Informally, it gives an asymptotic closed form for recurrences
of the form T(n) = a T(n/b) + f(n)
The behavior of T is dominated either by an nlogba
term arising from the recursion, or by f(n), according to how
big (asymptotically) f is compared to nlogba
Why floor/ceiling in theorem? Notation?
Uses in theorem statement acknowledge that T is only defined on
natural numbers, so references to n/b are short-hands for
floor or ceiling of n/b
Applicable to multiple sub-problem sizes?
No, theorem is about T(n) = a T(n/b), doesn’t allow for
anything other than n/b as sub-problem size
Do sums work the same for n not exact power of b?
Yes, once certain asymptotic approximations are established
Overview of proof structure
Part 1: theorem holds if n = bk for natural number k
Lemma 4.2 - T(n) = Θ(nlogba) + sum
Lemma 4.3 - closed forms for sum
Lemma 4.4 - combine 4.2 and 4.3 to prove Master Theorem
Part 2: adapt part 1 for general n
Consider how “adapting” works for third case in Lemma 4.3