Posted on

Table of Contents

Definition

Let f(x1,x2,...,xn)f(x_1,x_2,...,x_n) be a polynomial in n variables with integer coefficients and f(x1,x2,...,xn)0(m)f(x_1,x_2,...,x_n)\equiv 0(m). A solution is that f(a1,a2,...,an)0(m)f(a_1,a_2,...,a_n)\equiv 0(m). Thus fˉ(aˉ1,aˉ2,....,aˉn)=0ˉ\bar f(\bar a_1, \bar a_2,....,\bar a_n)=\bar 0.

Proposition 1

For axb(m)ax\equiv b(m), consider d>0d > 0 and (a,m)=d(a,m)=d. axb(m)ax\equiv b(m) has solutions and have exactly d solutions iff dbd|b.

For a solution x0x_0, then there are other solutions given by x0+m,x0+2m,x0+(d1)mx_0 + m', x_0+2m', x_0+(d-1)m', where m=m/dm'=m/d.

Proof:

If dbd|b, we shall have axmy=dax'-my'=d. So we have my=daxmy'=d-ax', times b/db/d on both side we have myb/d=baxb/dmy'b/d=b-ax'b/d. Now we have a solution x=xb/dx=x'b/d.

If x0x_0 is a solution, we have my0=bax0my_0=b-ax_0. Thus b=my0ax0b=my_0-ax_0. That is b(a,m)    dbb\in(a,m) \implies d|b.

If x0,x1x_0,x_1 are two solutions, ax0b(m)ax_0\equiv b(m), ax1b(m)ax_1\equiv b(m). Therefore, a(x0x1)0(m)    ma(x0x1)a(x_0-x_1)\equiv 0(m) \implies m|a(x_0-x_1).

By dividing dd, we have ma(x0x1)m'|a'(x_0-x_1). And as we know, (m,a)=1(m',a')=1. So m(x0x1)m'|(x_0-x_1).

Now we can have a linear combination: x1=x0+kmx_1=x_0+km' where kZk\in\Z.

Corollary 1

If (a,m)=1(a,m)=1, then axb(m)ax\equiv b(m) has one and only solution.

Proof:

Clearly, the number of solution is (a,m)=1(a,m)=1.

Corollary 2

If p is prime and a≢0(p)a \not\equiv 0(p) , there is one and only solution for axb(p)ax\equiv b(p).

Proof:

(a,p)=1(a,p)=1.

Unit

Proposition 2

As previously discussed, axb(m)ax\equiv b(m) is equivalent to aˉx=bˉ\bar a x=\bar b over Z/mZ\Z/m\Z.

So we can define a unit aˉ\bar a iff aˉx=1ˉ\bar ax=\bar1, that is ax1(m)ax\equiv 1(m) is solvable. Thus iff (a,m)=1(a,m)=1, aˉ\bar a is a unit.

It follows there are ϕ(m)\phi(m) units in Z/mZ\Z/m\Z​.

So in Z/pZ\Z/p\Z, every nonzero element is a unit, that makes it a field.

But for a composition number m=m1m2m=m_1m_2. We have mˉ10ˉ,mˉ20ˉ\bar m_1 \ne \bar 0, \bar m_2 \ne \bar 0 , however, m1m2=mˉ=0ˉ\overline{m_1m_2}=\bar m = \bar 0. So Z/mZ\Z/m\Z is not a field.

Corollary 2 (Euler's Theorem)

If (a,m)=1(a,m)=1, then aϕ(m)1(m)a^{\phi(m)}\equiv 1 (m)

Proof:

Since (a,m)=1(a,m)=1, aˉ\bar a is a unit. So aˉϕ(m)=1ˉ\bar a ^{\phi(m)}=\bar 1.

Corollary 3 (Fermat's Little Theorem)

If p is prime and pap \nmid a, then ap11(p)a^{p-1}\equiv 1(p).

Proof:

Since (a,p)=1(a,p)=1, aϕ(p)1(p)a^{\phi(p)}\equiv 1 (p). Thus ϕ(p)=p1\phi(p)=p-1.