Revision history [back]

Thanks for the quick answer. I extended the precision

R=RealIntervalField(100)
n=28
x=range(n+1);a=range(n)
x[0]=R(pi).upper()
for i in range(n):
    a[i] = int(x[i])
    x[i+1]=1/(x[i]-a[i])
    print(a[i])

but now a[28] is wrong (compared to continued_fraction). So 100 bits is not enough. How I discover it?