be approximated arbitrarily well relative to the denominator's square.^[2] In metric number theory, the theorem connects to the typical behavior of real numbers under Lebesgue measure. Khintchine's theorem refines this by showing that almost every real number \alpha (in the sense of Lebesgue measure) admits infinitely many rational approximations p/q satisfying |\alpha - p/q| < c / q^2 for every constant c > 0.^[25] This implies that the Dirichlet bound can be strengthened by any constant factor for the vast majority of irrationals. A notable exception consists of badly approximable numbers, defined as those irrationals \alpha for which there exists a constant c > 0 such that |\alpha - p/q| > c / q^2 holds for all integers p, q with q > 0. These numbers resist approximations better than a fixed multiple of the Dirichlet bound and include all quadratic irrationals, such as the golden ratio \phi = (1 + \sqrt{5})/2, whose continued fraction partial quotients are bounded.^[26] While this set has Lebesgue measure zero, it has full Hausdorff dimension 1.^[27] The theorem also informs the geometric structure of exceptional sets in approximation theory. Specifically, the set of real numbers \alpha that admit infinitely many approximations satisfying |\alpha - p/q| < 1/q^\tau for some \tau > 2 has Hausdorff dimension $2/\tau < 1, as established by Jarník's theorem; such sets capture numbers approximable far better than the Dirichlet baseline but occupy a fractal subset of the reals with reduced dimensionality.^[28]

In cryptography

Dirichlet's approximation theorem plays a crucial role in cryptanalysis of the RSA cryptosystem by guaranteeing the existence of rational approximations that enable recovery of the private key when it is sufficiently small. In Wiener's attack, the theorem ensures that if the private exponent d satisfies d < N^{1/4}/3, where N is the modulus, there exist integers k and d such that \left| \frac{e}{N} - \frac{k}{d} \right| < \frac{1}{2d^2}, approximating e/\phi(N) closely enough to identify k/d as a convergent in the continued fraction expansion of e/N.^[29] This approximation stems from the relation ed = 1 + k \phi(N), which rearranges to \frac{k}{d} \approx \frac{e}{\phi(N)}, and Dirichlet's theorem provides the bound to ensure such fractions exist and can be found efficiently by computing the continued fraction convergents of e/N (at most \log_2 N candidates).^[29] Once a suitable convergent k/d is obtained, solving ed - kN \approx 1 yields \phi(N), allowing factorization of N via the quadratic formula. The attack's efficiency relies on the theorem's guarantee of good approximations, limiting the search to a polynomial-time number of convergents, and Legendre's theorem to verify which convergent corresponds to the correct k/d by checking if |ed - kN| is small.^[29] This method succeeds because small d makes the approximation error \left| \frac{e d - k N}{d N} \right| < \frac{3}{\sqrt{N} d} sufficiently low, exploiting the bound |\phi(N) - N| < 3\sqrt{N}.^[29] Boneh and Durfee extended Wiener's approach to handle larger private exponents up to d < N^{0.292} by combining continued fraction approximations with Coppersmith's lattice-based method for finding small roots of univariate polynomials. Here, Dirichlet's theorem again provides the initial good approximation to seed the lattice construction, where the polynomial f(x) = e x - 1 + k \phi(N) (with \phi(N) approximated) is analyzed modulo N, allowing recovery of roots corresponding to small d - x.^[29] This improvement maintains the reliance on Dirichlet for the approximation guarantee while using LLL lattice reduction to push the boundary beyond Wiener's limit. These attacks fail for balanced keys where d \approx N/2, as the approximation error exceeds $1/d^2, preventing sufficiently accurate convergents from revealing \phi(N) or enabling small-root finding.^[29]

