You are on page 1of 5

Dutch mathematician T.J.

Stieltjess claim to have solved Riemanns hypothesis has


received attention in the scholarly literature , especially after the 1985 proof of the falsity of the
Mertens conjecture, a stronger version of Stieltjes 1885 claim to have proven, in todays
notation, that ()/

is bounded. [Borwein, 2009, 69] Number theorists today are skeptical of


Stieltjes claim that he had a proof of an equivalent of Riemanns hypothesis, and with good
reason. And yet, revisiting his attempt at proof, something that has never been done, can yield
insights on the history of understanding the problem. His letters offer a little studied resource on
this first attempt to solve Riemanns conjecture, an attempt that, while lacking rigor by
contemporary standards, still has not been entirely rejected. Studying Stieltjes letters can still
shed light,not only on the assumptions of his 19
th
century mathematical contemporaries but also
on those assumptions we have inherited from them.
Reconstruction of Stieltjes proof of RH:
Definition 1.1 The Liouville function is defined by
() = ()

,
where is the number of, not necessarily distinct, prime factors of .
Theorem 1.2 The Riemann hypothesis is equivalent to the statement that
for every fixed > 0,

() + () ++()

+
;
=
This translates to the following statements: The Riemann hypothesis is equivalent to the
statement that an integer has equal probability of having an odd number of or an even
number of distinct prime factors (in the precise sense given above). The sequence
{()}
=

= {, , , , }
behaves more or less like a random sequence of s and -s in that the difference between
the number of s and -s is not much larger than the square root of the number of terms.
(Borwein, 2008, 6)
We will use the equivalent statement The Riemann hypothesis is equivalent to
() +() ++ ()

+

(Borwein, 2008, 46)
1


Cauchys Cours dAnalyse (1827) ; Note II, Theorem XV: Let ,

, be any real
quantities. If these quantities are not equal to each other, then the numerical value of the sum

+

+
is less than the product

2
+

2
+

2
+
so that we have


val.num. ( +

+) <

2
+

2
+

2
+

[Cauchy, 2009, 302]
The last part is equivalent to | +

+ | <

2
+

2
+

2
+

Let = (1)

, k being the number of prime factors, either distinctive, for the Mbius
function () or not, for the Liouville function ().
|
1
+
2
+ +

1
2
+
2
2
+ +

2


1
if we interpret the statement in terms of the frequencies, this is just the statement that the sequences of the
averages of the partial sums have the limit 1/2" (italics in the original). [Varadajaran 126 Cesaro. ] Jahnke 171 says
its true!. Euler Ayoub has lambda

From the above we have that substitution gives us
|
1
+
2
++

(1
1
)
2
+(1
2
)
2
+ + (1

)
2


|
1
+
2
+ +

| <

(1
1
)
2
+ (1
2
)
2
++(1

)
2

And if we let

or

we are done since absolute convergence implies convergence and



=1
1

=1


=1
(0)

=1

1
2

QED
2


More rigorously (?),we recall the AM-QM inequality (Nahin, 334), generalized in Jensens
inequality for any real
1
,
2
, ,

1
+
2
++

1
2
+
2
2
++




2
Note that Cauchy gives us that
|
1
+
2
+ +

= (

2
,

2
,

2
)
= ((1)
2
, (1)
2
, (1)
2
)

=1

as Stieltjes claimed.

with equality iff
1
=
2
= , =

. Now, multiplying by we get



1
+
2
++

1
2
+
2
2
+ +



Following Cauchys Course we have

1
+
2
+ +

1
2
+
2
2
++

2


So now we can prove that the following is a constant.

1
2
+
2
2
+ +

2
=

Let = (1)

. From the above we have that substitution gives us


1
+
2
+ +

(1)
2
+ (1)
2
+(1)
2
+

We then have that

1
+
2
+ +

<

(1
1
)
2
+(1
2
)
2
++ (1

)
2


1
+
2
++

<

(0)

1
+
2
++

<

1
2


And if we let =

or
1
we are done.

QED.

Version 2: Recall the quadratic mean, or root mean square.

1

=1

=1


Now, let

=1
be the partial sums of the Mbius function and multiply both sides by n.
1

()
1

2
()

=1

=1

()
1

2
()

=1

=1

() (
2
)
1

2
()

=1

=1

()
2
()

=1

=1

()

=1

2
()

=1


()

=1
|()|

=1

()

=1
1

=1


Via zeta regularization (or Grandis series /Diracs comb , Eulers characteristic, (0) =
1
2
)
()

=1

1
2

QED

Note: For (0) =
1
2
see also (Kaneko 2003) and Eulers Beau rapport
http://eulerarchive.maa.org/pages/E352.html

Juan Marin

You might also like