Privacy Boost uses LeanIMT (Lean Incremental Merkle Tree) as the Merkle tree for recording internal state. The biggest feature of LeanIMT is that incremental appends are extremely efficient. In a typical binary Merkle tree, the depth is fixed, and each time a new leaf is added, the entire path must be recalculated while filling with zero hashes. In contrast, LeanIMT increases the tree depth dynamically according to the number of leaves, and when the right child is empty, it adopts a method of lifting the left child value directly without using zero nodes. Thanks to this structure, the number of Poseidon2 hash computations required for each insert is minimized, and the cost of adding new UTXOs onchain remains very low.In addition, LeanIMT stores only the minimum information required at each level, and is designed so that only the “paths that need to change” are updated whenever a new leaf is added. Because it does not require recalculating all levels like a full Merkle tree, it is an efficient structure for Privacy Boost.As a result, LeanIMT can be seen as a data structure for maintaining the UTXO state of the shielded pool onchain at low cost.