Comment on page
0-1 背包问题
给定 n 个物品的 重量 weight，价值 value，以及一个容量为 W 的背包，求如何装入物品，使背包内物品价值最大
对于一件物品，只能选择或不选择（0-1）
- value[n] —— n 个物品的价值表
- weight[n] —— n 个物品的重量表
- MaxWeight —— 背包的最大容量
- MaxValue —— 背包可获得的最大价值
- items —— 选择装入背包的物品列表
Input:val[] = {60, 100, 120}wt[] = {10, 20, 30}MaxWeight = 50Output: MaxValue = 220items = [1,2]
Solution is below!
使用二维数组 map[N][W] ，N 为放入背包的最大物品个数， W 为背包中物品的最大重量， 从 map[0][0]开始构建数组，（map[0][j] = 0, map[i][0] = 0）
对于 map[i][j]，其可能取值只有两种情况：
- 1.第i个物品不放人背包，则map[i][j] = map[i-1][j]
- 2.第i个物品放入背包， 则 map[i][j] = map[i-1][j-weight[i]] + value[i]
当 weight[i] 小于当前最大背包容量 j 时，只能取第1种
Python
C++
def OneZero_Knapsack_DP(Weight:list, Value:list, MaxWeight):
size = len(Weight)
# 初始化二维数组，加1是增加为0的情况
totalValue = [[0] * (MaxWeight + 1) for i in range(size + 1)]
for i in range(size + 1):
for j in range(MaxWeight + 1):
if i == 0 or j == 0:
totalValue[i][j] = 0
elif j >= Weight[i - 1]:
# 第i个物品可以放入背包
# 取二者最大值
totalValue[i][j] = max(totalValue[i - 1][j],
totalValue[i - 1][j - Weight[i - 1]] + Value[i - 1])
else:
# 第i个物品不能放入背包
totalValue[i][j] = totalValue[i - 1][j]
return totalValue[size][MaxWeight]
# 优化方法，
# 由于在计算可放入物品数量为k时的最大价值 totalValue[k][j] 时，
# 只需知道可放入物品数为 k-1 时的最大价值 totalValue[k-1][j] 和 totalValue[k-1][j-w[k]]，
# 因此我们只需要保留最大价值与背包容量的关系，即只需构建数组 totalValue[MaxWeight+1]，
# 用totalValue[w]的新值与旧值来表示放入 i 个物品的最大价值和 放入 i-1 个物品时的最大价值
# 然后用同样的方法进行迭代
def OneZero_Knapsack_DP2(Weight:list, Value:list, MaxWeight):
size = len(Weight)
totalValue = [0 for i in range(MaxWeight + 1)] # 初始化数组 totalValue[MaxWeight+1]
for i in range(size):
for j in range(MaxWeight , Weight[i]-1, -1): # MaxWeight <= j <= Weight[i]
totalValue[j] = max(totalValue[j], Value[i] + totalValue[j - Weight[i]])
# 当背包当前容量小于第i个物品的重量时，i不放入背包，
# 因此 totalValue[j < weight[i]] 的值不用改变
return totalValue[MaxWeight]
// Naive method
int MaxValue( vector<int>value, vector<int>weight, int MaxWeight, int size ) {
if ( size == 0 ) return 0;
if ( weight[size - 1] > MaxWeight ) {
return MaxValue( value, weight, MaxWeight, size - 1 );
}
return max(
value[size - 1] + MaxValue( value, weight, MaxWeight - weight[size - 1], size - 1 ),
MaxValue( value, weight, MaxWeight, size - 1 )
);
}
int OneZero_Knapsack_Naive( vector<int>value, vector<int>weight, int MaxWeight ) {
return MaxValue( value, weight, MaxWeight, value.size() );
}
/*
Dynamic Programming approach for 0-1 knapscak
Not work when value or weight include float items !!
*/
#define index(a,b) ((MaxWeight+1)*(a) + (b))
int OneZero_Knapsack_DP( vector<int>value, vector<int>weight, int MaxWeight ) {
int size = value.size();
vector<int> totalValue;
totalValue.resize( ( MaxWeight + 1 )*( size + 1 ) ); // Use 1-D array to represent matrix tatalValue[size+1][MaxWeight+1]
for ( int i = 0; i <= size; i++ ) {
for ( int j = 0; j <= MaxWeight; j++ ) {
if ( i == 0 || j == 0 )
totalValue[index( i, j )] = 0;
else if ( j >= weight[i - 1] ) {
totalValue[index( i, j )] = max(
totalValue[index( i - 1, j )], // Not include the ith item for max weight j
value[i - 1] + totalValue[index( i - 1, j - weight[i - 1] )] ); // Include the ith item
}
else
// Exceed the max weight, so cannot include the ith item
totalValue[index( i, j )] = totalValue[index( i - 1, j )];
}
}
return totalValue[index( size, MaxWeight )];
}
Last modified 4yr ago