April 19th, 2006, 2:56 am
I guess if I take 1 million+1, it does the job. Let me prove this. To make life easier, let us denote 1,000,000 by m. Then what I am going to prove is,the first four digits of (1+m)^m are different.We know, for any m, we have,(1+1/m)^m < e < (1+1/m)^(1+m).Thus, for us,e - (1+1/m)^m < (1+1/m)^(1+m) - (1+1/m)^m = (1+1/m)^m times (1/m) < e times (1/m) < 1/ 100000 (Since we have m = 1 million)Now, m^m (e - (1+1/m)^m) < m^m times e - (1+m)^m < m^m times (1/100000)We know e=2.7182...Hence, the first four digists of (1+m)^m will be 2718 because the error is at least 5 digits away.So 1+m (i.e. 1000001) does the job.I hope I didn't make too many errors here. And I am sorry I am not familar with typing formulas with Latex.