找回密码
 快速注册
搜索
查看: 1095|回复: 12

3x + 1 问题

[复制链接]

443

主题

1519

回帖

1万

积分

积分
11660

显示全部楼层

realnumber 发表于 2014-6-9 19:18 |阅读模式
从任意一个正整数开始,重复对其进行下面的操作:如果这个数是偶数,把它除以 2 ;如果这个数是奇数,则把它扩大到原来的 3 倍后再加 1
。序列是否最终总会变成 4, 2, 1, 4, 2, 1, „ 的循环?      这个问题可以说是一个“坑”——乍看之下,问题非常简单,突破口很多,
于是数学家们纷纷往里面跳;殊不知进去容易出去难,不少数学家到死都没把这个问题搞出来。
已经中招的数学家不计其数,这可以从 3x + 1 问题的各种别名看出来: 3x + 1 问题又叫 Collatz 猜想、 Syracuse 问题、 Kakutani 问题、 Hasse 算法、 Ulam 问题等等。
后来,由于命名争议太大,干脆让谁都不沾光,直接叫做 3x + 1 问题算了。   
  3x + 1 问题不是一般的困难。这里举一个例子来说明数列收敛有多么没规律。从 26 开始算起, 10 步就掉入了“421 陷阱”: 26, 13, 40, 20, 10, 5, 16, 8, 4, 2, 1, 4, 2, 1, „      
但是,从 27 开始算起,数字会一路飙升到几千多,你很可能会一度认为它脱离了“421 陷阱”;但是,经过上百步运算后,它还是跌了回来:  27, 82, 41, 124, 62, 31, 94, 47, 142, 71, 214, 107, 322, 161, 484, 242, 121, 364, 182, 91, 274, 137, 412, 206, 103, 310, 155, 466, 233, 700, 350, 175, 526, 263, 790, 395, 1186, 593, 1780, 890, 445, 1336, 668, 334, 167, 502, 251, 754, 377, 1132, 566, 283, 850, 425, 1276, 638, 319, 958, 479, 1438, 719, 2158, 1079, 3238, 1619, 4858, 2429, 7288, 3644, 1822, 911, 2734, 1367, 4102, 2051, 6154, 3077, 9232, 4616, 2308, 1154, 577, 1732, 866, 433, 1300, 650, 325, 976, 488, 244, 122, 61, 184, 92, 46, 23, 70, 35, 106, 53, 160, 80, 40, 20, 10, 5, 16, 8, 4, 2, 1, 4, 2, 1, „
wenku.baidu.com/view/d3a297d2a58da0116c174990.html

443

主题

1519

回帖

1万

积分

积分
11660

显示全部楼层

 楼主| realnumber 发表于 2014-6-9 19:26
本帖最后由 realnumber 于 2024-11-9 11:53 编辑 出现次数超过一半的元素     
令 U 是一个有限集,S1 , S2 , „ , Sn 都是 U 的非空子集,它们满足任意多个集合的并集仍然在这些集合里。证明,一定能找到某个元素,它出现在了至少一半的集合里。      不可思议,即使是最基本最离散的数学研究对象——有限集——里面,也有让人崩溃的未解问题。      1999 年, Piotr Wojcik 用一种非常巧妙的方法证明了,存在一个元素出现在了至少 n/log2n 的集合里。不过,这离目标还有很大一段距离。




