Re: challenge (simplyfying quicksort code)

On 27.03.2024 12:04, fir wrote:

the chellenge is simplify such quicksort code as far as you can
 
conditions
1) it should work proper way
2) should be not slower or only slightly slower
(so thsi is not optimistation for speed its more for being simple but
staing reasonably fast)
3) by simplify i mean maybe removing soem if or some other
element but not necassary rewriting names for shorter etc as names are
not much meaningfull here  (so by simplifying i mainy understand
removing something or some recomposition for something better/clearer)
4) the basic quicksort has two calls inside this one is
opitmised to have only one , i think both versions are okay
but maybe thsi one with one call inside is maybe better
 
 
void QuicksortInts7(int* table, int lo, int hi)
{
 
    int i=lo; int j=hi;
 
    while (i<hi)
    {
        int pivot =table[(lo+hi)/2];
        while (i<=j)          // Partition
        {
            while (table[i]<pivot) i++;
            while (table[j]>pivot) j--;
            if (i<=j)
            {
                int t=table[i];table[i]=table[j];table[j]=t;
                i++; j--;
            }
        }
 
        if (lo < j){ QuicksortInts7(table, lo, j);} //recursion
        lo=i; j=hi;
    }
}
 
 
is there a way to simplify this?

Your conditions listed above appear to be quite fuzzy and I think your
view of "simplification" is a very subjective one, so it's probably
impossible to provide an answer that suits you. But I can give you my
own opinions below oriented on some keywords/conditions you used...

Simplicity is something that keeps the intention of the algorithm
clear, so I'd stay with two QS() calls, as in the original algorithm.
(Your point 4.)

Your algorithm has worst complexity O(N^2), so it's quite meaningless
to speak about "slightly slower" or "reasonable fast". (Your point 2.)
If folks want some speed guarantee then they'd enhance the algorithm
by well established means[*] (which would require more code, though,
so likely goes against your "challenge").

I wouldn't trust any home-brewed algorithm, especially if (as in this
case) you changed or omitted conditions from the original algorithm.
(See your own follow-up in this thread.) The least required would be
the preconditions stated (e.g. is 'hi' the index of the last element
in the table or the subsequent element). And the code should have
sufficient comments to be able to work on it, in the first place. So
given that "it should work proper way" is hard to [easily] verify.
(Your point 1.)

PS: I would suggest that you try to formally or analytically verify
whether your variant is working correctly. (Or check it with *lots*
of test data.)

Janis

[*] For example, the Pascal code of the CDC mainframe library I used
40 years ago already had median-of-3 pivot selection, and linear sort
implemented for data partition subsets of less than 10 elements.