Question
Leetcode - 1425. Constrained Subsequence Sum
Train of thought
- dp[i] = j : j represents the maximum sum that last index is i, dp[0] = nums[0].
- Review k elements, and find maximum.
class Solution {
public int constrainedSubsetSum(int[] nums, int k) {
int n = nums.length;
// dp[i] = j : j represents the maximum sum that last index is i
int[] dp = new int[n];
dp[0] = nums[0];
// loop to calculate the value of dp array
int maxSum = dp[0];
for (int i = 1; i < n; i++) {
// review k elements, and find maximum
int maxAgo = Integer.MIN_VALUE, mE = Math.max(i - k, 0);
for (int m = i - 1; m >= mE; m--) {
if (dp[m] > maxAgo) {
maxAgo = dp[m];
}
}
// calculate dp[i]
if (maxAgo < 0) {
dp[i] = nums[i];
}
else {
dp[i] = maxAgo + nums[i];
}
// update maximum
if (dp[i] > maxSum) {
maxSum = dp[i];
}
}
return maxSum;
}
}
Optimize
We can optimize the cost time of find maximum element by remember first maximum and second maximum, special pay attention to step by step looping the array to calculate it that mark sure we can first time to find the second maximum.
class Solution {
public int constrainedSubsetSum(int[] nums, int k) {
int n = nums.length;
// dp[i] = j : j represents the maximum sum that last index is i
int[] dp = new int[n];
dp[0] = nums[0];
// loop to find the value of dp array
int maxSum = dp[0], maxAgo = dp[0], maxAgoIndex = 0, secondMaxAgo = Integer.MIN_VALUE, secondMaxAgoIndex = -1;
for (int i = 1; i < n; i++) {
// change maximum to second maximum
if (i - maxAgoIndex == k + 1) {
maxAgo = secondMaxAgo;
maxAgoIndex = secondMaxAgoIndex;
//nums: 100, -10, -10, -10, -2, -2, -10, -100, 15, -5, -10, 10, 2, -10, 5, 20
//dp : 100, 90, 90, 80, 88, 86, 78, -14, 15, 10, 5, 20, 22, 12, 27, 47
// maybe has skip element 78: greater 88, second greater 86
if (secondMaxAgoIndex + 1 != i) {
secondMaxAgo = dp[secondMaxAgoIndex + 1];
secondMaxAgoIndex++;
}
else {
secondMaxAgoIndex = -1;
}
}
// calculate dp[i]
if (maxAgo < 0) {
dp[i] = nums[i];
}
else {
dp[i] = maxAgo + nums[i];
}
// update maximum
if (dp[i] > maxSum) {
maxSum = dp[i];
}
// get greater ago
if (dp[i] >= maxAgo) {
maxAgo = dp[i];
maxAgoIndex = i;
secondMaxAgoIndex = -1;
continue;
}
// init second maximum
if (secondMaxAgoIndex == -1) {
secondMaxAgo = dp[i];
secondMaxAgoIndex = i;
continue;
}
// update second maximum
if (dp[i] >= secondMaxAgo) {
secondMaxAgo = dp[i];
secondMaxAgoIndex = i;
}
}
return maxSum;
}
}