This paper explores the irreducibility of polynomials derived from prime numbers' k-adic expansions, extending Murty's theorem by considering scaled primes and analyzing polynomial factors.
Contribution
It proves a stronger version of Murty's theorem for scaled primes and introduces techniques to analyze polynomial factors for larger multipliers.
Findings
01
Proved irreducibility for scaled primes with multiplier less than k
02
Developed methods to control polynomial factors with larger multipliers
03
Extended existing theorems on prime-based polynomials
Abstract
In 2002, M.Ram Murty showed that if p is a prime with k-adic expansion :p=∑i=0naiki , then the polynomial f(x)=∑i=0naixi is irreducible in Z[x].When k=10 , it's a result of A.Cohn. I think this kind of polynomials is really interesting and worse to speak more. So I plan to find more conclusions about this kind of polynomials. In the first section of this article, author proves a stronger version of this theorem that if we multiply prime p by a factor t that is smaller than k ,the conclusion also holds. In the second section, author further consider larger multiplier t ,and gives a technique to control one of the factors of the polynomial.
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsMeromorphic and Entire Functions · advanced mathematical theories · Analytic Number Theory Research
Full text
Primes and Irreducible Polynomials
Boyang Zhao
Abstract.
In 2002, M.Ram Murty showed that if p is a prime with k-adic expansion :p=∑i=0naiki , then the polynomial f(x)=∑i=0naixi is irreducible in Z[x].[1]When k=10 , it’s a result of A.Cohn.[2]When I was in Nanjing University,I have proved this in my own way(Since I did not publish that article, I cannot cite it in a formal way[3]).
In the first section of this article,author proves a stronger version of this theorem that if we multiply prime p by a factor t that is smaller than k ,the conclusion also holds.In the second section, author further consider larger multiplier t ,and gives a technique to control one of the factors of the polynomial.
1. Basic Proof of the Theorem
Definition 1.1**.**
[TABLE]
, where 0⩽ai⩽k−1,ai∈Z,fk,m(k)=m,m>0
It’s obvious that ai are the coefficients of m’s k-adic expansion.So it can be named as the polynomial generated by k-adic expansion of m
Lemma 1.2**.**
If z is a non-zero root of fk,m, it has the following properties:
[TABLE]
Proof.
Assume fk,m=xt∑i=0naixi, because we just consider non-zero roots, W.L.O.G, t=0,fk,m=∑i=1naixi,a0=0
[TABLE]
If ∣z∣<k1,f(z)=0,then
[TABLE]
Contradict. Thus ∣z∣>k1
Similarly, If ∣z∣>k,f(z)=0⇒
[TABLE]
Contradict, thus we have proved (i)
[TABLE]
If Re(z)>0, then 1/z=zˉ/∣z∣2⇒Re(1/z)=∣z∣2Re(z)>0
If ∣z∣⩾21+4k−3, we have:
[TABLE]
Because an⩾1,an−1⩾0Re(z1)>0 and ∣z∣⩾21+4k−3⇒∣z∣>1
[TABLE]
Contradict, thus we have proved (ii)
[TABLE]
If Re(z)>k>0, by (ii) we can know that ∣z∣<21+4k−3
We have the following conclusions:
[TABLE]
Let k(x,y)=4x3−3xy2∂x∂k(x,y)=12x2−3y2∂y∂k(x,y)=−6xy
If k⩾4 then any factor of g is bigger than k when x=k
⇒t≥g(k)>k ,contradict.
When k=3,
3−ai>3,(3−3)2>1.6>1,(1.6)2=2.56>3−1=k−1≥t
So g has no factors of degree 1 and g has at most 1 functor of degree 2
Let g=b2x2+b1x+b0,with root z and z. 1⩽b2⩽2,1⩽b0⩽2,,
Then 2=g(3)=b2(3−z)(3−z)>b2(3−∣z∣)2=b2(3−b2b0)2
So there must be b2=1 and b0=2 or 2>b2(3−b2b0)2>2 contradict.
g(3)=9+3b1+2=2⇒b1=−3⇒g(x)=x2−3x+2
However g(1)=g(2)=0⇒fk,tp(1)=0 impossible.
In conclusion, ∀0<t⩽k−1,fk,tp is irreducible.
∎
Corollary 1.5**.**
If fk,m reducible, g∣fk,m and g(k)=n with highest order coefficient a¿o
Then
[TABLE]
Proof.
[TABLE]
Thus degg<degk−kan
∎
Corollary 1.6**.**
p,q* are primes, ∀k, if fk,pq reducible, ∃ irreducible polynomial g and h such that g(k)=p,h(k)=q*
Proof.
If ∃s(x)∣fk,pq,s(k)=1 .
Then
[TABLE]
Thus if fk,pq reducible, fk,pq=gh,g>1,h>1
\Rightarrow\left\{\begin{array}[]{l}g(k)=p\\
h(k)=q\end{array}\right. or \left\{\begin{array}[]{l}g(k)=q\\
h(k)=p\end{array}\right.
