In article <11************ ********@e56g20 00cwe.googlegro ups.com>,
ki***********@g mail.com wrote:
What I want to do is write a program to test lexicographical
combinations of various sets of numbers to a predefined number and
write that true values to a text file.
Example:
Number choosen between 1 and 50
Number set = 6
Value test = 150
Save to = 150.txt
So what happens is you are using 50 numbers in sets of 6 to test the
sets to equal to 150 and
save to file 150.txt
01 02 03 04 05 06 = 21 False -> Discard
01 10 20 30 40 49 = 150 True -> Store number set in text file
Any ideas would be appreciated as I am fairly new to C++ and would like
to make the program flexible to test any numbers and number sets.
Personally I would start by writing a generic algorithm with the
following signature and semantics:
template<class Function, class BidirectionalIt erator, class Size>
Function
for_each_combin ation(Bidirecti onalIterator first,
BidirectionalIt erator last,
Size k, Function f);
Requires: [first, last) is a valid range.
Effects: For each combination of elements in [first, last) taken k at a
time, permutes those elements into the range [first, first+k) and calls
f(first, first+k). If k < 1, no calls to f are made. If k >
last-first, it is truncated to last-first.
Notes: The effects clause does not imply random access. Iterator
addition used only for concise statements.
Returns f.
This is a valuable general purpose tool.
Then I would write a test functor more geared to your problem which has
an operator taking a range of int's, sums them up (perhaps using
std::accumulate ), compares the sum to the proper number (perhaps given
during the functor's constructor), and if equal, process the desired
range in the desired way.
Finally a driver program could create the array, and call
for_each_combin ation with the appropriately created test functor.
The implementation of for_each_combin ation is definitely the hard part
in this exercise. But fwiw, there are 183723 combinations of numbers in
the range [1, 50], where six of them add up to 150 (out of a total of
nearly 16 million combinations).
There are cheaper algorithms for this particular computation. But
having the for_each_combin ation function in your tool box is worth the
time creating it. It is much more general, and if nothing else, would
help you test a more specialized but higher performing algorithm for
this problem.
-Howard
