| |
| |
| |
|
Page: 1 2 3 4 5 6 7
Comments:
<0> what type of printing binary trees deffine it in the only way
<1> can u plz...?
<2> can i subtract/divide/multiply std::vector::iterator s ? is iterator arithmetic supported beyond plus?
<3> yes
<1> i wonna write to a binary file..
<3> they're random access iterators
<4> dextre, subtract
<5> SSSsssSSS nope, you are specifically banned
<1> i wonna write to a binary file..
<3> er
<3> not multiply and divide rthough
<2> Cowmoo: i got 2 iterators i want to get the average position of.. (a+b)/2
<4> what would divide and multiply mean for an iterator/pointer ?
<5> and they probably banned you because you're irritating
<3> what would division even mean?
<3> use std::distance dextre
<0> what type of printing binary trees deffine it in the only way
<2> Cowmoo: i'm using "equal_range", which gives 2 iterators marking a range,
<3> but I dunno, what you're doing seems suspect
<2> Cowmoo: thanks, i like that
<5> dextre why do you think you need iterator arithmetic anyhow?
<2> varjr: yeah, i don't think it's something i would implement into vector::iterator myself.. i happen to need the average position returned by std::equal_range, which does a binary search
<2> std::distance sounds good
<5> dextre I presume your container is sorted
<2> yeah, it's sorted ;)
<2> i chose equal_range because there might be duplicate elements, in which case i'd like the avg position
<5> what are you going to do if distance is even
<2> isn't even? then i'd round down
<5> and why do you need the position of the "middle" of the range
<5> IS even odd actually had a middle
<2> you're right
<1> i wonna write to a binary file.. i did this: http://www.noidea128.org/sourcefiles/16262.html
<2> varjr: round down then
<1> but when I open data.bin it was writen rgular not binary... why is that?
<5> why do you want the middle, instead of the beginning or end
<2> i'd like to build a function that finds k nearest neigbors in a list of vectors, and do it very fast..
<1> help me.. ok.?
<5> SSSsssSSS that's pure C...not interested in helping you
<2> so i figure i'd sort the positions in each dimension
<3> haha dextre
<3> AI ***ignment?
<2> no. thesis
<3> oh
<3> I had to write a k-NN implementation couple of weeks ago
<5> dextre dextre what are you "nearest neighboring" on ?
<2> i'm borrowing this trick from axis-aligned bounding box collision detection i read about some time ago
<2> the fastest way to get kNN seems to be this
<1> so where to?
<1> get help beside #c?
<2> u sort once, then query time becomes m log n, with n dimension vectors, and m of them
<1> cause I'm banned overthere..
<5> SSSsssSSS yes, and I know the person who banned you,,,,,, you must have been a real pest
<6> Well, other than repeating the same sentence about 15 times within 10 seconds, he's almost fine :p
<2> varjr: i'm nearest neiboring on a vector of m points with n dimensions
<5> are you clustering things? or you have a specific place you want to be near to in n-space?
<2> i actually need to have the neighbors, k of them
<7> please i need a good documentation about how using API on MSVC
<5> for a specific point? or what??
<6> G3M: MSDN
<2> neighbors of a position where there may or may not be a point
<2> the sort need only be done once
<5> a point exists.....doesn't need anything else
<7> MrAshe : yeah but i found nothing :/
<7> there search engine ****s
<5> so you have n coordinates and you want the k nearest things to it
<2> and for queries the rest is just binary search on each dimension
<2> yeah
<5> you going to then search the subset that is k*n elements?
<2> i COULD build a distance matrix of the entire dataset
<2> and sort a single row of that matrix to yield the k nearest neighbors in log n steps
<5> dextre that's expensive you ned up w/ m^2/2 entries in it
<5> m being the number of elements in the entire collection
<1> where I can get help in #c?
<1> where I can get help in c?
<1> not ic channel #c
<1> other channel..
<1> knows?
<2> vaw: yeah, but query time will be very fast, and it can even be m Time, not log n
<5> log M
<2> yeah
<5> so you have to decide whether the sort is worth it...... depends on the query vs data-changes ratio
<5> btw, when I'm doing that kind of thing I never do sqrt distance squared is good for comparing also
<5> and it makes the math quite a bit faster
<2> well, right now, i'm maintaining N sorted lists, 1 per dimension.. the distance matrix has M^2 / 2 terms
<2> vawjr: good idea, thanks
<2> each sorted list has M entries
<2> so it's space(N*M) vs space(M^2/2)
<2> steps(N log M) vs steps(M)
<2> tough call
<5> don't forget the time needs to be factored in
<2> time needs to be factored in?
<5> yeah, if you're measuring the cost of different approaches
<2> implementation time?
<5> implementation + a*number of times it's run
<5> a being execution time
<2> implementation time aside, which do u like better?
<5> if this is only gonna run once that's a big differnce between running every millisecond
<2> axis sorting, or distance matrix
<5> how many times is it gonna run...... distance 1/2 matrix requires (after the equal-range call) at most k looks
<5> and NO distance calculations to perform....straight compares
<5> asix requires calculating k*n distances for each search
<2> query(time(N log M)) vs query(time(M))
<2> init(time(N* M log M)) vs init(time(M log M))
<2> space(N*M) vs space(M^2/2)
<2> query for either require no distance calculations,
<2> but building the distance matrix does, extensively, and the sorted list approach doesn't
<0> hello
<8> hi
<0> i have postorder print of the binar tree how can i print the tree from it ?
<0> do you have any idea ?
<2> left->print(); right->print(); this->print();
<0> not its not good
<2> yes it IS
<3> fozo: this is a comon algorithm, look it up online
<2> * preorder */ this->print(); left->print(); right->print();
<9> this->print() should be this->printSelf();
<9> otherwise it's just endless recursion
<0> looked not found on goodle
<2> yeah,
<2> goodle
<2> lol
<3> lol
<2> yodle
<0> :)
<9> http://en.wikipedia.org/wiki/Tree_traversal
<2> is a distance matrix really that bad?
<2> it has attractive time complexity
<3> sure why not
<3> if the space complexity is no bother
<10> hmm
<10> damnation, I need a better vector art program than inkscake
<2> what's a fast way to get the N smallest values in a list?
<2> keeping the complexity as close to N as possible
<3> you could do O(n), where n is the size of your list
<3> that's the fastest
<2> yeah,
<2> but how do i implement this
<3> use the Select algorithm
<3> find the median Nth element, and use Partition with it as the pivot
<2> ah
<2> so it's like binary search
<3> no
<3> well
<3> in the sense that they're both divide-and-conquer algorithms I suppose?
<2> std::nth_element ?
<3> ohh, you were tlaking C++
<3> haha yea, that'd do
<2> what an awesome coincidence
<2> stumbled onto the right function
<2> is nth_element fast?
Return to
#c++
or
Go to some related
logs:
What play contains the line 'something is rotten in the state of Denmark' ?
how to manstrubate
#linux
#AllNiteCafe
zic skype
apt-remove --purge
#linux
volleyball tips/instructions
lymaria
bibanul
|
|