打个草稿,与1楼无关
p>3为素数,a,b为正整数,a>b
其实是打算$(1+x)^{ap}=(1+x)^{p}(1+x)^p(1+x)^{(a-2)p}$两边二项展开,不太好表达,先按以下展开
$(1+x)^{ap}=(1+x)^{2p}(1+x)^{(a-2)p}$两边二项展开,由$x^{bp}$的系数,可得
$C_{ap}^{bp}=\color{blue}{C_{2p}^0C_{(a-2)p}^{bp}}+\color{green}{C_{2p}^{1}C_{(a-2)p}^{bp-1}+...+C_{2p}^{p-1}C_{(a-2)p}^{(b-1)p+1}}+\color{blue}{C_{2p}^{p}C_{(a-2)p}^{(b-1)p}}+\color{green}{C_{2p}^{p+1}C_{(a-2)p}^{(b-1)p-1}+...++C_{2p}^{2p-1}C_{(a-2)p}^{(b-2)p+1}}+\color{blue}{C_{2p}^{2p}C_{(a-2)p}^{(b-2)p}}$
若绿色2部分都有
$\color{green}{C_{2p}^{1}C_{(a-2)p}^{bp-1}+...+C_{2p}^{p-1}C_{(a-2)p}^{(b-1)p+1}}=0 \mod  p^3$---(1)
$\color{green}{C_{2p}^{p+1}C_{(a-2)p}^{(b-1)p-1}+...++C_{2p}^{2p-1}C_{(a-2)p}^{(b-2)p+1}}=0\mod p^3$---(2)成立
则蓝色部分$C_{ap}^{bp}=\color{blue}{C_{2p}^0C_{(a-2)p}^{bp}}+\color{blue}{C_{2p}^{p}C_{(a-2)p}^{(b-1)p}}+\color{blue}{C_{2p}^{2p}C_{(a-2)p}^{(b-2)p}}(??=C_{a-2}^{b}+C_2^1C_{a-2}^{b-1}+C_{a-2}^{b-2}) \mod p^3$成立,??处,接下来对a,b用数学归纳法证明,完.

以下证明绿色部分(1),其中$C_{2p}^i=C_p^0C_p^i+C_p^1C_p^{i-1}+...+C_p^iC_p^0=2C_p^i \mod p^2,i=1,2,3,..,p-1$
$C_{(a-2)p}^k=\frac{(a-2)p}{k}C_{(a-2)p-1}^{k-1},k=bp-1,bp-2,...,(b-1)p+1$(这些k都有 $k\nmid p$)
$C_p^i=\frac{p}{i}C_{p-1}^{i-1},i=1,2,3,..,p-1$

如此要证明(1)成立,只需要证明
$\sum_{i=1}^{p-1}\frac{(a-2)}{i(bp-i)}C_{p-1}^{i-1}C_{(a-2)p-1}^{bp-i}=0 \mod p$,分母理解为逆元,以下同.或参考6楼的习题2
由Lucas定理,只需要证明
$\sum_{i=1}^{p-1}\frac{1}{i^2}C_{p-1}^{i-1}C_{a-1}^{b-1}C_{p-1}^{p-i}=0 \mod p$
即只需证明$t\sum_{i=1}^{p-1}\frac{1}{i^2}C_{p-1}^{i-1}C_{p-1}^{i-1}=(C_p^1C_p^1+C_p^2C_p^2+....+C_p^{p-1}C_p^{p-1})=C_{2p}^p-2=0 \mod p$  $t=((p-1)!)^4$
(2)大概也可以这样完成吧.

点评

https://en.wikipedia.org/wiki/Union-closed_sets_conjecture  发表于 2024-9-21 03:36

830

主题

4862

回帖

3万

积分

积分
36159

显示全部楼层

isee 发表于 2014-6-9 20:41
数学史话?小史?

66

主题

416

回帖

3566

积分

积分
3566

显示全部楼层

Tesla35 发表于 2014-6-9 22:08
3x+1问题m67的书里有说到

830

主题

4862

回帖

3万

积分

积分
36159

显示全部楼层

isee 发表于 2014-6-9 23:33
matrix67?

471

主题

945

回帖

9837

积分

积分
9837

显示全部楼层

青青子衿 发表于 2014-6-21 12:06
回复 5# isee
zh.wikipedia.org/wiki/考拉兹猜想
奇偶归一猜想(英语:Collatz conjecture),又称为3n+1猜想、冰雹猜想、角谷猜想、哈塞猜想、乌拉姆猜想或叙拉古猜想,是指对于每一个正整数,如果它是奇数,则对它乘3再加1,如果它是偶数,则对它除以2,如此循环,最终都能够得到1。
en.wikipedia.org/wiki/Collatz_conjecture
PS:中文的维基百科,和英文的wikipedia介绍篇幅天壤之别,看来英文学得好还占优势!

