Bidirectional breadth-first search: Examples & Code

Algorithm Visualizations November 16, 2026 javascript

Published by Dimitris Kougioumtzis · Updated

Expand complete BFS layers from node 0 and from the highest node on reversed edges. Weights are ignored.

Expand complete layers from the source and from the target on reversed edges, joining parent paths at a meeting vertex.

Open the 3D visualization

Bidirectional breadth-first search example state preview
Preview of the example's final state. Complete values are available in the worked example and interactive timeline.

When to use Bidirectional breadth-first search

One minimum-hop route between known endpoints.

Choosing an alternative

One-sided BFS explores only from the source; Dijkstra incorporates weights.

Worked example: input and state changes

Example input

0,1,1;1,2,1;2,0,1;2,3,1;3,4,1;4,3,1

These checkpoints follow the actual example timeline. State details show the stored values and decisions; use the visualization to inspect every intervening step.

  1. Step 0: Initialize graph state.

    State at this checkpoint
    {
      "values": [
        "-",
        "-",
        "-",
        "-",
        "-"
      ],
      "active": [],
      "kind": "graph",
      "chosen": [],
      "visited": []
    }
  2. Step 3: Discover 1 from the source side.

    State at this checkpoint
    {
      "values": [
        "F:0 B:-",
        "F:1 B:-",
        "F:- B:-",
        "F:- B:-",
        "F:- B:0"
      ],
      "active": [
        0,
        1
      ],
      "kind": "graph",
      "chosen": [],
      "visited": [],
      "forward": [
        0,
        1,
        -1,
        -1,
        -1
      ],
      "backward": [
        -1,
        -1,
        -1,
        -1,
        0
      ]
    }
  3. Step 5: Discover 3 from the target side.

    State at this checkpoint
    {
      "values": [
        "F:0 B:-",
        "F:1 B:-",
        "F:- B:-",
        "F:- B:1",
        "F:- B:0"
      ],
      "active": [
        4,
        3
      ],
      "kind": "graph",
      "chosen": [],
      "visited": [],
      "forward": [
        0,
        1,
        -1,
        -1,
        -1
      ],
      "backward": [
        -1,
        -1,
        -1,
        1,
        0
      ]
    }
  4. Step 8: Expand reverse BFS layer at 3.

    State at this checkpoint
    {
      "values": [
        "F:0 B:-",
        "F:1 B:-",
        "F:2 B:-",
        "F:- B:1",
        "F:- B:0"
      ],
      "active": [
        3
      ],
      "kind": "graph",
      "chosen": [],
      "visited": [],
      "forward": [
        0,
        1,
        2,
        -1,
        -1
      ],
      "backward": [
        -1,
        -1,
        -1,
        1,
        0
      ]
    }
  5. Step 10: Shortest path has 4 edges: 0 > 1 > 2 > 3 > 4.

    State at this checkpoint
    {
      "values": [
        "F:0 B:-",
        "F:1 B:-",
        "F:2 B:2",
        "F:- B:1",
        "F:- B:0"
      ],
      "active": [],
      "kind": "graph",
      "chosen": [
        0,
        1,
        3,
        4
      ],
      "visited": [
        0,
        1,
        2,
        3,
        4
      ],
      "result": 4,
      "path": [
        0,
        1,
        2,
        3,
        4
      ],
      "forward": [
        0,
        1,
        2,
        -1,
        -1
      ],
      "backward": [
        -1,
        -1,
        2,
        1,
        0
      ]
    }

Expected final state

Shortest path has 4 edges: 0 > 1 > 2 > 3 > 4.

Complexity and input bounds

Family: Graphs. Time complexity: O(V + VE) with edge-array scans.

Space complexity: O(V) frontiers, distances, and parents.

Working-space bounds describe algorithm state and any explicitly noted output, excluding the visualization's saved timeline. Recording frames can use more memory than running the function alone. Here n is sequence length, V and E are vertex and edge counts, and other parameters refer to the named input dimensions.

Directed edges as source,target,weight; separated by semicolons. Nodes 0–7, positive weights 1–99. Start node is 0. At most 16 edges. Weights are ignored. Vertices include every index from 0 to the highest endpoint.

The playground validates these limits before starting. Downloaded functions are callable implementations, with parameter and output-buffer contracts in their comments; they do not all repeat the playground's input validation.

Common mistakes and limitations

Reverse search must follow incoming original edges, and complete layers support the stopping argument.

Further reading

For background on algorithm analysis and the JavaScript function interface:

Editorial and testing approach · Browse all algorithm guides

Complete JavaScript implementation

javascript
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
// Edge arrays from/to/weight; 1..8 vertices indexed 0..vertices-1, 0..16 edges, endpoints in bounds. Weights ignored. Output buffer at least 256 integers. Directed shortest unweighted path from 0 to vertices-1. Returns edge count or -1; writes path nodes including both endpoints. Reference snippets assume validated playground inputs. Fixed scratch buffers have 256 slots. Sorting variants may use a different traversal strategy than the animation.
function algorithm(from, to, weight, edges, vertices, output) {
    let forward = Array(256).fill(0); let backward = Array(256).fill(0); let parent = Array(256).fill(0); let next = Array(256).fill(0); let queueF = Array(256).fill(0); let queueB = Array(256).fill(0); let reversed = Array(256).fill(0);
    let u = 0; let v = 0; let index = 0; let headF = 0; let tailF = 1; let headB = 0; let tailB = 1; let endF = 0; let endB = 0; let meet = -1; let best = 1000000; let count = 0;
    for (u = 0; u < vertices; u++) { forward[u] = -1; backward[u] = -1; parent[u] = -1; next[u] = -1; }
    forward[0] = 0; backward[vertices - 1] = 0; queueF[0] = 0; queueB[0] = vertices - 1;
    while (headF < tailF && headB < tailB && meet < 0) {
        endF = tailF; endB = tailB;
        while (headF < endF) { u = queueF[headF++];
            for (index = 0; index < edges; index++) { v = to[index]; if (from[index] == u && forward[v] < 0) { forward[v] = forward[u] + 1; parent[v] = u; queueF[tailF++] = v; } }
        }
        while (headB < endB) { u = queueB[headB++];
            for (index = 0; index < edges; index++) { v = from[index]; if (to[index] == u && backward[v] < 0) { backward[v] = backward[u] + 1; next[v] = u; queueB[tailB++] = v; } }
        }
        for (u = 0; u < vertices; u++) { if (forward[u] >= 0 && backward[u] >= 0 && forward[u] + backward[u] < best) { meet = u; best = forward[u] + backward[u]; } }
    }
    if (meet < 0) { return -1; }
    u = meet; while (u >= 0) { reversed[count++] = u; u = parent[u]; }
    for (index = 0; index < count; index++) { output[index] = reversed[count - 1 - index]; }
    u = next[meet]; while (u >= 0) { output[count++] = u; u = next[u]; }
    return best;
}

Source

Early primary research on bidirectional and heuristic search. The playground uses complete BFS layers and directed-edge reversal.