Longest

Alright this question has been messing with my brain a bit I’m solving the “Longest Repeating Character Replacement” problem (LeetCode 424). Quick idea of the problem: You’re given a string and a number k. You can change at most k characters in any substring to make all characters the same. Find the longest possible substring you can get. So the usual sliding window solution is: * keep a frequency map * track the count of the most frequent character (maxFreq) * if (window size - maxFreq) > k, shrink the window Here’s the part that’s confusing me… When I shrink the window, I don’t update maxFreq. So technically it can become wrong (like bigger than the actual max frequency inside the current window). That means sometimes: * the condition (window size - maxFreq) <= k might say the window is valid * but in reality, it might not be But somehow this still passes all test cases?? Why does this work? Why don’t we need to recompute maxFreq when the window shrinks? Feels like this should break the logic, but it doesn’t… what am I missing here? PS : I tried asking AI, but i still dont get it, also i wrote this desc with AI cause i didnt know how to phrase it well, thank you