Divide & Conquer Recurrence Tree
Expand a balanced recurrence one level at a time. Track the number of subproblems, their sizes, the work per node, and the total work at each level. Compare that exact finite sum with its asymptotic growth.
Model: T(1) = 1; internal-node work is exactly f(n) = nᶜ(logᵦ n)ᵏ. Choosing n = bᴸ avoids rounding subproblem sizes. Reset hides the expansion again.
Recurrence and current step
Recursion tree
Large levels are collapsed into a labeled group. The number of nodes and all cost totals still include every recursive call. Arrows into a group represent the collection of calls, not a single parent-child edge.
Work by level
| Level d | Nodes aᵈ | Size n / bᵈ | Cost / node | Level cost |
|---|
T(n) = Σ[d = 0 … L − 1] aᵈ f(n / bᵈ) + aᴸT(1)