B.cpp

// #pragma GCC optimize("Ofast,unroll-loops")
// #pragma GCC target("avx,avx2,fma")

#include <bits/stdc++.h>
using namespace std;

#define ll long long
#define ull unsigned long long
#define dd double
#define ld long double
#define sl(n) scanf("%lld", &n)
#define si(n) scanf("%d", &n)
#define sd(n) scanf("%lf", &n)
#define pll pair <ll, ll>
#define pii pair <int, int>
#define mp make_pair
#define pb push_back
#define all(v) v.begin(), v.end()
#define inf (1LL << 62)
#define loop(i, start, stop, inc) for(ll i = start; i <= stop; i += inc)
#define for1(i, stop) for(ll i = 1; i <= stop; ++i)
#define for0(i, stop) for(ll i = 0; i < stop; ++i)
#define rep1(i, start) for(ll i = start; i >= 1; --i)
#define rep0(i, start) for(ll i = (start-1); i >= 0; --i)
#define ms(n, i) memset(n, i, sizeof(n))
#define casep(n) printf("Case %lld:", ++n)
#define pn printf("\n")
#define pf printf
#define EL '\n'
#define fastio std::ios_base::sync_with_stdio(false);cin.tie(NULL);cout.tie(NULL);

const ll sz = 1e3 + 10, sz2 = 1e6 + 10;
ll mat[sz][sz], rowSel[sz2],  colSel[sz2];

priority_queue <ll> row, col;

int main()
{
    ll n, m, k, p;
    sl(n), sl(m), sl(k), sl(p);

    for1(i, n)
        for1(j, m)
            sl(mat[i][j]);


    for1(i, n) {
        ll sum = 0;
        for1(j, m)
            sum += mat[i][j];

        row.push(sum);
    }

    for1(j, m) {
        ll sum = 0;
        for1(i, n)
            sum += mat[i][j];

        col.push(sum);
    }

    for1(i, k) {
        ll now = row.top();
        row.pop();

        rowSel[i] = rowSel[i-1] + now;

        row.push(now - m*p);
    }

    for1(i, k) {
        ll now = col.top();
        col.pop();

        colSel[i] = colSel[i-1] + now;

        col.push(now - n*p);
    }


    ll ans = -inf;
    for(ll i = 0; i <= k; i++) {

        ll colTake = k-i, neg = colTake*i*p;

        ans = max(ans, rowSel[i] + colSel[colTake] - neg);
    }

    cout << ans << EL;

    return 0;
}