This is a math problem published in a periodical in 1877.
Find the least integral values of x and y that will satisfy the equation:
x^2 - 9781*y^2 = 1.
I thought the least integral roots are (1, 0) and (-1, 0).
Also I noticed the possible x should be around 99*y, so I manually caculated about 50 pairs, no luck.
Then I wrote a program to look through the vicinity at (99*y - 10 < x < 99*y + 10, y) with y values going up to 10^9. I have a few lines of crap print out because the computer was messed up when the number is too big. Like these:
The roots are ---
x = -1, y = 0
The roots are ---
x = 1, y = 0
The roots are ---
x = -1177971319, y = 161635332
The roots are ---
x = 2125380257, y = 542070584
The roots are ---
x = 879440713, y = 572868844
The roots are ---
x = -1134323873, y = 856212344
Then I thought at that time (1877), pleople just did not have computers. Then I try to prove it.
Then I spent about 20 minutes to prove that there is no other integral roots using some dumb basic methods. By checking last digit of 9781*y^2 + 1, I rule out 9, 1, 4, 6 as the last digit of y. If the last digit of y is 7 or 3, then 9781*y^2 + 1 will end with 10, 30, 50, 70 or 90. If the last digit of y is 5, then last 2 digits of 9781*y^2 +1 are 26. No integer has square value ending in 26. If the last digit of y is 8 or 2, then y must be form of 50k + 2 or 50k -2, and x will take form 10m+5, then we will have m(m+1)=9781k*(25k+1), which cannot be true. If the last digit of y is 0, then x must take form of 50k +1, and y is take the form of 10p, we will have 2500k^2+100k +1 = 978100m^2. This cannot be true. So this Diophantine equation has only roots (1, 0) and (-1, 0).
Can any one have a better solution to prove this? Thanks.