思路:
1. 对于每一个节点, 返回其所在子树所能提供的最大值, 且该最大值必须是单支的, WA 过
2. max(0, max(lf, rt))+root->val, 可以仅返回根节点, WA 过
3. 需要维护一个全局最优解 ans, WA 过
代码:
class Solution { public: int ans; int solve_dp(TreeNode *root) { if(root == NULL) return 0; int sum = root->val; int lf = 0, rt = 0; if(root->left) lf = solve_dp(root->left); if(root->right) rt = solve_dp(root->right); if(lf > 0) sum += lf; if(rt > 0) sum += rt; ans = max(ans, sum); return max(0, max(lf, rt))+root->val; } int maxPathSum(TreeNode *root) { ans = -100000000; solve_dp(root); return ans; } };