Lazy Propagation
Introduction
Lazy propagation is a technique used to delay the updates in segment trees. It is used to optimize the time complexity of range updates in segment trees.
Lazy Propagation
Explanation
Segment tree with lazy propagation is a data structure for the pair of a monoid $(M, \ast)$ and a set of functions $F \subset M^M$ which satisfies the following property:
- $\forall f\in F, \; \forall a,b\in M, \; f(a\ast b) = f(a)\ast f(b)$
i.e. $f$ is an endomorphism of the monoid $(M, \ast)$.
- $F = \mathrm{End}(M)$
An endomorphism is a homomorphism from a set to itself, and a homomorphism is a function that preserves the structure of the set. (Read more about homomorphisms here.)
Consider an array $A=[a_1,\cdots,a_N]$ of size $N$. Segment tree with lazy propagation can carry out two types of queries in $O(\log N)$ time complexity:
- Range query: Given a range $[l,r]$, find the result of applying the operation $\ast$ to all elements in the range. In other words, calculate $a_l \ast a_{l+1} \ast \cdots \ast a_r$.
- Range update: Given a range $[l,r]$ and a function $f\in F$, update all elements in the range by applying the function $f$ to them: $[\cdots,a_l,\cdots,a_r,\cdots] \to [\cdots, f(a_l), \cdots, f(a_r),\cdots]$
The idea is to store the updates in the nodes of the segment tree and apply them only when needed. Such new tree nodes are called lazy nodes and the updates are called lazy updates.
graph TD
subgraph s1 ["1/[1:4]"];
f1(4); l1( ):::lazy;
end;
subgraph s2 ["2/[1:2]"];
f2(2); l2( ):::lazy;
end;
subgraph s3 ["3/[3:4]"];
f3(2); l3( ):::lazy;
end;
subgraph s4 ["4/[1]"];
f4(1); l4( ):::lazy;
end;
subgraph s5 ["5/[2]"];
f5(1); l5( ):::lazy;
end;
subgraph s6 ["6/[3]"];
f6(1); l6( ):::lazy;
end;
subgraph s7 ["7/[4]"];
f7(1); l7( ):::lazy;
end;
s1:::round---s2:::round & s3:::round;
s2---s4:::round & s5:::round;
s3---s6:::round & s7:::round;
classDef lazy fill:#ccf;
classDef round rx:10,ry:10,fill:#ffe;
The binary tree above is an example of a segment tree with lazy propagation for the array $[1,1,1,1]$. In each subgraph, the left node represents the sum of all elements in the range, and the right node represents the lazy update.
When we update the range $[2:4]$ by adding $1$ to all elements, the segment tree will be updated as follows.
graph TD
subgraph s1 ["1/[1:4]"];
f1(7); l1( ):::lazy;
end;
subgraph s2 ["2/[1:2]"];
f2(3); l2( ):::lazy;
end;
subgraph s3 ["3/[3:4]"];
f3(4); l3(1):::lazy;
end;
subgraph s4 ["4/[1]"];
f4(1); l4( ):::lazy;
end;
subgraph s5 ["5/[2]"];
f5(2); l5(1):::lazy;
end;
subgraph s6 ["6/[3]"];
f6(1); l6( ):::lazy;
end;
subgraph s7 ["7/[4]"];
f7(1); l7( ):::lazy;
end;
s1:::round---s2:::round & s3:::round;
s2---s4:::round & s5:::round;
s3---s6:::round & s7:::round;
classDef lazy fill:#ccf;
classDef round rx:10,ry:10,fill:#ffe;
When we update the range $[1:3]$ by adding $1$ to all elements, the segment tree will be updated as follows.
graph TD
subgraph s1 ["1/[1:4]"];
f1(10); l1( ):::lazy;
end;
subgraph s2 ["2/[1:2]"];
f2(5); l2(1):::lazy;
end;
subgraph s3 ["3/[3:4]"];
f3(5); l3( ):::lazy;
end;
subgraph s4 ["4/[1]"];
f4(1); l4( ):::lazy;
end;
subgraph s5 ["5/[2]"];
f5(2); l5(1):::lazy;
end;
subgraph s6 ["6/[3]"];
f6(3); l6(2):::lazy;
end;
subgraph s7 ["7/[4]"];
f7(2); l7(1):::lazy;
end;
s1:::round---s2:::round & s3:::round;
s2---s4:::round & s5:::round;
s3---s6:::round & s7:::round;
classDef lazy fill:#ccf;
classDef round rx:10,ry:10,fill:#ffe;
When we query the sum of the elements in the range $[3,4]$, the segment tree will be updated and the nodes will be visited as follows.
