You are on page 1of 5

A self - criticism to my fourth work --

WHERE THERE IS SMOKE , THERE IS SMOKE


An accustomed parody of the american
quantum physicist Wheeler J.(1911-2008)
* * *
N. V. TU YEN
( Vietnam )

At first , I must apologize to my readers - familiar and unfamiliar -


for my exorbitant optimism and overhurry, as regards which I shall
explain myself in the following three paragraphs .
As to the readers , who are interested in the achievement of the
work only , please run your eye over the first two paragraphs , while
to others , who like going into details of this undecided work , do
read and the last one , please .
Furthermore , please remember some formulas and notations
mentioned in [ 7 ] , such as : p1 = min pi contained in each
factorizations formula ; 1/ni with even ni is simply named even unit
fraction , with odd ni named odd unit fraction ; 1/( n1 , n2 , , nm ) =
1/n1+1/n2+ ....+1/nm , all which were noted in [ 7 ] .

1
After what I had written in my last fourth work named Non-
existence of any k- perfect odd numbers [ 7 ] , I set about its
submission to an official referee . For this reason , I knew that my
first unavoidable task will have to change my bet mentioned in the
last paragraph of [ 7 ] into a proved theorem . Concretely , I shall
have to prove a theorem, where the wonderful fraction , denoted
( pi 1 )/ pi has been equaled by the sums of both even and odd
different unit fractions uniquely .

2
During the whole of two months , I tried hard to find this proof ,
but finally I was to recognize the non-uniqueness of a such waiting
theorem . More concretely , while the formulas ( 1 ) and ( 2 ) [ 7 ]
identify the fraction ( pi 1 )/ pi with the sums of both even and odd
different unit fractions only , the trial errors assembling can be
identified this fraction with the sums of all different odd unit
fractions . For example , when pi = 3 , by ( 1 ) and ( 2 ) we get
( pi 1 )/ pi = 2/3 = 1/2 + 1/6 , while by assembling play we have
2/3 = 1/3 + 1/3 = 1/( 5, 9, 45 ) + 1/( 7, 15, 21, 33, 113, 1174635 ) , i. e. we
can get a sum of nine different odd unit fractions entirely .

1
In a word , the equalization of an odd unit fraction when using the
well-known formula ( 1 ) , we obtain the sums of either only even
or both even and odd different unit fractions , but when using the
trial-errors assembling we can get the sums of different odd unit
fractions only . Go into some details , we can see :
1/5 = 1/( 6, 30 ) = 1/( 9, 23, 25, 225, 1035 ) ;
1/7 = 1/( 8, 56 ) = 1/( 9, 49, 441, 111, 16317 ) and so on .
*
Thus , my mistake here consists in a wrong estimation about these
two formulae ( 1 ) and ( 2 ) by considering them as an unique
possibility to expand a given fraction into its equal sums of other
different unit fractions .
By this mistake I had lost my bet noted in the last paragraph of
[ 7 ] , as it were . I think the attentive readers had knew it just on
reading the Wheeler parody of an idiom No smoke without fire
about which I was hopeful .

3
Some lemmas joined up with the self-criticism

Lemma 1 . An even unit fraction 1/n can not identify with any
sums of different odd unit fractions only .
Proof - Let this even unit fraction 1/n be 1/2k with k N . Assume
m m m m 1
1/ 2k = 1 / ni with all odd ni , then we get ni 2k. ( n j )i ,
i 1 i 1 i 1 j 1; j i

where its left side is an odd number , while its right side is an even
number . This is an absurdity , hence the lemma is true .

Lemma 2 ( false ) . An odd unit fraction 1/n can not identify with
any sums of different odd unit fractions only .
Proof - Let this odd unit fraction 1/n be 1/( 2k+1 ) with k N .
m
Suppose 1/( 2k+1 ) = 1 / n
i 1
i ( a ) with all odd ni . Then we have :

m m m 1

ni (2k 1) (
i 1 i 1

j 1; j i
n j )i ( b ) where both two sides of ( b ) are

identically odd numbers . It means , as to the parity, this case is possible.


