July 2002
Intermediate to advanced
320 pages
8h 15m
English
By the “integer square root” function, we mean the function
To extend its range of application and to avoid deciding what to do with a negative argument, we assume x is unsigned. Thus, 0 ≤ x ≤ 232 − 1.
For floating-point numbers, the square root is almost universally computed by Newton’s method. This method begins by somehow obtaining a starting estimate g0 of
Then, a series of more accurate estimates is obtained from
The iteration converges quadratically—that is, if at some point is accurate to
Read now
Unlock full access