Complete Binary Tree

Heaps are stored in arrays without explicit pointers because the tree is complete: every level filled left-to-right.

Index:  0 1 2 3 4 5 6 7 8 9 ...
        |
        root
        | |
        | left  child of root
        |
        right child of root

For node at index i (0-indexed):
   left  = 2i + 1
   right = 2i + 2
   parent = (i - 1) // 2

This is the classic Eytzinger layout. Cache-friendly: parent and children are at predictable offsets, so prefetchers love it.

Insertion always at the END (next available slot). Removal of root: swap with the last element, shrink, then sift-down.

Memory: just a flat array. Heap with N entries uses N * 8 bytes (for 64-bit numbers). Compare to red-black tree which has 3 pointers per node (24 bytes overhead).