Merge-style sparse dot product

Data Structures 1 implementation

Mathematical Formulation

Computes x·y = Σ x_i * y_i where x and y share index i Uses sorted merge to find common indices in linear time.

Complexity

O(k1 + k2) where k1, k2 are the number of nonzeros @pre Both vectors must be sorted by increasing index

Implementations

CoinUtils

CoinPackedVector stores a sparse vector as parallel arrays of indices and values. Unlike CoinShallowPackedVector, this class owns its storage and supports modification operations.