logo

P-adic Numbers in Number Theory 📂Number Theory

P-adic Numbers in Number Theory

Definition 1

A pp-adic valuation for a prime number pp and an integer aZa \in \mathbb{Z} is defined as follows with respect to vpv_{p} and is represented as aa of pp-adic valuation. vp(a):=sup{eZ:pea} v_{p} (a) := \sup \left\{ e \in \mathbb{Z} : p^{e} \mid a \right\}

Theorem 2

  • [0]: For every prime number pp vp(0)= v_{p} (0) = \infty
  • [1]: vp(xy)=vp(x)+vp(y)v_{p} (xy) = v_{p}(x) + v_{p}(y)
  • [2]: vp(x+y)min{vp(x),vp(y)}v_{p} (x+y) \ge \min \left\{ v_{p} (x) , v_{p} (y) \right\}
  • [3]: If nNn \in \mathbb{N}, x,yZx , y \in \mathbb{Z}, and prime number pp is gcd(n,p)=1p(xy)pxpy \gcd (n,p) = 1 \\ p \mid (x \mp y) \\ p \nmid x \\ p \nmid y then vp(xn±yn)=vp(x±y) v_{p} \left( x^{n} \pm y^{n} \right) = v_{p} \left( x \pm y \right)
  • [4]: If x,yZx , y \in \mathbb{Z}, prime number p2p \ne 2 is p(xy)pxpy p \mid (x - y) \\ p \nmid x \\ p \nmid y then vp(xpyp)=vp(xy)+1 v_{p} \left( x^{p} - y^{p} \right) = v_{p} \left( x - y \right) +1

  • bab \mid a represents that bb is a divisor of aa.
  • The term valuation means assigning a value, which could be directly translated into Korean as 값매김, but since both terms are somewhat lacking, it is recommended to use the English term Valuation.

Description

When I was taking an undergraduate course in cryptography, I noticed that the textbook used the form of quotient ring rather than the ring of integers modulo pp, and I asked the professor why. He answered that it was to differentiate it from the research on pp-adic numbers, which is a major branch in number theory. Studying the pp-adic numbers might as well be opening a big curtain in the field of number theory.

Simply put, a pp-adic valuation of a given natural number is nothing but looking at how many powers of pp are multiplied. For example, for a prime number p=3p = 3, the 33-adic valuation of 63=327163 = 3^{2} \cdot 7^{1} is v3(63)=2v_{3} (63) = 2, and v7(63)=1v_{7} (63) = 1. On the other hand, for invisible numbers when factoring, for example, the 22-adic valuation is trivially 20632^{0} \mid 63, so it’s v2(63)=0v_{2} (63) = 0.

Proof

[0]

The fact that bb is a divisor of aa, that is, bab \mid a, means that there exists an integer kZk \in \mathbb{Z} that satisfies ak=ba k = b. For all eZe \in \mathbb{Z}, there exists k=0 k = 0 that satisfies pek=0p^{e} \cdot k = 0, so sup{eZ:pe0}=\sup \left\{ e \in \mathbb{Z} : p^{e} \mid 0 \right\} = \infty.

[1]

vp(a)v_{p}(a) is counting the power of pp in aa, so naturally vp(xy)=vp(x)+vp(y)v_{p} (xy) = v_{p}(x) + v_{p}(y) holds.

[2]

Let’s represent some X,YX,Y \in as the following for x,yx,y. x:=pvp(x)Xy:=pvp(y)Y x := p^{v_{p} (x) } X \\ y := p^{v_{p} (y) } Y Without losing generality, if we set it as vp(x)vp(y)v_{p} (x) \ge v_{p} (y), then x+y=pvp(y)(pvp(x)vp(y)X+Y) x+y = p^{v_{p}(y) } \left( p^{v_{p}(x) - v_{p}(y)} X + Y \right) Therefore, we obtain at least vp(x+y)vp(y)v_{p} (x+y) \ge v_{p}(y).

[3]3

Part 1.

If we factorize xnynx^{n} - y^{n}, xnyn=(xy)(xn1+xn2y+xyn2+yn1) x^{n} - y^{n} = (x-y) \left( x^{n-1} + x^{n-2} y + \cdots x y^{n-2} + y^{n-1} \right) According to the definition of vpv_{p}, unless the second factor t=0n1x(n1)tyt\sum_{t = 0}^{n-1} x^{(n - 1)-t} y^{t} includes pp as a divisor, vp(xnyn)v_{p} \left( x^{n} - y^{n} \right) or vp(xy) v_{p} \left( x - y \right) are the same. Writing this as an equation gives us: vp(xnyn)=vp(xy) v_{p} \left( x^{n} - y^{n} \right) = v_{p} \left( x - y \right)


Part 2. vp(xnyn)=vp(xy)v_{p} \left( x^{n} - y^{n} \right) = v_{p} \left( x - y \right)

