Hey there! Let's dig into why your Percolation program is freezing up with no output when running with n=200 and 100 trials. This could be either a bug causing an infinite loop or a performance bottleneck, but based on common pitfalls in this Coursera assignment, bugs are the far more likely culprit. Here's a breakdown of the most probable issues and how to check them:

1. Infinite Loops (Top Suspect) #

The most common reason for this kind of freeze is an infinite loop in your trial logic—specifically, repeatedly trying to open sites that are already open.

If your code doesn't track which sites are already open (e.g., no boolean[][] isOpen array to mark opened sites), or your random site selection doesn't skip already open sites, you could end up in a loop where you keep picking the same opened sites over and over instead of making progress toward percolation. For example, a flawed trial loop might look like this:

// Problematic: no check for already open sites
while (!percolates()) {
    int i = StdRandom.uniform(n) + 1;
    int j = StdRandom.uniform(n) + 1;
    open(i, j);
}

Even with n=200 (40,000 total sites), random chance could lead to stretches where you keep hitting open sites, causing the program to hang indefinitely.

2. Performance Bottlenecks (Less Likely for n=200) #

WeightedQuickUnionUF is extremely efficient (amortized nearly constant time per operation), so n=200 shouldn't cause performance issues on its own. That said, if you're skipping key optimizations, you might see slowdowns that feel like a freeze:

• No virtual nodes: If you check percolation by iterating all top-row sites and all bottom-row sites to see if any pair is connected, that's O(n) checks per percolation test. For n=200, that's 400 operations per check—manageable, but if combined with an infinite loop, it'll make the freeze worse.
• Unnecessary repeated calculations: If your open() or percolates() methods do redundant work (like recalculating connections from scratch every time), that could add up, but this is rare with WeightedQuickUnionUF.

3. Other Bug Candidates #

• Incorrect index conversion: If you're converting (row, column) coordinates to the 1D index used by WeightedQuickUnionUF incorrectly (e.g., off-by-one errors), you might break connectivity checks. This could make percolates() never return true, trapping your trial loop forever.
• Uninitialized state: If you forgot to initialize your isOpen array (or never mark sites as open when you call open()), your program will never make progress toward percolation, leading to an infinite loop.

Quick Debugging Steps #

• Add progress logging: Insert a print statement that tracks the number of open sites (e.g., System.out.println("Open sites: " + openCount)). If this number stops growing or grows extremely slowly, you're definitely stuck opening already-open sites.
• Validate percolates(): Manually open all sites in a small test case (e.g., n=3) and call percolates()—if it returns false, your connectivity logic is broken.
• Check virtual node setup: If you're using virtual nodes (top and bottom), make sure you're connecting top-row sites to the top virtual node when they're opened, and bottom-row sites to the bottom virtual node. A missing connection here will prevent percolates() from ever returning true.

内容的提问来源于stack exchange，提问作者mc2017