3149

主题

8387

回帖

6万

积分

$\style{scale:11;fill:#eff}꩜$

积分
65396
QQ

显示全部楼层

hbghlyj 发表于 2022-7-20 08:29
本帖最后由 hbghlyj 于 2024-9-20 19:31 编辑 陶哲轩的博文中带有 Collatz 猜想的标签的博文:terrytao.wordpress.com/tag/collatz-conjecture/

The 3x+1 Problem: An Overview (arxiv.org/pdf/2111.02635.pdf)
The Ultimate Challenge: The 3x+1 Problem (Libgen)

3149

主题

8387

回帖

6万

积分

$\style{scale:11;fill:#eff}꩜$

积分
65396
QQ

显示全部楼层

hbghlyj 发表于 2022-7-21 00:09
A 2-ADIC EXTENSION OF THE COLLATZ FUNCTION
A Formal Proof of Hensel’s Lemma over the p-adic Integers

Definition 3.1. The p-adic valuation function $V_{p}: \mathbb{Z} \rightarrow \mathbb{Z} \cup\left\{+\infty\right\}$ is defined below:
$$
V_{p}(x)= \begin{cases}+\infty & \text { if } x=0 \\ \max \left\{z \mid p^{z} \text { divides } x\right\} & \text { if } x \in \mathbb{Z} \text { and } x \neq 0\end{cases}
$$
Note that by the Fundamental Theorem of Arithmetic, $V_{p}(x y)=V_{p}(x)+V_{p}(y)$. It is thus natural to extend $V_{p}$ to the rational numbers $a / b$ as follows: $V_{p}(a / b)=$ $V_{p}(a)-V_{p}(b)$ for all $a, b \in \mathbb{Z}$ such that $a, b \neq 0$, We can now extend the fact that $V_{p}(x y)=V_{p}(x)+V_{p}(y)$ to our extended p-adic valuation function:
Lemma 3.3. For all $x, y \in \mathbb{Q}, V_{p}(x y)=V_{p}(x)+V_{p}(y)$
Proof. If $x y \in \mathbb{Q}$ and $x y \notin \mathbb{Z}$, then let $x=q / r$ and $y=s / t$ so that $V_{p}(x y)=V_{p}(q s / r t)=V_{p}(q s)-V_{p}(r t)$ by the definition of the p-adic valuation function. So since this lemma is true for all integers, $V_{p}(q s)-V_{p}(r t)=V_{p}(q)+V_{p}(s)-V_{p}(r)-V_{p}(t)=V_{p}(q)-V_{p}(r)+V_{p}(s)-V_{p}(t)=V_{p}(q / r)+V_{p}(s / t)$, which by definition equals $V_{p}(x)+V_{p}(y)$
We can now present three properties of the p-adic valuation function:
(1) For all $x \in \mathbb{Z}, V_{p}(x) \geq 0$.
Proof. Let $y=V_{p}(x)=\max \left\{z \mid p^{z}\right.$ divides $\left.x\right\}$. Since $x$ must be an integer, $p^{0} \mid x$. And since $0 \in\left\{z \in \mathbb{Z}\right.$ such that $\left.p^{z} \mid x\right\}, V_{p}(x)=\max \left\{0, z_{1}, z_{2}, z_{3}, \ldots\right\} \geq$ $0 .$
(2) For all $x \in \mathbb{Q}, V_{p}(x)=V_{p}(-x)$.
Proof. For $x \in \mathbb{Z}$, this property follows directly from the fact that if $p^{k} \mid z$, then $p^{k} \mid-z$. For $x \in \mathbb{Q}$, this property follows from Lemma (3.3) and the application of this property to integers as described in the previous case.
(3) For all $x, y \in \mathbb{Q}, V_{p}(x+y) \geq \min \left\{V_{p}(x), V_{p}(y)\right\}$
Proof. Without loss of generality, let $z=V_{p}(x)=\min \left\{V_{p}(x), V_{p}(y)\right\}$. So choose $a, b \in \mathbb{Z}$ such that $x=a p^{z}$ and $y=b p^{z}$ with $a$ not divisible by $p$. Factoring out $p^{z}$ results in $x+y=p^{z}(a+b)$. Furthermore, $V_{p}(x+y)=$ $V_{p}\left(p^{z}(a+b)\right)=V_{p}\left(p^{z}\right)+V_{p}(a+b)$ by Lemma (3.3). This contracts to $V_{p}(x+y)=z+V_{p}(a+b)$. And since $V_{p}(a+b) \geq 0$ by property (1), $V_{p}(x+y) \geq z=V_{p}(x)=\min \left\{V_{p}(x), V_{p}(y)\right\}$.

