I have just solved this great problem: http://codeforces.net/problemset/problem/429/A
I have solved the problem by using DFS on the given tree, like most others did, and like the editorial suggests.
As I was coding the solution, I was sure that I would get a stack-overflow, and then I would reimplement the solution using BFS. But I did not..
In the problem, it is very possible that the tree is a path (a unary tree where each node has at most one child), and its depth could be 10^5. In fact the test case #2 is a path.
A stack of depth 10^5 should be an overflow right? I have not used any linker settings to increase the stack size.
So what I'm wondering is: is it just a missing test case? Or does codeforces compilers do something special so that we don`t get a stack overflow even at depth 10^5?
I`m wondering this, because lots of times, it is easier to code the recursive DFS solution than iterative BFS solution. If possible I wish not to care about the stack-overflows.
Thank you in advance.
Default stack size in codeforces is 256MB
That explains it. Thank you very much :)
So a typical codeforces problem will have the memory limit of 256MB. It is heap memory right? So in total we have 256MB heap + 256MB stack = 512 MB memory?
No,we don't.12693743 shows that stack memory also counts. (memory shown:262144KB=256MB,while this program only used 150MB heap memory)
nice practical proof! thank you :)
this post by MikeMirzayanov says that the g++ compile option is
g++.exe -static -DONLINE_JUDGE -lm -s -x c++ -Wl,--stack=268435456 -O2 -o {filename}.exe {file}
in that,-Wl,--stack=268435456
says that the stack size is 268435456 bytes(=256MB) which is far more than what your program will use. (In worse case, your program will use 100000*3*4=1200000 bytes)