[抄题]:
Given an array of integers nums
and a positive integer k
, find whether it's possible to divide this array into k
non-empty subsets whose sums are all equal.
Example 1:
Input: nums = [4, 3, 2, 3, 5, 2, 1], k = 4
Output: True
Explanation: It's possible to divide it into 4 subsets (5), (1, 4), (2,3), (2,3) with equal sums.
[暴力解法]:
时间分析:
空间分析:
[优化后]:
时间分析:
空间分析:
[奇葩输出条件]:
[奇葩corner case]:
dfs中的cc是:组数k直接等于1
dfs的退出条件是:cur_sum == target
[思维问题]:
不懂“是否”题为啥要用dfs:这一步能不能、下一步能不能,每个元素都要算到,所以用boolean DFS
[英文数据结构或算法,为什么不用别的数据结构或算法]:
[一句话思路]:
DFS扩展的依据是:下一步没有访问过 visited = f,就设置visited = t &回溯。中间不要直接return,否则后面没法做。
[输入量]:空: 正常情况:特大:特小:程序里处理到的特殊情况:异常情况(不合法不合理的输入):
[画图]:
[一刷]:
- dfs的target直接写成target就行了,就是个参数
[二刷]:
- dfs的start直接写成start就行了,就是个参数 traverse中要变成i + 1
[三刷]:
找到k组之后,继续找k - 1 组,变量变化之后就要控制它的最后值 是否为0或1 了
[四刷]:
[五刷]:
[五分钟肉眼debug的结果]:
[总结]:
有变量k的次数递减,就必须写退出条件。DFS最重要的反而是退出条件
[复杂度]:Time complexity: O(n) Space complexity: O()
[算法思想:迭代/递归/分治/贪心]:
[关键模板化代码]:
[其他解法]:
[Follow Up]:
[LC给出的题目变变变]:
[代码风格] :
[是否头一次写此类driver funcion的代码] :
[潜台词] :
class Solution { public boolean canPartitionKSubsets(int[] nums, int k) { int sum = 0; //ini: get sum for (int num : nums) sum += num; int[] visited = new int[nums.length]; //cc: k == 1 if (k == 1) return true; if (k <= 0 || sum % k != 0) return false; //canPartition return canPartition(0, nums, visited, 0, k, sum / k); } public boolean canPartition(int start, int[] nums, int[] visited, int curSum, int k, int target) { if(k==1) return true; //exit :curSum == target if (curSum == target) return canPartition(0, nums, visited, 0, k - 1, target); //backtracing for (int i = start; i < nums.length; i++) { if (visited[i] == 0) { visited[i] = 1; //do not return directly if(canPartition(i + 1, nums, visited, curSum + nums[i], k, target)) return true; visited[i] = 0; } } return false; } }