graph TD
subgraph s1 ["1/[1:4]"];
f1(10); l1( ):::lazy;
end;
subgraph s2 ["2/[1:2]"];
f2(5); l2( ):::lazy;
end;
subgraph s3 ["3/[3:4]"];
f3(5); l3( ):::lazy;
end;
subgraph s4 ["4/[1]"];
f4(2); l4(1):::lazy;
end;
subgraph s5 ["5/[2]"];
f5(3); l5(2):::lazy;
end;
subgraph s6 ["6/[3]"];
f6(3); l6(2):::lazy;
end;
subgraph s7 ["7/[4]"];
f7(2); l7(1):::lazy;
end;
s1:::round---s2:::round & s3:::round;
s2---s4:::round & s5:::round;
s3---s6:::round & s7:::round;
s5:::highlight; s6:::highlight;
classDef lazy fill:#ccf;
classDef round rx:10,ry:10,fill:#ffe;
classDef highlight fill:#888,stroke:#444,stroke-width:2px;
Code
Let’s see the sample code.
const int N;
const int TREE_SIZE = 1 << ((int)ceil(log2(N)) + 1);
data A[N];
Node tree[TREE_SIZE];
Node_lazy lazy[TREE_SIZE];
Node merge(Node a,Node b){ /* merge two nodes */ }
Node conv(data a){ /* convert data to node */ }
Node identity(){ /* return identity node */ }
Node_lazy merge_lazy(Node_lazy a,Node_lazy b){ /* merge two lazy nodes */ }
Node_lazy identity_lazy(){ /* return identity lazy node */ }
Node apply_lazy(Node nd,Node_lazy lz){ /* apply lazy node to the node */ }
Node init(int nd,int l,int r){
if(l==r) return tree[nd] = conv(A[l]);
int m = (l+r)/2;
return tree[nd] = merge(init(nd*2,l,m),init(nd*2+1,m+1,r));
}
void propagate(int nd,int l,int r){
if(lazy[nd]==identity_lazy()) return;
lazy[nd*2] = merge_lazy(lazy[nd*2],lazy[nd]), lazy[nd*2+1] = merge_lazy(lazy[nd*2+1],lazy[nd]);
tree[nd*2] = apply_lazy(tree[nd*2],lazy[nd]), tree[nd*2+1] = apply_lazy(tree[nd*2+1],lazy[nd]);
lazy[nd] = identity_lazy();
}
Node update(int nd,int l,int r,int s,int e,Node_lazy lz){
if(e<l or r<s) return tree[nd];
if(s<=l and r<=e){
lazy[nd] = merge_lazy(lazy[nd],lz);
return tree[nd] = apply_lazy(tree[nd],lz);
}
propagate(nd,l,r);
int m = (l+r)/2;
return tree[nd] = merge(update(nd*2,l,m,s,e,lz),update(nd*2+1,m+1,r,s,e,lz));
}
Node Query(int nd,int l,int r,int s,int e){
if(e<l or r<s) return identity();
if(s<=l and r<=e) return tree[nd];
propagate(nd,l,r);
int m = (l+r)/2;
return merge(Query(nd*2,l,m,s,e),Query(nd*2+1,m+1,r,s,e));
}
Example
Let’s think of an example of calculating the sum of elements in a range and updating the elements in a range by adding a constant value. Then, the code will be as follows:
struct Node{
int sum,len;
Node(int sum=0,int len=0):sum(sum),len(len){}
};
Node merge(Node a,Node b){
return Node(a.sum+b.sum,a.len+b.len);
}
Node conv(int a){
return Node(a,1);
}
Node identity(){
return Node(0,0);
}
int merge_lazy(int a,int b){
return a+b;
}
int identity_lazy(){
return 0;
}
Node apply_lazy(Node nd,int lz){
return Node(nd.sum+nd.len*lz,nd.len);
}