Legendre's theorem

Legendre's theorem provides a characterization of the best rational approximations to an irrational number \alpha in terms of its continued fraction expansion. Specifically, if p/q is a rational number in lowest terms with q > 0 such that \left| \alpha - \frac{p}{q} \right| < \frac{1}{2q^2}, then p/q must be one of the convergents in the continued fraction expansion of \alpha.^[30] This result, originally proved by Adrien-Marie Legendre in 1830, identifies the precise rationals that achieve the strongest approximations under this bound, linking Diophantine approximation directly to the theory of continued fractions.^[30] The proof relies on the properties of continued fraction convergents and intermediate fractions, known as semi-convergents. For any rational p/q not equal to a convergent, the distance to \alpha satisfies \left| \alpha - \frac{p}{q} \right| > \frac{1}{2q^2}, as derived from the recurrence relations defining the convergents p_n/q_n and the mediant construction between consecutive convergents. By considering cases where p/q lies between two consecutive convergents and analyzing the minimal distance via the continued fraction algorithm, the inequality forces p/q to coincide with a convergent.^[30] This theorem complements Dirichlet's approximation theorem by offering a constructive method to identify the infinitely many rationals p/q satisfying \left| \alpha - \frac{p}{q} \right| < \frac{1}{q^2}, since every continued fraction convergent achieves this stronger bound: \left| \alpha - \frac{p_n}{q_n} \right| < \frac{1}{q_n^2}.^[31] The convergents, generated sequentially, thus realize the approximations guaranteed by Dirichlet in an explicit manner. In practice, Legendre's theorem underpins algorithms for computing optimal rational approximations, such as those based on the Euclidean algorithm, which iteratively produces the continued fraction expansion and its convergents to approximate \alpha.^[32] This connection enables efficient numerical methods for Diophantine problems requiring high-precision rational representations.

Hurwitz's theorem

In 1891, Adolf Hurwitz sharpened Dirichlet's approximation theorem by establishing a stronger bound on the quality of rational approximations to irrational numbers. Specifically, for any irrational number \alpha, there exist infinitely many integers p and q > 0 (with \gcd(p, q) = 1) satisfying

\left| \alpha - \frac{p}{q} \right| < \frac{1}{\sqrt{5} q^2}.

This improves upon Dirichlet's constant of 1 by a factor of \sqrt{5}. Furthermore, \sqrt{5} is the optimal constant: if it is replaced by any larger value c > \sqrt{5}, then there exist irrational numbers \alpha (notably, equivalents of the golden ratio) for which the inequality holds for only finitely many p/q.^[33] The proof of Hurwitz's theorem leverages the continued fraction expansion of \alpha = [a_0; a_1, a_2, \dots], where the convergents p_n/q_n yield the best rational approximations. For these convergents, the approximation error satisfies | \alpha - p_n/q_n | < 1/(a_{n+1} q_n^2). If the partial quotients a_n are unbounded—which holds for all irrationals except those with bounded quotients—then infinitely many a_{n+1} \geq 2, ensuring the error is less than $1/(\sqrt{5} q_n^2) for infinitely many convergents. The constant \sqrt{5} arises from optimizing over the minimal possible large quotients, with the boundary case tied to quotients of 1.^[34] The golden ratio \phi = (1 + \sqrt{5})/2, with continued fraction [1; \overline{1}] consisting entirely of partial quotients equal to 1, exemplifies the sharpness of the bound. Its convergents are ratios of consecutive Fibonacci numbers F_{n+1}/F_n, and these achieve approximations on the order of $1/(\sqrt{5} q^2), but no better universally. Equivalents of \phi, such as \phi + k for integer k, share this property and demonstrate that no single constant larger than \sqrt{5} works for all irrationals.^[33] Hurwitz's theorem has profound implications for classifying badly approximable numbers, defined as irrationals \alpha for which there exists some c(\alpha) > 0 such that | \alpha - p/q | > c(\alpha)/q^2 for all integers p, q > 0. These are precisely the numbers whose continued fraction partial quotients are bounded above, with the golden ratio equivalents attaining the infimal c(\alpha) = 1/\sqrt{5}. The theorem thus delineates the optimal universal constant for quadratic Diophantine approximations, influencing subsequent developments in metric number theory and the study of continued fractions.^[35]