∎
2. Conditions when t is bigger than k
By corollary 2, we can know that if fk.pq reducible, what are its irreducible factors values when x=k .And by corollary 1 we can control its degree. So we can found out what one of its factor looks like. I will give some examples to show how to estimating the degree of the factor that values smaller.
Theorem 2.1**.**
When k⩾4,k⩽t<2k,(p,k)=1, fk,tp is reducible ⇔ one of its factors is x+t−k.
Proof.
Let fk,+p=gh, p∣h. If h(k)>p⇒h(k)≥2p⇒g(k)<k. By proof of Theorem 1.4 we can know that g(k)>k, controdict.
Hence h(k)=p,g(k)=t.
When k>5 ,by Corollary1.5 ,it’s easily to know that degg<2
Let g=ax+b , Because all real root of fk,tp not bigger than 0⇒−ab≤0
Thus a,b>0,g(k)=ak+b=t<2k⇒a<2⇒a=1,b=t−k.
When k=4,5,degg<3
Let g=ax2+bx+c.
Because g(k)≡c≡t(modk) , k<t<2k,c≥0⇒t−c=k
Thus:
[TABLE]
If Δ≥0 , g has roots that is real and bigger than 0 , impossible.
If g=x2−3x+3 , assume h=∑i=0mcmxi
[TABLE]
If c1=1 ,then by induce , if ∀i<s,ci=1 ,for i = s:
[TABLE]
Thus h(x)=∑i=0nxi
However , an−1=1∗1−3∗1<0 ,contradict.
If c1=2 , a2=c2∗3−2∗3+1⇒a2=2 , a3=c3∗3−2∗3+2⇒c3=2
By induction , if ∀1<i<s , ci=2 then when i=s ,we have:
as=3∗cs−3∗2+2⇒cs=2
However , an−1=2∗1−2∗3=−4 ,contradict.
In conclusion , deg g = 1, g=x+t−k.
Next we need to prove h is irreducible.
If h(x)=p(x)∗q(x) W.L.O.G p(k)=1,q(k)=P.
∀ root z of p(x), it is also a root of f(x)⇒Re(z)<k.
Then p(k)>p(k−1)⇒p(k−1)≤0 .That’s impossible. So h is irreducible.
Thus fk,tp reducible ⇔∃g=x+t−k, h irreducible, fk,tp=gh.
∎
It is really hard when k=2,3. But we can consider f3,4p,f3,5p,f2,3p ,during the proof of these three conditions,except using Corollary1.5 ,we also use the technique of predict coefficients of g(x) to make contradiction and to use the fact that g(k)>g(k−[k−21]−1)
Proposition 2.2**.**
f3,4p* reducible if and only if x+1∣f3,4p.*
Proof.
When t=4,(3−3)6>4⇒ deg g<6 and 5 is odd, g(3)>3×(33)4>4
g has a real root smaller than 0. ⇒g(3)>3×(33)2=4.8>4 contradict.
If degg=2,
g(3)=4⇒g can’t have two real roots,or g(3)>32=9>4 contradict.
Let g(x)=ax2+bx+c1⩽a⩽2,1≤c⩽2
g(3)=4≡1(mod3)⇒c=1
So there are only two types of g,g=x2−2x+1 or g=2x2−5x+1,however,all of these polynomials has positive real roots.They can’t be factors of f3,4p
And we also hold the conclusion that h(x) is irreducible.
So f3,4p reducible if and only if x+1∣f3,4p.
∎
Proposition 2.3**.**
f3,5p* is reducible ⇔x+2∣f3,5p*
This proposition looks like the former one at first.Indeed they are truely similar at the first step,however ,in the last step ,we have to face the factor x2−2x+2 which its roots meet the all the properties that fk.m has.It need other method to do with this factor.
Proof.
[TABLE]
Let g=∑0nbixi,it obvious that 0<bn<3 and b0=2.
If deg g=6:
Assumeg=a∏i=13(x−zi)(x−zˉi)
∣z1∣2∣z2∣2∣z3∣2⩽2⇒∃is.t∣zi∣<2
Thusg(3)>(3−3)4(3−2)2>5 ,impossible.
If deg g=4:
Let g=a(x2−2Re2(z1)x−∣z1∣2)(x2−2Re(z2)x−∣z2∣2)
Since we have known all possible pairs of (ci,ci−1) ,and the only pair that (cm,cm−1) can be taken to make an−1≥0 is (1,2).
So from the tree ,we can know that the possible sequences of ci are all in this form:{12⋯2l110⋯0l2⋯12⋯2lt1} .It is a chain that starts from (1,2) and ends with (2,1) in the tree.
There must be even number of 1s.
So h(3)≡numberof1s≡0(mod2),2∣n
∎
In conclusion, f3,5p reducible if and only if x+2∣f3,5p
∎
Proposition 2.5**.**
f2,3p* is reducible if and only if x+2∣f2,3p or x2−x+1∣f2,3p*
When k=2, there’s a problem that for other k,k−k>1 However, 2−2<1, so we can’t control the degree of g by previous ways.