Since we assume p(xy)p \mid (x-y), we have xy0(modp)x - y \equiv 0 \pmod{p}, namely xy(modp)x \equiv y \pmod{p}, so t=0n1x(n1)tytt=0n1x(n1)txtnxn1(modp) \begin{align*} \sum_{t = 0}^{n-1} x^{(n - 1)-t} y^{t} \equiv& \sum_{t = 0}^{n-1} x^{(n - 1)-t} x^{t} \\ \equiv& n\cdot x^{n-1} \pmod{p} \end{align*} However, assuming pxp \nmid x and gcd(n,p)=1\gcd (n,p) = 1 gives t=0n1x(n1)tyt≢0(modp) \sum_{t = 0}^{n-1} x^{(n - 1)-t} y^{t} \not\equiv 0 \pmod{p} Therefore, we obtain the following: vp(xnyn)=vp(xy) v_{p} \left( x^{n} - y^{n} \right) = v_{p} \left( x - y \right)


Part 3. vp(xn+yn)=vp(x+y)v_{p} \left( x^{n} + y^{n} \right) = v_{p} \left( x + y \right)

It does not go beyond a change in sign from the equation obtained in Part 2. Substituting y-y instead of yy gives vp(xn(y)n)=vp(x(y)) v_{p} \left( x^{n} - (-y)^{n} \right) = v_{p} \left( x - (-y) \right) Thus, the following is obtained: vp(xn+yn)=vp(x+y) v_{p} \left( x^{n} + y^{n} \right) = v_{p} \left( x + y \right)

[4]

Part 1.

vp(xpyp)=vp(xy)+1 v_{p} \left( x^{p} - y^{p} \right) = v_{p} \left( x - y \right) +1 This means that when you factorize (xy)(x-y) and (xpyp)\left( x^{p} - y^{p} \right), pp is exactly added by 11. To demonstrate this, let’s factorize xpypx^{p} - y^{p} in the same manner as in Proof [3]. xpyp=(xy)(xp1+xp2y+xyp2+yp1) x^{p} - y^{p} = (x-y) \left( x^{p-1} + x^{p-2} y + \cdots x y^{p-2} + y^{p-1} \right) That pp is exactly multiplied by 11 means the second factor t=0p1x(p1)tyt\sum_{t = 0}^{p-1} x^{(p - 1)-t} y^{t} appears as a multiple of pp but not as a multiple of p2p^{2}. pt=0p1x(p1)tytp2t=0p1x(p1)tyt p \mid \sum_{t = 0}^{p-1} x^{(p - 1)-t} y^{t} \\ p^{2} \nmid \sum_{t = 0}^{p-1} x^{(p - 1)-t} y^{t}


Part 2. pt=0p1x(p1)tytp \mid \sum_{t = 0}^{p-1} x^{(p - 1)-t} y^{t}

Since we assume p(xy)p \mid (x-y), we have xy0(modp)x - y \equiv 0 \pmod{p}, namely xy(modp)x \equiv y \pmod{p}, so t=0p1x(p1)tytt=0p1x(p1)txtpxp10(modp) \begin{align*} \sum_{t = 0}^{p-1} x^{(p - 1)-t} y^{t} \equiv& \sum_{t = 0}^{p-1} x^{(p - 1)-t} x^{t} \\ \equiv& p\cdot x^{p-1} \\ \equiv& 0 \pmod{p} \end{align*} Therefore, it’s pt=0p1x(p1)tytp \mid \sum_{t = 0}^{p-1} x^{(p - 1)-t} y^{t}.


Part 3. p2t=0p1x(p1)tytp^{2} \nmid \sum_{t = 0}^{p-1} x^{(p - 1)-t} y^{t}

Since it’s xy(modp)x \equiv y \pmod{p}, any kZk \in \mathbb{Z} can be set as y=x+kpy = x + kp. Fixing the index as t=1,,p1t = 1, \cdots , p-1 and expanding x(p1)tytx^{(p - 1)-t} y^{t} over xx gives us in (modp2)\pmod{p^{2}} x(p1)tytx(p1)t(x+kp)tx(p1)t(xt+txt1kp+t(t1)2xt2k2p2+)x(p1)t(xt+txt1kp)+O(p2)xp1+tkpxp2(modp2) \begin{align*} x^{(p - 1)-t} y^{t} \equiv& x^{(p - 1)-t} \left( x + kp \right)^{t} \\ \equiv& x^{(p-1)-t} \left( x^{t} + t x^{t-1} kp + {{ t(t-1) } \over { 2 }} x^{t-2} k^{2} \cdot p^{2} + \cdots \right) \\ \equiv& x^{(p-1)-t} \left( x^{t} + t x^{t-1} kp \right) + O \left( p^{2} \right) \\ \equiv& x^{p-1} + tkpx^{p-2} \pmod{p^{2}} \end{align*} Returning to the original series gives t=0p1x(p1)tytt=0p1[xp1+tkpxp2]pxp1+p(p1)2kpxp2pxp1+p12kxp2p2pxp1(modp2)≢0(modp2) \begin{align*} \sum_{t = 0}^{p-1} x^{(p - 1)-t} y^{t} \equiv& \sum_{t = 0}^{p-1} \left[ x^{p-1} + tkpx^{p-2} \right] \\ \equiv& p x^{p-1} + {{ p(p-1) } \over { 2 }} kpx^{p-2} \\ \equiv& p x^{p-1} + {{ p-1 } \over { 2 }} k x^{p-2} \cdot p^{2} \\ \equiv& p x^{p-1} \pmod{p^{2}} \\ \not \equiv& 0 \pmod{p^{2}} \end{align*} Therefore, it’s p2t=0p1x(p1)tytp^{2} \mid \sum_{t = 0}^{p-1} x^{(p - 1)-t} y^{t}.