Thue–Siegel–Roth theorem

The Thue–Siegel–Roth theorem represents a culmination of efforts to refine Dirichlet's approximation theorem for algebraic irrational numbers, establishing sharp bounds on the quality of rational approximations. In 1909, Axel Thue initiated this line of research by proving that for an algebraic irrational \alpha of degree d \geq 2, there are only finitely many rationals p/q satisfying |\alpha - p/q| < 1/q^{\frac{d}{2} + 1 + \epsilon} for any \epsilon > 0.^[36] Carl Ludwig Siegel advanced this in 1921, improving the exponent to \sqrt{8d/3} + \epsilon through more sophisticated analytic methods, particularly for quadratic irrationals initially. Klaus Roth's groundbreaking work in 1955 extended these results to any algebraic irrational, achieving the near-optimal exponent of $2 + \epsilon, independent of the degree d.^[37] The theorem states that if \alpha is an algebraic irrational of degree at least 2, then for every \epsilon > 0, there exist only finitely many rationals p/q (in lowest terms, with q > 0) such that

\left| \alpha - \frac{p}{q} \right| < \frac{1}{q^{2 + \epsilon}}.

This bound holds regardless of the degree of \alpha, marking a significant sharpening of Dirichlet's guarantee of infinitely many approximations within $1/q^2.^[37]^[38] Proofs of the theorem rely on constructing auxiliary polynomials with integer coefficients that vanish to high order at points related to \alpha, leveraging tools like Siegel's lemma for existence and Wronskian determinants to control growth.^[39] Roth's original approach combined these with inductive estimates on the minimal index of polynomials near \alpha, avoiding reliance on p-adic valuations but emphasizing real-analytic bounds.^[37] Modern expositions often incorporate p-adic methods to handle uniformity or effective versions, while generalizations via Wolfgang Schmidt's subspace theorem (1972) extend the result to simultaneous approximations in higher dimensions using non-archimedean places and linear forms in logarithms.^[39] The theorem's significance lies in confirming that Dirichlet's exponent of 2 is essentially optimal for algebraic irrationals, as no better uniform exponent is possible beyond finitely many exceptions, in stark contrast to transcendental numbers like Liouville's constructed examples, which admit approximations exceeding any exponent.^[37] This distinction underpins much of transcendence theory, delimiting the "algebraic" from the "transcendental" in Diophantine contexts.^[38]

