ACM PKU Eight Puzzle

[fushar]

http://acm.pku.edu.cn/JudgeOnline/problem?id=1077
For anyone that has got accepted in this problem, please help me! I got WAs so many... what else I might miss? I just used A* with priority_queue. It works for all TC I created myself....

[mf]

Just do a simple breadth-first search.

You can find some I/O from the original ACM contest here:
http://www.acm.inf.ethz.ch/ProblemSetAr ... hCen/1998/

[fushar]

DId you mean that I should do BFS from final state to ALL reachable states?
Oh one more, how did you mark a state as visited?

[mf]

DId you mean that I should do BFS from final state to ALL reachable states?

There's only about 9!/2 = 181440 reachable states. 1 second is more than enough time to check them all.

Oh one more, how did you mark a state as visited?

Convert permutation into an integer in any way, and store them in std::set<int>.
(Or a boolean array, if you're not lazy, and the range of integers is small enough)

[fushar]

Oh, I see. So if the queried state is not visited, then I should output "unsolvable". Right?
Anyway thanks for your help. I'll try to recode my solution.

[fushar]

Uh, I just used plain BFS without heuristic, and got TLE....

[mf]

My old BFS solution for UVa's problem 652 gets accepted at pku, running 0.438 seconds.

It doesn't use STL, though as I was coding in C - it converts each permutation into its sequential number (from 0 to 9!-1) and uses a boolean array to mark visited states.

[fushar]

How did you do the conversion?

[mf]

Like this:

Code: Select all

int fact[10];  // a table of factorials

int encode(int a[9]) {
	int c[10], i, k, r;
	for (i = 0; i < 9; i++) c[i] = i;
	for (r = 0, k = 0; k < 9; k++) {
		r += c[a[k]] * fact[8 - k];
		for (i = a[k]; i < 9; i++) c[i]--;
	}
	return r + 1;
}

void decode(int a[9], int r) {
	int c[10], i, k;
	for (i = 0; i < 9; i++) c[i] = i;
	for (r--, k = 0; k < 9; k++) {
		i = r / fact[8 - k];
		r %= fact[8 - k];
		a[k] = c[i];
		for (; i < 9; i++) c[i] = c[i + 1];
	}
}

[fushar]

Hey thank you so much.......!!!!!

[fushar]

I'm so tired getting WAs, what's wrong with my code?

Code: Select all

Deleted AC code

[mf]

Your program prints \0 character at the beginning of output:

Code: Select all

user@localhost:/tmp$ g++ -o p p.cc && ./p >output
2 3 4 1 5 x 7 6 8
user@localhost:/tmp$ hexdump -C output 
00000000  00 75 6c 6c 64 64 72 75  72 64 6c 6c 75 72 64 72  |.ullddrurdllurdr|
00000010  75 6c 64 72 0a                                    |uldr.|
00000015
user@localhost:/tmp$

[fushar]

What a silly bug..
Just changed

Code: Select all

if (x != -2)
	{
		trace(P[x]);
		printf("%c", M[x]);
	}

to

Code: Select all

if (P[x] != -2)
	{
		trace(P[x]);
		printf("%c", M[x]);
	}

and got AC!!!!!!!
Thanks mf!!

[Angeh]

Like this:

Code: Select all

int fact[10];  // a table of factorials

int encode(int a[9]) {
	int c[10], i, k, r;
	for (i = 0; i < 9; i++) c[i] = i;
	for (r = 0, k = 0; k < 9; k++) {
		r += c[a[k]] * fact[8 - k];
		for (i = a[k]; i < 9; i++) c[i]--;
	}
	return r + 1;
}

void decode(int a[9], int r) {
	int c[10], i, k;
	for (i = 0; i < 9; i++) c[i] = i;
	for (r--, k = 0; k < 9; k++) {
		i = r / fact[8 - k];
		r %= fact[8 - k];
		a[k] = c[i];
		for (; i < 9; i++) c[i] = c[i + 1];
	}
}

is this a Known Algorithm? can you explain more why to code like this ?:)

[fushar]

It's just a simple algorithm to convert a permutation into its index (from 0 to 9!-1) and vice versa.