3149

主题

8387

回帖

6万

积分

$\style{scale:11;fill:#eff}꩜$

积分
65396
QQ

显示全部楼层

hbghlyj 发表于 2022-7-21 01:30
Definition 3.4. The $p$-adic distance function $d_{p}: \mathbb{Q} \times \mathbb{Q} \rightarrow \mathbb{Q} \geq 0$ is defined as follows:
$$
d_{p}(x, y)= \begin{cases}\frac{1}{p^{V_{p}(x-y)}} & \text { if } x \neq y \\ 0 & \text { if } x=y\end{cases}
$$
Theorem 3.6. The rational numbers $\mathbb{Q}$ together with the $p$-adic distance function (3.5) form an ultrametric space.

Proof. (Property 1) This is an obvious consequence of the definition of $d_{p}(x, y)$.
(Property 2) If $x=y$, then by definition $d_{p}(x, y)=0=d_{p}(y, x)$. But if $x \neq y$, then it is apparent that $d_{p}(x, y)$ equals $d_{p}(y, x)$ if and only if $V_{p}(x-y)$ equals $V_{p}(y-x)$ which follows directly from Property (2) of the p-adic valuation function[2].
(Property 3) Note that $x-y=x-z+z-y$. So by property (3) of the p-adic valuation function, $V_{p}(x-y)=V_{p}(x-z+z-y) \geq \min \left\{V_{p}(x-z), V_{p}(z-y)\right\}$. Without loss of generality assume $d_{p}(x, z)=\max \left\{d_{p}(x, z), d_{p}(y, z)\right\}$. From the definition of the p-adic distance function, this occurs if and only if $V_{p}(x-z)=\min \left\{V_{p}(x-\right.$ $\left.z), V_{p}(y-z)\right\}$. So $V_{p}(x-y) \geq V_{p}(x-z)$. And it follows that $p^{V_{p}(x-y)} \geq p^{V_{p}(x-z)}$. So $\frac{1}{p^{V_{p}(x-y)}} \leq \frac{1}{p^{V_{p}(x-z)}}$. Which results in the equation: $d_{p}(x, y) \leq d_{p}(x, z)=$ $\max \left\{d_{p}(x, z), d_{p}(y, z)\right\}$. So $d_{p}(x, y)$ forms an ultrametric space over $\mathbb{Q}$.

4. Completeness

Definition 4.1. A sequence $\left(a_{0}, a_{1}, \ldots\right)$ of elements from a metric space $M=$ $\{X, d(x, y)\}$ is Cauchy if for all $\epsilon>0$, there exists an $N \in \mathbb{N}$ such that for all $m, n \geq N, d\left(a_{m}, a_{n}\right)<\epsilon$.  [5]

Definition 4.2. A metric space $M$ is complete if every Cauchy sequence in $M$ converges to an element of $M$.

