DEMONSTRATE MATHEMAT1CA

Vol. XI No 1 1978
D . 4 i c t t • d to P r a f • • • o r
Stofaa Stiuxevlci

Henryk Iwaniec

ON THE PROBLEM OF JACOBSTHAL

1. I n t r o d u c t i o n
J a c o b s t h a l ' s problem i s to e s t i m a t e f o r a given r the
maximal l e n g t h C ( r ) of a sequence of c o n s e c u t i v e i n t e g e r s
ee3h d i v i s i b l e by one of r a r b i t r a r i l y chosen primes ( f o r
r e f e r e n c e s and h i s t o r y of the problem see [ l ] ) .
The f u n c t i o n C ( r ) can be considered by the s i e v e method.
I f A s t a n d s f o r a sequence of X c o n s e c u t i v e i n t e g e r s and
Q i s t h e product of r primes q - p ^ . ^ j , t h e n the s i f t i n g
function

S(«A,Q) =
) 1
a eA
(a,Q)=1
counts t h e s e elements of A which a r e coprime t o Q. Thus
the problem i s how l a r g e X and r must be i n o r d e r to have

SW,Q) > 0 .

I t i s w e l l known due t o J u r k a t and R i c h e r t [4] t h a t the s i e -

v i n g l i m i t of the l i n e a r s i e v e (being occured h e r e ) e q u a l s 2 ,
eo i t e a s i l y f o l l o w s

C(r) < c ( t ) r 2 + £

f o r a l l e > 0 . Of course by the s i e v e method the exponent 2

cannot be r e d u c e d , however some small improvements a r e s t i l l
p o s s i b l e i f the e r r o r term i n the s i e v e problem i s a l l o w e d .
- 225 -

Unauthenticated
Download Date | 1/11/18 1:51 AM
2 H.Iwaniec

For a given r > 1 l e t C 0 ( r ) be the maximal length of

the sequence of consecutive integers each d i v i s i b l e by one
of the f i r s t r primes. Then the results of [4] imply

C Q (r)<g r 2 expflog r ) 1 3 / U

while in [ 3 ] i t i s proved

(1 ) C0(r) r 2 log 2 r .
p
Jacobsthal asked whether C ( r ) = c 0 ( r ) an<^ whether C ( r ) r .
The aim of this paper i s to prove (1) f o r C ( r ) . Modify-
ing the arguments used in [3] we shall show that s l i g h t l y
more i s true.
T h e o r e m . There exists an absolute constant c > 0
such that f o r a r b i t r a r i l y chosen primes q 1 , . . . q p , r > 1 each
i n t e r v a l of the length

2
contains at least r integer numbers coprime to q-j...qr.
C o r o l l a r y . We have

C(r) r 2 log2 r .

2. The shifted sieve

The idea of the shifted sieve was inspired by the work of
Halberstam and Richert [2] .
Let i be a f i n i t e sequence of integers, Q a positive
square-free number and

SW,Q) = ) 1.
aeA
(a, <0=1

- 226 -

Unauthenticated
Download Date | 1/11/18 1:51 AM
 On the problem of Jacobsthal 3

For a l l d | Q l e t us introduce quantities

a €cii
as o (mod d)

and assume that they s a t i s f y the inequality

(R) |p d | " fTdll ^ A

where the constants A, B, X as w e l l as the multiplicative
function f ( d ) are > 1.
Let there be given another square-free number P and a
multiplicative function g ( n ) 1 f o r a l l n | P. Suppose
that Q and P have the same number of divisors and l e t
1 be one-one multiplicative correspondence between them.
Finally suppose that f o r a l l n I P, d I Q, n = 1(d) we have

(2) g(n) < f(d).

We are now in position to formulate

Lemma 1 , I f the sequence { ^ - n } n ) p of r e a l num-
bers s a t i s f i e s one of the following conditions

(+) 6m = ^n^XZ fOP 811 m1

n|m n|m

(-) = ^n^XZ, iU(n) f°r a11 m ' P

n|a n|m
then

(3) (i-iHr)IZrrra5P7*4Ci*J,,<">
q|Q n |P pm nIP

- 227 -

Unauthenticated
Download Date | 1/11/18 1:51 AM
4 H.Iwaniec

or

u> ^ « » i f l (1 - m j z n («¿Vd - 4
E ki » r W

qlQ 11IP p l n n|P

according as (+) or (-) holds.

