Binary Search Drill 12-Day

The whole template lives here. Off-by-one is the entire game. Burn lo/hi/mid into muscle memory before you touch anything harder.

Not strictly binary search — but the 'running minimum' habit is the same monotonic-thinking muscle. Use it as a warm-up between BS reps.

Two binary searches: one for leftmost, one for rightmost. The 'bias the search to one side' trick generalizes to lower/upper bound problems.

Flatten the matrix mentally: treat row-major indices as 1D. Or do two binary searches. Either answer works — defend your pick.

First problem to break the 'binary search needs a sorted array' mental model. Compare mid to mid+1 to decide direction.

Binary search where you compare mid to RIGHT, not to a target. The mental model shift — 'compare to anchor, not target' — is the interview signal.

Decide which half is sorted, then check if target lives there. The branching is fiddly — write the four cases out loud before coding.

Binary search inside a hash map's value list. Practice the data-modeling step: 'where does the search happen?' is the whole design question.

Binary search on the ANSWER, not an index. Once you see this shape once, you see it everywhere — name it the moment you spot it.

Same shape as Koko. lo = max(weights), hi = sum(weights), then BS on capacity. Drill the 'find minimum X that satisfies a predicate' template.

Bridge problem: not pure BS, but the 'find the largest valid X' instinct from Koko/capacity carries over. Use it to sharpen the predicate-thinking.

Capstone. Binary search on the partition. The hardest BS problem in the canon — if you can narrate it, you can narrate any BS variant.