Example 4.3. The metric space $M=\{\mathbb{Q}, \|\}$, defined in Example (2.4), is not a complete metric space.

Proof. Consider the sequence $\left(a_{i}\right)$ where each $a_i$ is the irrational number $e=2.718 \ldots$ rounded to the $i^{t h}$ decimal place. It is obvious that the sequence is Cauchy because for any $\epsilon$, choose $N$ such that $10^{-N}<\epsilon$ so that $d\left(a_{m}, a_{n}\right)=\left|a_{m}-a_{n}\right|$ is less than $\epsilon$. But since $e$ is not a rational number, and since this Cauchy Sequence converges to $e, M$ is not a complete metric space.

3149

主题

8387

回帖

6万

积分

$\style{scale:11;fill:#eff}꩜$

积分
65396
QQ

显示全部楼层

hbghlyj 发表于 2022-7-21 01:36
5. The Field of p-adic Numbers

Definition 5.1. The field of $p$-adic numbers $\mathbb{Q}_{p}$ can be defined for each $p$ prime as the completion of $\mathbb{Q}$ under the $p$-adic distance function $d_{p}(x, y)$. In other words, $\mathbb{Q}_{p}$ is the set of the limits of all Cauchy sequences in $\mathbb{Q}$.

The real numbers are a completion of the rational numbers under the standard absolute value metric, and each real number can be written as $x=\pm \sum_{i=n}^{\infty} a_{i} 10^{-i}$ where each $a_{i}$ is a single digit (i.e. $a_{i} \in\{0,1,2,3,4,5,6,7,8,9\}$ ). We have just defined the base-10 decimal representation of $x$. So, for example, $82.34=8 \cdot 10^{1}+2 \cdot 10^{0}+3 \cdot 10^{-1}+4 \cdot 10^{-2}+0 \cdot 10^{-3}$. This idea extends perfectly to $p$-adic numbers.
Theorem 5.2. (Hensel's Representation of the p-adic numbers): There is a natural bijection between $\mathbb{Q}_{p}$ and $\left\{\sum_{i=s}^{\infty} a_{i} p^{i} \mid n \in \mathbb{Z}, a_{i} \in\{0,1, \ldots, p-1\}, s \in \mathbb{Z}\right\}$.

A proof of this theorem would require making precise the definition of completion, which is beyond the scope of this paper. But the main idea is as follows: the mapping takes the sum $\sum_{i=s}^{\infty} a_{i} p^{i}$ to the sequence $\left(b_{n}\right)$ where $b_{n}=\sum_{i=s}^{n} a_{i} p^{i}$. This is clearly a Cauchy sequence. The remainder of the proof shows that every Cauchy sequence "has the same limit as" one of these partial-sum sequences; where "has the same limit as" is an equivalence relation on Cauchy sequences that may be made precise.

Definition 5.3. The p-adic Integers are defined (using the Hensel Representation) as $\mathbb{Z}_{p}=\left\{\sum_{i=0}^{\infty} a_{i} p^{i}\right\}$
Theorem 5.4. There is a natural embedding of $\mathbb{Q}$ in $\mathbb{Q}_{p}$.
Proof. For each $q \in \mathbb{Q}$, examine the constant sequence $A=(q, q, q, q, q, q, \ldots)$ with $a_{i}=q$ for all $i \in \mathbb{N} . \quad A$ is Cauchy because for any $a_{m}$ and $a_{n}, d_{p}\left(a_{m}, a_{n}\right)=$ $d_{p}(q, q)=\frac{1}{p^{V_{p}(0)}}=0$ which is by definition less than $\epsilon$ for all $\epsilon>0$. It is equally evident that the sequence $A$ converges to $q$.

