2008年2月13日 星期三

Elementary Proof of "Sum Of Reciprocals Of All Primes Are Divergent"

兩個月前有兩位朋友講關於證明 "Sum Of Reciprocals Of All Primes Are Divergent"(若設 pk 為第 k 個質數,"Sum Of Reciprocals Of All Primes"就是 1 / p1 + 1 / p2 + 1 / p3 + ...)。我向他們說我有這個定理的elementary proof。昨日終於找到了。

中學生可能不太明白我在講甚麼,所以先解一解題。有一些series,如 1 + 1/2 + 1/4 + 1/8 + ...,可以知道這個series的和是有限的,而且upper bound是2。而且也可以看到它與 2 是越來越接近的,亦即是數學上所講的趨向 2。這類與某個數越來越接近的series,我們叫它convergent series。

但另一些series,如 1 + 2 + 3 + ...,可以見到當項數越來越多時,它的值會越來越大,而且無upper bound,這類series我們叫它divergent series。

那麼, 1 + 1/2 + 1/3 + 1/4 + ... 是convergent series還是divergent series呢?答案是divergent的,證明的idea就是證明它是無upper bound的,有興趣者可以試一試寫出證明。順帶一提,這個series是有名堂的,叫harmonic series。

1 / p1 + 1 / p2 + 1 / p3 + ...也是divergent的。這並不明顯,因為質數比整數疏得多,也就是說 1 / p1 + 1 / p2 + 1 / p3 + ... 比 1 + 1/2 + 1/3 + 1/4 + ... 少了很多項。現在寫出證明於PDF檔內(英文)。

順帶賣賣廣告,以上的PDF檔是用OpenOffice.org的Writer弄出來的。看起來還不錯吧? ^.^

6 則留言:

Kahoo 提到...

可否解釋一下 Σ(1/a_n) < Σ(1/p_k + 1/p_{k+1} + ...)^n 這一步?

Marco_Dick 提到...

因為a_n的質因數分解中的所有質數都在 {p_k,p_{k+1},...}內。

若a_n可以分解成r個質數之積,則在R.H.S. when n = r 時,可以從構造(1/p_k + 1/p_{k+1} + ...)^n 的 expansion中找到 1 / a_n。故L.H.S.內的每一個term都會在R.H.S.出現。

koopa koo 提到...

Standard reference for your proof is in Apostal's Introduction to Analytic Number Theory, chapter 1. Another proof, which is longer but less tricky uses Riemann Zeta function, and it's Euler product representation.

Marco_Dick 提到...

Hi Koopa. I read this proof from a piece of A4 paper I printed long time ago. Thanks for pointing out the standard reference.

May you also point out any reference for the proof using Euler Product Representation?

koopa koo 提到...

If I remember correctly, it's in a book called: From Fermat to Minkowski. Lectures on the Theory of Numbers and Its Historical Development by Opolka. You can also find it as an exercise from my notes on ENT.

Singmay 提到...

You can find some more proofs here

http://en.wikipedia.org/wiki/Proof_that_the_sum_of_the_reciprocals_of_the_primes_diverges

I think the proof that Koopa had mentioned is probably the first one.