The good news is that, g(2)>g(1), so if g(2)=n, then g(2)g(1)≤n1 can be a good way to control degree of g.
Now , we got following important constants to estimate values of factors of g(1)g(2) by the range of roots:
[TABLE]
Their roles are similar to the role of k−k in Corollary 1.5.Because in all conditions , factors of g(1)g(2) are bigger than 1.So we can control degree of g by these constants.
Thus, there’s the estimate of degree of g :
[TABLE]
If degg = 7, then n=1,m=3,⇒∣a∣∣z1∣2∣z2∣2∣z3∣2=1
By lemma1 ,∣a∣>21⇒∣z1∣∣z2∣∣z3∣<2,∃js.t∣zj∣<2, so regardless what Re(zj) is , s(zj)>2
Then g(1)g(2)>23×2×(2,17)2>3,controdict.
If degg=6,then ∣a1∣∣a2∣∣z1∣2∣z2∣2=1
∣a1∣,∣a2∣>21⇒∣z1∣2∣z2∣2<4⇒∃∣zi∣<2
Thus g(1)g(2)>23×2×1.217×34>3 controdict.
If degg=5,then ∣a∣∣z1∣2∣z2∣2=1⇒∣z1∥z2∣<2⇒∃∣zi∣<2
Thus 9(1)g(2)>23×2×1.217>3 controdict.
If degg=4 :
∣z1∣2∣z2∣2=1⇒∣z1∣∣z2∣=1 WLOG ∣z1∣<1.
Then if ∣z2∣⩽2⇒g(1)g(2)>2×49>3.Impossible.
If ∣z2∣>2⇒∣z1∣<21⇒s(z1)>23+2229+22>2.5
Here s(zj)>1+1+∣zj∣2−2Re(zj)3−2Re(zj), the bigger ∣zj∣ is and the smaller Re(zj) is ,the smaller s(zj) is. So s(zj)>s(−21)=23+2229+22>2.5
Then g(1)g(2)>2.5×1.2=3. Controdict.
If degg=3.
Case 1. ∣a1∣∣a2∥a3∣=1,a1,a2,a3<0
Let g=x3+ax2+bx+1
[TABLE]
[TABLE]
Then g(−21)=−81+4a−a−23+1<−21−43a<0,however, g(0)=1>0
So ∃ a real root ai s.t ∣ai∣<21, by lemma1 ∣ai∣>21 contradict.
Case 2: ∣a1∣∣z∣2=1
If ∣a1∣<1,∣a1∣>21⇒∣z∣2<2⇒∣z∣<2⇒g(1)g(2)>23⋅2=3 contradict.
Then ∣a1∣>1⇒∣z∣<1 and ∣Re(z)∣<1
Because g(2)=3 ,then we can assume g(x)=x2+ax2−(3+2a)x+1
If a⩾−1⇒g(2)=3>0,g(1)=−1−a≤0 .There is a real root bigger than 0 ,impossible.
If a<−1,g(−21)=−81+4a+23+a+1=25+45a−81 , because a≤−2 , g(−21)<0
So ∣a∣<21 , impossible.
If degg=2,g=x2+ax+1,g(2)=3⇒g=x2−x+1
If degg=1,g=x+1
Indeed f2,3×3=x3+1=(x2−x+1)(x+1),so f2,3p can have these two factors.
In conclusion, f2,3p reducible ⇔x+1∣f2,3p or x2−x+1∣f2,3p.
⇔f2,3p(−1)=0 or f2,3p(21+3i)=0
∎
Remark 2.6**.**
At last,we have done all conditions of fk,tp when t<2k.The conclusion that ′if x2−2x+2∣f3,5l ,then 2∣l′ is really interesting.The tree is generated only depend on some initial coefficient of h(x) and the constant term of g(x)!
I hope the way of proving Propositon2.5 could be improved,because while t becomes bigger,the condition gets worse.Maybe ′Trees′ could help to prove some polynomials can’t be a factor of f2,tp.
I also have some guess that I couldn’t prove now.For example, I think if g(x)∣fk,n.and g(k)=p for some primes , then g(k)∣fk,pα for some integer α. Then this kind of polynomials could be a new way to represent prime numbers.
ACKNOWLEDGEMENTS.
I would like to thank Xuejun Guo who guided my former thesis and let me know this interesting question and Stonybrook University where provide my a good environment to consider this question.
Bibliography3
The reference list from the paper itself. Each links out to its DOI / PubMed record.
1[1] Murty, M. Ram. ”Prime numbers and irreducible polynomials.” The American mathematical monthly 109.5 (2002): 452-458.
2[2] Brillhart, John, Michael Filaseta, and Andrew Odlyzko. ”On an irreducibility theorem of A. Cohn.” Canadian Journal of Mathematics 33.5 (1981): 1055-1059.
3[3] Boyang Zhao, Prime numbers and Irreducible polynomials of Integral Coefficient,graduation thesis of Nanjing University,2021