1992 BMO1 Q5


By Tim Norman on Wednesday, October 17, 2001 - 11:28 am:

a1 , a2 , a3 , ..., an , ... are positive integers such that for all n1, an+1 > an and aa n=3n

i) Find a1 to a9

ii) Find a100

I have got a solution but I am not happy with it (though it is correct), I may post it later but now I have to go to lessons!


By Tim Norman on Wednesday, October 17, 2001 - 02:40 pm:

Hello again.

Here is my solution:

Try a few a1 s:

If a1 =n where n> 3 then aa 1=an =3 while a1 =n> 3 hence a1 > an , but a1 should be always less than an , so n£ 3

If a1 =3 then aa 1=a3 =3 x 1=3, and you get a1 =3 and a3 =3, since a1 should be less than a3 , so n is not 3

If a1 =1 then aa 1=a1 =3 x 1=3, so you have a1 =1 and a1 =3 which is impossible, so n is not 1

It turns out that a1 =2 and you can then work out a1 to a9 except a4 and a5 . But since a3 =6 and a6 =8 and an are all integers, so a4 and a5 must equal to 7 and 8 respectively.

---

For the second part,

I worked out:

a11 =20, then aa 11=a20 =11*3=33, a33 =20*3=60, a60 =99, therefore a99 =180;

a12 =21, then a21 =36, a36 =63, a63 =108, a108 =189

there are 8 integers between 180 and 189 (exclusive) and 8 terms of a in this range, so a100 =181


Is there a better way of doing it? (It took me ages to get there!)

Thanks.


By Kerwin Hui on Wednesday, October 17, 2001 - 03:32 pm:

This question is very similar to 1992 BMO1 question 5. First, let me write an as a_(n). You can establish the value of a_(3n ) and a_(2*3n ) for all n (inductively), which is not difficult. Now, apply a simple argument to show a_(3n +k)=a_(3n )+k and correspondingly a_(2*3n +k)=3n+1 +3k. Hence you can work out the value of an with ease. [the BMO question asked for a_(1992)]

Kerwin