References

[1]
[PDF] 8. Dirichlet's Theorem and Farey Fractions
There is a very simple, and useful, theorem due to Dirichlet which tells us how well a real number can be approximated by a rational number a/q in terms of the.
[2]
[PDF] LECTURE 12: DIOPHANTINE APPROXIMATION 1. Dirichlet Theorem
Dirichlet Theorem. Many important ideas in Number Theory stem from notions of Diophantine approximation, which is to say rational approximations to real ...
[3]
[PDF] On a theorem of Davenport and Schmidt - UCLA Mathematics
Jul 13, 2020 · In 1842 Dirichlet [13] applied the pigeonhole principle to give good approximations of real numbers by rationals. One form of his theorem in one ...
[4]
[PDF] Diophantine Approximations
Theorem 1: (Dirichlet's Approximation Theorem) : If ɑ is irrational, then there exist infinitely many rational numbers p/q such that. (3) q p. −α. <. 2. 1 q.<|control11|><|separator|>
[5]
[PDF] DIRICHLET'S THEOREM ON DIOPHANTINE APPROXIMATION ...
We will discuss the history in more detail in §4. In the present paper we significantly generalize Theorems 1.1 and. 1.2 by using a homogeneous dynamics ...
[6]
Dirichlet's theorem with an arbitrarily small constant for algebraic ...
Feb 7, 2023 · Dirichlet's theorem on diophantine approximation asserts that, for every irrational real number α, there are infinitely many ... |α−pq|<1q2 ...Square-free diophantine approximation - MathOverflowPrecise asymptotic of diophantine approximation - MathOverflowMore results from mathoverflow.net
[7]
[PDF] Irrationality Exponent, Hausdorff Dimension and Effectivization
Dec 31, 2015 · Rational numbers have irrationality exponent equal to 1. An immediate consequence of Dirichlet's Approximation. Theorem (see Bugeaud, 2004, ...
[8]
A question on subgroup-restricted irrationality measures
Nov 25, 2013 · By the Dirichlet approximation theorem, the irrationality measure of any irrational number is at least 2, and by what Yann Bugeaud calls "an ...Inhomogeneous Diophantine approximation and irrationality exponentDirichlet's approximation only using prime power as denominatorMore results from mathoverflow.net
[9]
https://webusers.imj-prg.fr/~pierre.lochak/textes/gorki.pdf
[10]
[PDF] SIMULTANEOUS DIOPHANTINE APPROXIMATION IN CLASSICAL ...
Theorem (Dirichlet): For any α ∈ Rn and any real Q > 1 there exists an integer q ∈ Z+, with 1 ≤ q<Q, such that hqαi ≤ Q−1/n. It turns out that one needs only ...<|control11|><|separator|>
[11]
[PDF] An Effective Version of Kronecker's Theorem on Simultaneous ...
Oct 5, 2001 · The vector θ induces a mapping fθ : x 7→ x + θ mod Zn in the torus Tn . This mapping can be interpreted as a dynamical system in Tn. For more ...
[12]
[PDF] Fourier series, Weyl equidistribution 1. Dirichlet's pigeon-hole ...
Oct 14, 2015 · • Dirichlet's pigeon-hole principle, approximation theorem ... Proof: For each m in the range 1 ≤ m ≤ N + 1, choose n = nm so that mα ...
[13]
[PDF] Diophantine Approximations
Jul 4, 2022 · By the Pigeonhole Principle, there exists two integers. 0 ≤ j < k ≤ n such that {j · α} and {k · α} belong in the same sub-interval.
[14]
[PDF] 3. The Geometry of Numbers
Theorem 3.10 (Minkowski's convex body theorem, III). Let Λ be a lattice in Rn, and let C be a convex body in Rn which is symmetric about 0. If C is closed.
[15]
[PDF] Minkowski's Theorem and Its Applications - CMU Math
Theorem. (Dirichlet's approximation). For any α ∈ (0,1) and N ∈ N, there exists a pair of natural numbers m, n with m ⩽ N, such that α − n m. <. 1. mN . Proof.
[16]
[PDF] math 6370: assignment #3
Problem 4: Prove the following theorem of Dirichlet using Minkowski's theorem. Dirichlet's approximation theorem: For any α ∈ R and integer N ≥ 1, there are ...
[17]
[PDF] Geometric Number Theory Lenny Fukshansky
DIRICHLET'S THEOREM. 63. Minkowski-style proof of Dirichlet's theorem. With notation as in the statement of Theorem 4.3.1, let us define two binary linear forms ...<|control11|><|separator|>
[18]
Publications: 1839–1845 | SpringerLink
Dec 30, 2018 · The following spring, on April 14, 1842, Dirichlet read a historic memoir titled “Generalization of a statement from the doctrine of ...
[19]
Verallgemeinerung eines Satzes aus der Lehre von den ...
Verallgemeinerung eines Satzes aus der Lehre von den Kettenbrüchen nebst einigen Anwendungen auf die Theorie der Zahlen · P. G. L. Dirichlet, L. Kronecker ...
[20]
[PDF] DIOPHANTINE APPROXIMATION
