SRM 439

D2 1000p PalindromeFactory

 

邮局问题。设Dp[i][j]为从串[i,j]得到回文串的最小步数，然后进行两种操作：删除一个字符、删除两个字符。

 

class PalindromeFactory { string str; int dp[50][50]; int get ( int l, int r ) { if ( l == r || l == r - 1 && str[l] == str[r] ) return dp[l][r] = 0; if ( l == r - 1 && str[l] != str[r] ) return dp[l][r] = 1; if ( dp[l][r] != -1 ) return dp[l][r]; int &res = dp[l][r]; if ( str[l] == str[r] ) res = get ( l + 1, r - 1 ); else res = get ( l + 1, r - 1 ) + 1; res = min ( res, get ( l + 1, r ) + 1 ); res = min ( res, get ( l, r - 1 ) + 1 ); return res; } public: int getEditDistance(string initial) { str = initial; memset ( dp, 0xff, sizeof ( dp ) ); int ret = get ( 0, str.size () - 1 ); int i, j, n = str.size (); for ( i = 0; i < n; i ++ ) for ( j = i + 1; j < n; j ++ ) { memset ( dp, 0xff, sizeof ( dp ) ); swap ( str[i], str[j] ); ret = min ( ret, get ( 0, n - 1 ) + 1 ); swap ( str[j], str[i] ); } return ret; } }

 

D1 250p  PalindromePhrases

 

所求解其实就是二进制表达时1的个数。

 

D1 500p PalindromePhrases

 

相当有技术含量的DP。

 

考虑ab, edcba, cd这样的情况，从外向内构造解。

 

ab和edcba中的ba配对之后，还剩下edc，可以和cd配对。由于只有13个元素，首先考虑到的是bitmask。确定组合后剩下的字符可以做一状态。由于配对有两个方向：如例子中的cd是正方向配对，而edcba是反方向配对，因此将方向作为状态。总状态即为：DP(string 字符串, int 方向, int bitmask)

 

 #define MP make_pair class PalindromePhrases { map < pair < string, int >, long long > m[1 << 13]; vector < string > w[2]; int n; bool isP ( const string &s ) // palindromic { int i = 0, j = s.size () - 1; while ( i < j && s[i] == s[j] ) i ++, j --; return i >= j; } int getPrefix ( const string &a, const string &b ) { int i, ret = 0; for ( i = 0; i < a.size () && i < b.size (); i ++ ) { if ( a[i] == b[i] ) ret ++; else break; } return ret; } long long get ( const string &ex, int r, int used ) { //printf ( "%s %d %d/n", ex.c_str (), r, used ); if ( used == ( 1 << n ) - 1 ) { //printf ( "%s %d %d/n", ex.c_str (), r, used ); //printf ( "back %d/n/n", isP ( ex ) ); return isP ( ex ); } pair < string, int > key = MP ( ex, r ); if ( m[used].count ( key ) ) { //printf ( "%s %d %d/n", ex.c_str (), r, used ); //printf ( "%lld/n/n", m[used][key] ); return m[used][key]; } long long int &res = m[used][key]; res = 0; if ( isP ( ex ) ) res = 1; int i; for ( i = 0; i < n; i ++ ) if ( ( ( 1 << i ) & used ) == 0 ) { int wlen = w[r][i].size (); int plen = getPrefix ( w[r][i], ex ); if ( plen != wlen && plen != ex.size () ) continue; int t; if ( plen == wlen ) { res += ( t = get ( ex.substr ( plen ), r, ( 1 << i ) | used ) ); } else { res += ( t = get ( w[r][i].substr ( plen ), 1 - r, ( 1 << i ) | used ) ); } } //printf ( "%s %d %d/n", ex.c_str (), r, used ); //printf ( "%d/n/n", res ); return res; } }