Now , suppose among all the denominators ni in the equality ( a ) ,
there are both odd and even ni at the same time , then from ( b ) ,
its left side is an even number , while its right side will be an even
too if the number of the odd products of ( m-1 ) factors ni is an even
number , which is evidently possible too . Thus these two cases
considered - either all 1/ni are different odd , or are both even and
odd in the expanded sums - seem to be completely possible .

2
However , a counter-example of the lemma is more persuasive .
Using calculator Casio fx 570 vn plus by a trial errors manner , it
was easy to find , for example : 1/7 = 1/( 9, 49, 111, 441, 16317 ) ;
1/11 = 1/( 13, 121, 197, 1573, 154941, 309881, 4939813021 ) , etc . Here
there are counter-examples so the lemma 2 is false .

Lemma 3 . Let m be the number of distinct divisors ni of a


k-perfect even number npk then m has to compose by an even
number of odd divisors m1 and ( m m1 ) even divisors .
Proof - From the definition of a k- perfect even number npk we
m
have : n
i 1
i k .n pk . Because the right side of the equality is always

an even number , the left side has to be an even too , i. e. it has to


be composed by an even number m1 of the odd divisors evidently .

Lemma 4 . Let m be the number of distinct divisors ni of a


k- perfect odd number npk , then :
1 ) If k is an even number , m has to be an even number too .
2 ) If k is an odd number , m has to be an odd number too .
m
Proof - From the definition of a k- perfect odd number n
i 1
i k .n pk ,

As to 1 ) , because its right side is an even number , its left side


has to be composed of an even m of the odd divisors ni , so that
their sum is an even number too .
As to 2 ) , because its right side is an odd number , its left side has
to be composed of an odd m of the odd divisors ni , so that their
sums is an odd number too .
Geometrically speaking , in the last case 2 ) , there is necessarily an
integral tangent and ( m 1 )/2 integral secants of this k-perfect odd
number npk here ( see [ 8 ] ) .

Lemma 5 . The prime factorizations form of every ( k = 2c )- perfect


odd number has to express by the following formula :
c l
2b j
n pk piai . p j ( A ) , where : pi is an i-th Hilbertian prime,
i 1 j c 1

i. e. in form pi = 4si +1 , and pi > 1 ; ai = 4ti +1 is the power of pi ;


pj is an j-th non-hilbertian odd prime ; 2bj is the power of pj ;
c = k/2 ; l is the number of all primes in the factorizations formula
of the ( k = 2c ) perfect odd number npk ; Here , all Latin letters
with and without dummy variables are natural numbers .
Proof - This lemma, in fact , is a expansion of the delicate Eulers
factorization form of every ( k = 2 ) perfect odd number , that is :
l
ai 2b j
n p 2 p . p j
i ( a) . With this form and the definition of a
j 2

3
diperfect ( i . e . a 2- perfect ) odd number np2 , that is ( n p 2 ) 2.n p 2
l
2b j
it has to satisfy the equality : (n p 2 ) ( pia ). ( p j )
i
( b ).
j 2

From ( b ) , because on its right side , the first factor is always a


product of two subfactors , where the first is always one subproduct
of number 2 and one odd number noi, [ 1 ] , and the second factor
is an odd number undoubtedly . And all that stuff lead to a simple
induction of the lemma in the following manner :
For a diperfect odd number , there is necessarily one hilbertian prime
pi to have a k = 2.1 = 2 after its definition .
For a tetraperfect odd number , there are necessarily two hilbertian
primes pi to have k = 2.2 = 4 after its definition .
For a hexaperfect odd number , there are necessarily three
hilbertian primes to have k = 2.3 = 6 after its definition ; and so on.
And at last , for every ( k = 2c ) perfect odd number , its
factorization formula has to be composed exactly of ( c = k/2 )
hilbertian primes , as affirmed in the lemma .

*
REFERENCE
( continued )

[ 7 ] N.V. TU YEN , Non-existence of any k- perfect odd numbers ,


Posthumous works , 2015 .
[ 8 ] N.V. TU YEN , Towards reaching a portrait of an odd perfect
number , Posthumous works , 2014 .
[ 9 ] N.V. TU YEN , Continuation 1 of [ 8 ] , 2015 .

* * *

4
5

You might also like