Dijkstra implementation keeps giving wrong shortest paths.

[PanthersHowl]

My Dijkstra implementation keeps giving wrong shortest paths. Getting infinite loops or incorrect distances.

[MaxValueBuilder]

Here's a production-ready Dijkstra's algorithm with priority queue:

import heapq
from collections import defaultdict

def dijkstra(graph, start):
    # Priority queue: (distance, node)
    pq = [(0, start)]
    distances = {node: float('inf') for node in graph}
    distances[start] = 0
    prev_node = {}
    
    while pq:
        current_dist, current = heapq.heappop(pq)
        
        if current_dist > distances[current]:
            continue
            
        for neighbor, weight in graph[current].items():
            distance = current_dist + weight
            
            if distance < distances[neighbor]:
                distances[neighbor] = distance
                prev_node[neighbor] = current
                heapq.heappush(pq, (distance, neighbor))
    
    return distances, prev_node

# Example usage
graph = {
    'A': {'B': 4, 'C': 2},
    'B': {'A': 4, 'C': 1, 'D': 5},
    'C': {'A': 2, 'B': 1, 'D': 8},
    'D': {'B': 5, 'C': 8}
}

distances, prev = dijkstra(graph, 'A')
print(distances)  # {'A': 0, 'B': 2, 'C': 2, 'D': 7}

[SeifeddineJamei]

@PanthersHowl
Common Root Causes for Dijkstra Failures

If your Dijkstra’s algorithm is failing, it is likely due to one of these three logic errors:

1. Missing "Visit" Check (The Infinite Loop)
   Dijkstra can enter an infinite loop if you don't keep track of visited nodes. If you re-process a node that already has its shortest path finalized, you might cycle indefinitely.

• Fix: Use a visited set or only process a node if the distance in the queue is equal to the current known distance.

2. Not Using a Priority Queue (Min-Heap)
   If you use a regular queue (BFS style), it won't work for weighted graphs. You must always pick the node with the minimum distance.
   Fix: Use heapq in Python.

3. Negative Edge Weights
   Dijkstra’s algorithm cannot handle negative weights. It assumes that once a node is visited, its path is the absolute shortest.
   Fix: If your graph has negative edges, you must use the Bellman-Ford algorithm instead.

Correct Python Implementation (using heapq)

import heapq

def dijkstra(graph, start):
    # distances[node] = infinity by default
    distances = {node: float('infinity') for node in graph}
    distances[start] = 0
    
    # Priority Queue stores (distance, node)
    pq = [(0, start)]
    
    while pq:
        current_dist, current_node = heapq.heappop(pq)
        
        # 1. Skip if we found a better path already
        if current_dist > distances[current_node]:
            continue
            
        for neighbor, weight in graph[current_node].items():
            distance = current_dist + weight
            
            # 2. Relaxation Step
            if distance < distances[neighbor]:
                distances[neighbor] = distance
                heapq.heappush(pq, (distance, neighbor))
                
    return distances

Quick Checklist:

• Are you using heapq.heappush and heapq.heappop?
• Are you initializing distances to infinity?
• Are you sure there are no negative weights in your data?