Theorem 5.5. The embedding in Theorem (5.4) induces an embedding of $\mathbb{Z}^{\geq} \geq 0$ in $\mathbb{Z}_{p}$; and the unique Hensel Representation of each $z \in \mathbb{Z}^{\geq}=0$ is $\sum_{i=0}^{n} a_{i} p^{i}$ for some $n \in \mathbb{N}$

Proof. (Existence) This proof uses induction. For $z=0, n=0$, because the padic Hensel Representation of $z$ is $0 \cdot p^{0}+0 \cdot p^{1}+\ldots$ Now assume that our theorem holds for all $z \in \mathbb{Z}^{\geq 0}$ such that $z$ is less than or equal to a given $k$. Examine $k+1$. Let $q=\max \left\{s \mid k+1-p^{s}>0\right\}$. Since $0<k+1-p^{q} \leq k$, there exists an $n \in \mathbb{N}$ such that the p-adic Hensel Representation of $k+1-p^{q}$ is $\sum_{i=0}^{n} a_{i} p^{i}$. Also, $a_{i}=0$ for all $i \geq q$ since $k+1-p^{q}<p^{q}$. So for the integer $k+1, n=q$.
(Uniqueness) Assume that there exist two distinct p-adic Hensel Representations of $z \in \mathbb{Z} \geq 0$ : $\sum_{i=0}^{n} a_{i} p^{i}$ and $\sum_{i=0}^{m} b_{i} p^{i}$. If $z \equiv r\pmod p$ for some $r \in \mathbb{Z} \geq 0$, then $a_{0}$ must equal $r$ and $b_{0}$ must equal $r$. Since this is true for all $r \in \mathbb{Z} \geq 0, a_{0}=b_{0}$. So substracting the equal terms $a_{0}$ and $b_{0}$ from their respective summations, we must still be left with distinct coefficients since we began with distinct representations of $z$. Now assume that $a_{i}=b_{i}$ for all $i$ less than or equal to a given $k$. Examine $a_{k+1}$ and $b_{k+1}$. If $z \equiv t\pmod{p^{k+2}}$ for some $t \in \mathbb{Z} \geq 0$, then to preserve equality between each summation and $z, a_{k+1}$ must equal $q$ and $b_{k+1}$ must equal $q$ since each $a_{i} p^{i}$ equals $b_{i} p^{i}$ for $i \leq k$. So by induction, each coefficient $a_{i}$ must equal $b_{i}$ which contradicts our assumption that there exist two distinct p-adic Hensel Representations of $z$.

3149

主题

8387

回帖

6万

积分

$\style{scale:11;fill:#eff}꩜$

积分
65396
QQ

显示全部楼层

hbghlyj 发表于 2023-4-18 09:52
本帖最后由 hbghlyj 于 2024-9-20 19:28 编辑
realnumber 发表于 2014-6-9 12:26
http://wenku.baidu.com/view/d3a297d2a58da0116c174990.html

3149

主题

8387

回帖

6万

积分

$\style{scale:11;fill:#eff}꩜$

积分
65396
QQ

显示全部楼层

hbghlyj 发表于 2024-9-21 03:43
realnumber 发表于 2014-6-9 11:26

出现次数超过一半的元素

    令 U 是一个有限集,S1 , S2 , … , Sn 都是 U 的非空子集,它们满足任意多个集合的并集仍然在这些集合里。证明,一定能找到某个元素,它出现在了至少一半的集合里。

    不可思议,即使是最基本最离散的数学研究对象——有限集——里面,也有让人崩溃的未解问题。

    1999 年, Piotr Wojcik 用一种非常巧妙的方法证明了,存在一个元素出现在了至少 n/log2n 的集合里。不过,这离目标还有很大一段距离。


2024 年发表了一篇论文 https://arxiv.org/pdf/2405.03731,声称证明了这个猜想。
从论文来看,虽然篇幅很短,但它引用了其他论文。

这个证明正确吗?
这里有没有专家可以看一下稿件并发表一下意见?

手机版|悠闲数学娱乐论坛(第3版)

GMT+8, 2025-3-4 16:06

Powered by Discuz!

× 快速回复 返回顶部 返回列表