bianchengzhimei22（bianchengzhimei22）.docVIP

bianchengzhimei22（bianchengzhimei22） Find the maximum number of K In the interview, ask the following: Q: there are a lot of unordered numbers. Lets just assume that they are not equal. How do you choose the largest one? How many numbers? Answer: you can write like this: int array[100] Question: Well, what if there are more elements? Answer: that can be changed to: int array[1000] Q: what if we have a lot of elements, such as 100 million floating point numbers? Answer: A, ten, 100, 1000, 000 That can be written: float, array, [100, 000, 000] Q: can such programs be compiled and run? Answer: Well I never wrote so much 0 Analysis and solution [solution 1] When students write and write float array , they often fail to think of the data structure How is it implemented on the computer, from the stack of the current program (Stack), or from the Heap?, Or computer memory may not fit such a big thing? Lets first assume that the number of elements is not large, for example, in the thousands or so, in this case, then I Lets sort them out. Here, quick sorting or heap sorting is a good choice for their average The time complexity is O (N * log2N). Then remove the first K, O (K). Total time complexity O (N * log2N) + O (K) = O (N * log2N). You must have noticed that when K=1, the algorithm above is also O (N * log2N) complexity, Obviously, we can get results by comparing and exchanging by N-1 times. The above algorithm the whole array Sorting, and the original subject only requires the largest number of K, does not require the former K number in order, and also do not need After that, the N-K numbers are in order. How can you avoid sorting after N-K numbers? We need partial sorting algorithms, select rows Both ordering and exchanging sorting are good choices. Sort out the large numbers of the first K in the N numbers, and the complexity is O (N * K). Which one is better? O (N * log2N) or O (N * K)? It depends on the size of the K. Its the question you need to know in t