Discussion @ 1489. Points on a Parallelepiped @ Timus Online Judge

Show all messages Hide all messages

I do simple check to what part does the point below. What's wrong? WA#6. Some tricky tests? Thanks (-) Alexey 15 Oct 2006 11:23

Edited by author 15.10.2006 11:24

Re: I do simple check to what part does the point below. What's wrong? WA#6. Some tricky tests? Thanks (-) Olympic Bear 15 Oct 2006 16:06

Check each region by x and y, e.g.
if (b+c<=y && y<=b+c+b && c <= x && x<=c+a) // bottom region

Thanks. Now I have WA#7 ) Alexey 15 Oct 2006 18:59

I had some mistakes like
If y>=a+b+c ...
but must be
If y>=b+b+c

I have one check per one side.
For example
If y<=b - the point is on the lowest side (on the picture).
If x<=c - the point is on the leftest side.
etc.

So, it is enough to check, you see, but WA#7.

Thanks for help but still need it. )

Re: Thanks. Now I have WA#7 ) Tolstiy_BSU 18 Oct 2006 14:38

You need check
y > b+b+c and y < b+c..

because when point on the edge .... You can do mistake ...
(I was wrong in this)

Re: Thanks. Now I have WA#7 ) Alexey 18 Oct 2006 21:28

I have AC, so thank U very much!
But I still don't understand my mistake.
It is guaranted that the point is in the picture, so
I can only check one coordinate...

Cann't understand...

Re: Thanks. Now I have WA#7 ) Midnight_Kitty 27 Nov 2006 01:11

be careful when you do checks like
if (x <= a) then ...
it can produce WA, becouse if x = a, point can lie on another side

Re: Thanks. Now I have WA#7 ) Paul Diac 21 Sep 2007 02:23

Yes, that is true.
Thank you.
The test that makes the difference is:
-------
2 2 2
0 4
4 0
-------
before this correction the answer was 4.8989794856 (wrong!)
but the right answer is 2.8284271247 (verifyed with AC program).

Re: Thanks. Now I have WA#7 ) UXMRI: Sergey Baskakov, Raphail Akhmedisheff and Denis Nikonorov 26 Jun 2009 17:22

This is great test. Thank you!

Re: Thanks. Now I have WA#7 ) sergovoy 30 Oct 2017 01:20

TY, Paul Diac!

Edited by author 30.10.2017 01:21

no mes SuperLight 26 Jun 2009 19:31

Re: Thanks. Now I have WA#7 ) Partisan 26 Jun 2009 17:40

If you use real numbers, you can get situation, when point lies on none of 6 sides. So you should work with epsilon. Or you can use exact arithmetic, just multiply numbers by 100!

It means 100 and !, not 100! :)

Edited by author 26.06.2009 17:41

Discussion of Problem 1489. Points on a Parallelepiped