Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!bloom-beacon!gatech!pyr!jkr
From: jkr@pyr.gatech.EDU (J. Kenneth Riviere (JoKeR))
Newsgroups: comp.lang.pascal
Subject: Re: Perfect number program
Message-ID: <4166@pyr.gatech.EDU>
Date: Wed, 7-Oct-87 12:33:15 EDT
Article-I.D.: pyr.4166
Posted: Wed Oct  7 12:33:15 1987
Date-Received: Sat, 10-Oct-87 09:52:03 EDT
References: <9652@brl-adm.ARPA>
Reply-To: jkr@pyr.UUCP (J. Kenneth Riviere (JoKeR))
Organization: Georgia Institute of Technology
Lines: 25

In article <9652@brl-adm.ARPA> 11SSTEIN%GALLUA.BITNET@wiscvm.wisc.EDU writes:
> I am working on a perfect number program and I have faced some problems.
>     I tried mine and it takes TOO long. I used this program:
>
>    FOR I=2 TO 10000 DO
>    W=0 
>    FOR K=2 TO I DO 
>    IF I/K = TRUNC (I/K) THEN W=W+(I/K) 
>  IF W=I THEN PRINT W "IS A PERFECT NUMBER"
>

First of all, when you discover a divisor of I you have also discovered
another divisor of I (when you find that 2 is a divisor of 28 it is clear
that 28/2 means that 14 is also a divisor of 28).  Thus you should be able
to increment W by both K and I/K when your test is true.  Furthermore,
since you are checking both K and I/K it is not necessary for your inner
loop to progress past the square root of I since you will only rediscover
the same divisors from the other side (you would eventually find 14 again,
but with my approach you shouldn't need to loop past 5 to find divisors
for 28).  Hope this helps.


J. Kenneth Riviere    (JoKeR)    ISA, Georgia Tech, Atlanta, GA  30332
Internet:  jkr@pyr.gatech.edu                  Bitnet:  iadt1kr@gitvm1
uucp: ...!{decvax,linus,rutgers,ihnp4,hplabs,seismo}!gatech!gitpyr!jkr