P r o o f . We need only lower bound for the sifting
function, so we shall show here only (4). We have

8W.Q ) > Y Z \ h[6)

&eA Idla, dlQ

- E ^ K d ) Mdl > * C t t j t - "CK(d)l

dlQ d|Q d|Q

= X
y y^gl(d) ^ /"(k) II I BR r ( n J -
¿dlQ
_ 1 T d 7 2dkl QilkT - nIP nl
I

1
O - ^ C t t t ^ T T - l i ^ ^ -
qla
q|Q d|Q n|P

Prom (2 j we have f ( q ) ^ g(p), where p = l(qj. Hence

g g
V^ l(d) ^ l(d) .gn
n (f(q )-i) n (6(i(qj)-i > n (g(p)-u

dIQ q l d d|Q q|d

n|P P|n

This completes the proof of (4).

3. The proof of the theorem

Let A be a sequence of X consecutive integers. Then
the number ^ ¿ l of elements from A which are divisible
by d is approximately equal to X/d, precisely we have

- 228 -

Unauthenticated
Download Date | 1/11/18 1:51 AM
 On the problem of Jacobsthal 5

\*à\ - f | < 1 '

Hence condition (R) is satisfied with A = 1, B = 1 and
f(d) = d.
p
Let z > 2, y > z and P be the product of all pri-
mes p < z,

p.[-|P.
P<Z

It can easily be proved that the sequence [^-n]n|p<

for 2 1 < u
ju.(n) if n = P r -.P U . P 1 >...>P U . P 1 -..P 2 1 < y Pgi

0 otherwise

satisfies (-). Por n > y we have ^ = 0, henoe we have

a trivial estimate of the remainder term

< 7
X > n l
n|P
but it is too weak to prove (1). We quote from [3] two esti-
mates.
L e m m a 2. Por 4 < z 2 < y < z^ we have

(5) |xnl < y d o g y)" 2 .

n |P

1o
(6) r _ 2 et S(s-1) + 0 ( 1 ^
J X I P-1 ) s Vlog y ) '
pl
n|P

where s = log y/log z and j = 0,577... is the Buler con-

stant.
The proof is very complicated and utilizes some differen-
tial equations with shifted arguments. These results seem to
be best possible. Note that in the Selberg upper bound me-

- 229 -

Unauthenticated
Download Date | 1/11/18 1:51 AM
6 H.Iwaniec

thod the remainder term can also be estimated by the value of

order y/log y (see [ 5 ] ) .
Now we have everything to prove the theorem.
For a r b i t r a r i l y given primes q^ < . , . < q , r > 1 l e t us
put
<3 = • • '<) r i z = Pr, = Pit i = 1,...,r,

where p^ stands f o r the i - t h prime number. Then the fun-

ctions g(n) = n for n | P and f ( d ) = d f o r d I Q s a -
t i s f y ( 2 ) . Thus from Lemmas 1 and 2 we obtain

(7) SU,Q)>X |~| (1 - 1 )

j log(s—1 )

+ c (nn)l
If C i s a s u f f i c i e n t l y large absolute constant and

,-1
y = C z2, X = p| (l - J ) y/log z
q|Q
o p
then the r i g h t hand side of (7) i s > y/log z > r . This com-
pletes the proof of the theorem.
Added in the proofs. Recently R.C. Vaughan [6J derived
from [3] the estimate C(r) r2log4r.

REFERENCES

[ 1 ] P. E r d 6 s : On the integers r e l a t i v e l y prime to n

and on a number-theoretic function considered by Jacob-
s t h a l , Math. Scand. 11(1962) 163 - 170.
[ 2 ] H. H a l b e r s t a m , H.-E. R i c h e r t : Mean
value theorems f o r a c l a s s of arithmetic functions, Acta
Arith. 18(1971) 243-256.

- 230 -

Unauthenticated
Download Date | 1/11/18 1:51 AM
 On the problem of Jacobsthal 7

[3] H. I w a n i e c j On the error term in the linear

sieve, Acta Arith. 19(1971) 1-30.
[4] W.B. J a r k a t , H.E. R i c h e r t : An improve-
ment of Selberg's sieve method. I, Acta Arith. 16(1969)
207-216.
[5] A. S e 1 b e r g : Sieve methods, Proc. Sympos. Pure
Math. 20(1971) 311-351.
[6] R.C. V a i j g h a n : On the order of magnitude of
Jacobsthal s function, Proc. Edinburgh Math. Soc., 20
(1976-77) 329-331.

INSTITUTE OP MATHEMATICS, POLISH ACADEMY OP SCIENCES

Received October 1st, 1977.

- 231 -

Unauthenticated
Download Date | 1/11/18 1:51 AM
 Unauthenticated
Download Date | 1/11/18 1:51 AM