[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference rusure::math

Title:Mathematics at DEC
Moderator:RUSURE::EDP
Created:Mon Feb 03 1986
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:2083
Total number of notes:14613

1993.0. "Real numbers as infinite products" by EVTSG8::ESANU (Au temps pour moi) Fri Sep 01 1995 03:47

Let  (a(n))n  be a sequence of non-null real numbers, convergent to 0.
Prove that for every real number  x  there is a sequence of integers 
(mu(n))n such that

                      oo
		x =  /PI  mu(n) * a(n)
		    n = 0

---

This is the second part of the problem that I published in the American
Mathematical Monthly, Vol. 85, No. 10, December 1978, p.828, problem 6240.

I couldn't find any application to this. (?)

/Mihai.
T.RTitleUserPersonal
Name
DateLines
1993.1HANNAH::OSMANsee HANNAH::IGLOO$:[OSMAN]ERIC.VT240Fri Sep 01 1995 11:111
same questions....
1993.2re. .1EVTSG8::ESANUAu temps pour moiFri Sep 01 1995 12:331
Same answers (see 1992.*).
1993.3SolutionDECADA::YODERMFYThu Sep 07 1995 11:5632
It was a bit trickier to find a simple solution here than for the case of sums.

First, note that for x=0 any series with mu(0)=0 will do, and if we can solve
the problem for x>0, we can change the sign of mu(0) to get a solution for -x. 
So WLOG x>0.  We will always choose nonzero mu(i) with the same sign as a(i); so
WLOG we may also assume all a(i)>0 as well.  Now we can work more easily since
all relevant quantities will be positive.

Next, since a(n)->0, I wish to choose a point k beyond which all a(i)<1/2, and
let a(i)=1 before this point; this reduces the problem to a new one in which x'
is x divided by the product of the initial a(i), and a'(i)=a(i-k).

Let P(n) be product(i in 1..n)mu(i)*a(i), so P(n)=mu(n)*a(n)*P(n-1) for n>0, and
P(-1)=1 by convention.  Define inductively mu(n)=ceiling(x/(a(n)*P(n-1)).  (The
value of mu(n) depends on previous mu(i), but this is well-defined.)  It is
clear that mu(n) is always a positive integer.

We have mu(n) >= x/(a(n)*P(n-1)); multiplying by a(n)*P(n-1) and using the
definition of P(n) gives us P(n)>=x for all n>=0.  Similarly,

  mu(n) < 1 + x/(a(n)*P(n-1)), so

  P(n) < a(n)*P(n-1) + x = P(n)/mu(n) + x.

Using the first part, P(n) < a(n)*P(n-1) + x < P(n-1)/2 + x, using a(n)<1/2; by
induction this gives us P(n)<2(x+1) for n>=-1, so the P(n) are bounded above.

Using the second part, P(n) < x/(1-1/mu(n)) when mu(n)/=1.  From the definition
of mu(n) and from having the P(n) bounded and a(n)->0, we get mu(n)->infinity. 
So 1-1/mu(n)->1 and P(n) is sandwiched between x and a value that approaches x. 
So P(n)->x.

1993.4typo in .3DECADA::YODERMFYThu Sep 07 1995 14:383
Next, since a(n)->0, I wish to choose a point k beyond which all a(i)<1/2, and
let mu(i)=1 before this point...
     ^ I wrote a(i)
1993.5re. .3: Applications?EVTSG8::ESANUAu temps pour moiMon Sep 11 1995 06:176
Mike,

Do you see any applications to this?

Thanks,
Mihai.