Before discussing this, we prove a predecessor, due to Dirichlet. Theorem 1.1 (Dirichlet, 1842). Let α ∈ R. Then for every integer Q ⩾ 2 there are integers x, y ...
[21]
[PDF] arXiv:2010.14760v2 [math.NT] 18 Aug 2022
Aug 18, 2022 · Seeking a quantitative answer leads to the theory of metric Diophantine approximation. Dirichlet's theorem (1842) is the starting point in this ...Missing: Crelle | Show results with:Crelle
[22]
An essay on continued fractions - Leonhard Euler - ResearchGate
Aug 7, 2025 · An Essay on Continued Fractions ... The paper translated here, "De Fractionibus Continuis Dissertatio", represents Euler's first published work on ...Missing: bounds | Show results with:bounds
[23]
[PDF] Diophantine approximation, irrationality and transcendence
Jun 2, 2010 · Euler gave continued fractions expansions not only for e and e2: ... , History of continued fractions and Padé approximants, vol. 12 of ...
[24]
Lagrange and the Solution of Numerical Equations - ScienceDirect
Lagrange, J.-L.1808. Traité de la résolution des equations numériques de tous les degrés, 1798, revised in 1808, in Œuvres de Lagrange, J.-A. Serret, Ed ...
[25]
Diophantine Approximation and its Applications - SpringerLink
Euler proved in 1737 that the continued fraction expansion of any quadratic irrational number is periodic. The converse was proved by Lagrange in 1770.
[26]
Disquisitiones Arithmeticae on JSTOR
In this section we will treat first of the resolution of fractions into simpler ones and of the conversion of common fractions into decimals. We will then ...Missing: Diophantine approximation
[27]
A METHOD IN METRIC DIOPHANTINE APPROXIMATION
A generalization of Khintchine's theorem. Khintchine's Theorem [10], [11] on metric Diophantine approximation has been generalized in various directions [4] ...
[28]
[PDF] Badly approximable numbers over imaginary quadratic fields
Oct 14, 2018 · The obvious (and conjecturally only) algebraic numbers badly approximable over Q are quadratic irrationals. Here is an algebraic proof. b/2 c x ...
[29]
On badly approximable numbers | Mathematika | Cambridge Core
Feb 26, 2010 · As is well known, a single number β is badly approximable if and only if the partial quotients in its continued fraction are bounded.
[30]
Hausdorff Dimension and Diophantine Approximation - math - arXiv
May 28, 2003 · We begin with a brief treatment of Hausdorff measure and Hausdorff dimension. We then explain some of the principal results in Diophantine approximation.
[31]
[PDF] Twenty Years of Attacks on the RSA Cryptosystem 1 Introduction
Boneh, G. Durfee, and Y. Frankel. An attack on RSA given a fraction of the private key bits. In AsiaCrypt '98, volume 1514 of ...
[32]
[PDF] On a Theorem of Legendre on Diophantine Approximation - arXiv
Jul 15, 2024 · During the last three centuries many famous mathematicians came with interesting results in Diophantine approx- imations and continued fractions ...
[33]
[PDF] Increasing rate of weighted product of partial quotients in continued ...
May 31, 2022 · The next theorem, known as Legendre's Theorem, connects 1-dimensional Diophantine approximation with continued fractions. Theorem 2.1 (Legendre) ...
[34]
[PDF] arXiv:2211.01956v2 [math.HO] 7 Aug 2023
Aug 7, 2023 · Continued fractions exhibit numerous remarkable properties related to the Euclidean algorithm for integers or real numbers. Every rational ...
[35]
Ueber die angenäherte Darstellung der Irrationalzahlen durch ...
Cite this article. Hurwitz, A. Ueber die angenäherte Darstellung der Irrationalzahlen durch rationale Brüche. Math. Ann. 39, 279–284 (1891).
[36]
[PDF] APPROXIMATION BY CONTINUED FRACTIONS
Cohn [1] gave a simple proof of Hurwitz' theorem, that for any real irrational number x there are infinitely many rational num- bers p/q such that \x - p/q ...<|control11|><|separator|>
[37]
[PDF] Notes on Diophantine approximation and aperiodic order
Jun 24, 2016 · Therefore, Borel and Bernstein's theorem tells us that badly approximable numbers are not typical, in the sense of Lebesgue measure. However, it ...
[38]
https://ecroot.math.gatech.edu/siegel.pdf
[39]
https://arxiv.org/pdf/2502.00731.pdf
[40]
[PDF] An Outline of the Thue-Siegel Theorem - Ernie Croot's
May 22, 2007 · The purpose of this paper is to give a somewhat detailed outline of the proof of Siegel's theorem, where we focus more on the ideas rather than ...Missing: history | Show results with:history
[41]
https://arxiv.org/pdf/2502.00731.pdf