Easy
Balanced Binary Tree:Height or -1 on the Way Up, Not Max-Depth Twice
Height-balanced means |lh - rh| <= 1 at every node. Nested maxDepth at each node is correct and quadratic; one post-order walk can return the height or fail.
Read MoreTrees
23 questions
-
- Easy
Diameter:Left Height Plus Right Height, Not Nested Height Walks
The longest path through a node is leftHeight + rightHeight, counted in edges. Nested height walks are correct and quadratic; empty and a lone node are both 0.
Read More - Easy
Invert Binary Tree:Swap the Children, Keep the Same Nodes
Inverting a tree is a left/right swap at every node. A second tree of copied values looks mirrored and is not the tree callers still hold.
Read More - Easy
Maximum Depth:One Plus the Taller Child, Not Every Root-to-Leaf Walk
Depth is 1 + max(left, right). Enumerating every root-to-leaf path is correct and extra; empty is 0, a lone node is 1, and this prompt counts nodes not edges.
Read More - Easy
Path Sum:Remaining Hits Zero at a Leaf, Not an Internal Match
Recurse with target minus node.val. True only when a leaf's remaining is 0. Returning true because an internal node equals the target is the trap.
Read More - Easy
Same Tree:Walk Both at Once, Not Two Serializations
Two trees match only if a paired walk sees the same structure and values. Serializing both then comparing is correct and extra; a left child versus a right child is already false.
Read More - Easy
Sorted Array to BST:Mid of Each Slice Is the Root, Not Sequential Insert
Inserting a sorted array in order builds a right spine. The mid of each remaining slice is the root: BST order and height-balance from the same cut.
Read More - Easy
Subtree of Another Tree:Same Tree at Some Node, Not Only the Root
A match can hang off any node. Comparing only the two roots is Same Tree; call that check here, then try the children.
Read More - Easy
Symmetric Tree:Mirror Walk, Not Invert Then Compare
A tree is symmetric only if a paired walk sees mirrors. Inverting a copy then running Same Tree is correct and extra; left with left is the wrong pairing.
Read More
- Medium
BST Iterator:Pause Inorder Between Calls, Do Not Dump the Sorted List
Constructor stacks the left spine only. next() pops one visit and pushes the left spine of the right child — materializing every key up front is the brute that pays O(n) before the first call.
Read More - Medium
Count Good Nodes:Path Max on the Walk, Not a Nested Climb
A node is good if it is at least the max on the path from the root. Scanning that path at every node is quadratic; checking only the parent is not enough.
Read More - Medium
Flatten Binary Tree:Preorder Spine, Same Nodes, Left Null
A prev pointer in reverse post-order hangs right as next. Copying values into a list then building a new chain is not the tree callers still hold.
Read More - Medium
Kth Smallest in a BST:Stop Inorder at k, Do Not Dump the Sorted List
Inorder of a BST is already rank order. Materializing every key then indexing k-1 is correct and throws away the early exit — return on the kth visit.
Read More - Medium
LCA of a Binary Tree:Search Both Sides, Combine When Both Hit
No BST order, so both children get searched. If both return a hit, this node is the LCA — walking by key comparison is the other problem.
Read More - Medium
LCA of a BST:Walk the Split, Do Not Search Both Subtrees
Both keys smaller → left, both larger → right, split → this node. Path lists are correct and ignore the order you already paid for.
Read More - Medium
Level Order:Drain the Queue by Count, Not a Recursive Dump
Snapshot q.size() and poll that many nodes — that is one level. A second queue is extra furniture; DFS into depth buckets is a follow-up, not the default walk.
Read More - Medium
Path Sum III:Hash the Prefix Complement on the Walk, Not a Nested Start at Every Node
Count downward paths that sum to target by hashing prefixes on the current root-to-node walk. Nested starts are honest and quadratic; skip the decrement and siblings steal counts.
Read More - Medium
Rebuild the Tree:Preorder Names the Root, Inorder Cuts the Sides
The next unused preorder value is always the root; its inorder index is the left/right cut. Scanning inorder for every root still rebuilds the tree — in quadratic time.
Read More - Medium
Right Side View:Keep the Last Poll of Each Floor
Snapshot q.size() and record only the last poll of that drain. Walking only right children misses a left-subtree node that peeks out as the sole node on a deeper floor.
Read More - Medium
Validate BST:Bounds on the Walk, Not a Sorted Snapshot
A tree is a BST only if every node sits inside the range its ancestors allow. An inorder dump that looks sorted can still hide a violation if you check the wrong thing — and a left-child-only check is not enough.
Read More - Medium
Zigzag:Reverse Every Other Floor After the Same Drain
Snapshot size, collect the floor left to right, reverse the odd ones. A pair of ArrayDeques, or polling either end, are variants, not a second BFS.
Read More
- Hard
Maximum Path Sum:One Child Up, Both Children Through, Drop Negatives
Return node + max(0, leftGain, rightGain) — one child, because a path to the parent cannot fork. Start the global max at Integer.MIN_VALUE because all-negative trees exist.
Read More - Hard
Serialize:Keep Null Tokens, Not a Dense Heap Array
Level-order that drops holes cannot round-trip. Encode missing children as string tokens and never offer null to ArrayDeque; a 2^h slot dump wastes a spine.
Read More
Learn the structure
Representation and operations — not the problem set.