Hi everyone
I need help with this knapsack variant
Thanks
# | User | Rating |
---|---|---|
1 | tourist | 3985 |
2 | jiangly | 3814 |
3 | jqdai0815 | 3682 |
4 | Benq | 3529 |
5 | orzdevinwang | 3526 |
6 | ksun48 | 3517 |
7 | Radewoosh | 3410 |
8 | hos.lyric | 3399 |
9 | ecnerwala | 3392 |
9 | Um_nik | 3392 |
# | User | Contrib. |
---|---|---|
1 | cry | 169 |
2 | maomao90 | 162 |
2 | Um_nik | 162 |
4 | atcoder_official | 160 |
5 | djm03178 | 158 |
6 | -is-this-fft- | 157 |
7 | adamant | 155 |
8 | Dominater069 | 154 |
8 | awoo | 154 |
10 | luogu_official | 151 |
I need help with this knapsack variant
Thanks
Name |
---|
I cant figure out anything from hint but can share my approach. Just make multiple copies of the coin.
Say W={5,4} and K={2,3} then modify W as W1={5,5,4,4,4} and V as V1={100,100,70,70,70}. Forget K now.
Build dp table as dp[i][j]=minimum cost to get sum of j using first i coins. Complexity: O(S*sum(Ki)). You can do it in O(S) space.
For each object X, split its K_i copies into different objects with weight (2^a)*W_i and value (2^a)*V_i, where all the of 2^a of a particular X sum to K_i.
This probably seems a bit vague, so let me show some examples:
Object With 5 Copies: Split it into 3 objects with the weight of 1 copy, 2 copies, and 2 copies, allowing someone to take anywhere between 0 and 5 copies (we can easily see why 2 objects with the weight of 1 copy and 4 copies will not work).
Object With 12 Copies: Split it into 5 objects with the weight of 1 copy, 1 copy, 2 copies, 4 copies, and 4 copies.
We initially start with freq[1] = K_i, and loop for powers of 2 until we encounter freq[i] = 0 to split up the object's copies.
We now have a complexity of S*sqrt(sum of K_i).
EDIT: BTW @AmericanPsycho what book is this?
This book is Competitive Programming 3.
How is the square root coming?
Because 0 <= freq[i] <= 2 and the number of different freqs >= 1 will be O(sqrt(k)) since the sum of the first k natural numbers is O(k^2).