#line 2 "lib/prelude.hpp"
// #ifndef LOCAL
// #pragma GCC optimize("O3")
// #endif
#include <bits/stdc++.h>
using namespace std;
using ll = long long;
using vi = vector<int>;
using vvi = vector<vector<int>>;
using vll = vector<ll>;
using vvll = vector<vector<ll>>;
using vc = vector<char>;
#define rep2(i, m, n) for (auto i = (m); i < (n); i++)
#define rep(i, n) rep2(i, 0, n)
#define repr2(i, m, n) for (auto i = (n); i-- > (m);)
#define repr(i, n) repr2(i, 0, n)
#define all(x) begin(x), end(x)
template <class T> auto ndvec(int n, T e) { return vector(n, e); }
template <class... Ts> auto ndvec(int n, Ts... e) { return vector(n, ndvec(e...)); }
#if __cpp_lib_ranges
namespace R = std::ranges;
namespace V = std::views;
#endif
#line 3 "lib/io.hpp"

struct int1 {
  int val; int1(int a = 1) : val(a - 1) {}
  operator int() const { return val; }
};

template <size_t BufSize = 1 << 26> class stdin_reader {
 public: stdin_reader() { buf[fread(buf, 1, sizeof(buf), stdin)] = 0; } template <class T> enable_if_t<is_integral_v<T>> read(T& x) { skip(); [[maybe_unused]] bool neg = false; if constexpr (is_signed_v<T>) neg = *p == '-' ? (p++, true) : false; x = 0; while (*p > ' ') x = x * 10 + (*p++ & 0x0F); if constexpr (is_signed_v<T>) x = neg ? -x : x; } template <class T> void_t<decltype(&T::val)> read(T& x) { x = T((unsigned)(*this)); } void read(char &c) { skip(); c = *p++; } void read(char*& q) { skip(); q = p; while (*p > ' ') p++; *p = 0; } template <size_t N> void read(char (&s)[N]) { read(s); } void read(string& s) { skip(); char* p0 = p; while (*p > ' ') p++; s.assign(p0, p); } template <class T, void_t<decltype(tuple_size<T>::value)>* = nullptr> void read(T& x) { read_tuple_impl(x, make_index_sequence<tuple_size_v<T>>{}); } template <class T, class U> void read(pair<T, U>& x) { read(x.first), read(x.second); } template <class T, size_t N> void read(T (&a)[N]) { for (auto& e : a) read(e); } template <class T> operator T() { T x; return read(x), x; } template <class... Ts> void operator()(Ts&... xs) { (read(xs), ...); } int operator--() { return (int)*this - 1; } template <class T> T* arr(int n) { T* p = new T[n + 1]; rep(i, n) read(p[i]); return p; } template <class T> void vec(vector<T>& v, int n) { v.resize(n); for (auto& e : v) read(e); } template <class T> vector<T> vec(int n) { vector<T> v; return vec(v, n), v; } auto vi(int n) { return vec<int>(n); } auto vi1(int n) { auto v = vec<int>(n); rep(i, n) v[i]--; return v; } auto vll(int n) { return vec<ll>(n); } template <class... Ts> tuple<vector<Ts>...> vecs(int n) { tuple<vector<Ts>...> res; vecs_impl(res, n, make_index_sequence<sizeof...(Ts)>{}); return res; } template <class T> void vvec(vector<vector<T>>& v, int n, int m) { v.resize(n); for (auto& e : v) vec(e, m); } template <class T> vector<vector<T>> vvec(int n, int m) { vector<vector<T>> v; return vvec(v, n, m), v; } template <class... Ts> auto cols(int n) { return transpose(vec<tuple<Ts...>>(n)); } private: char buf[BufSize], *p = buf; void skip() { while (*p <= ' ') p++; } template <class T, size_t... Is> void read_tuple_impl(T& x, index_sequence<Is...>) { (*this)(get<Is>(x)...); } template <class T, size_t... Is> void vecs_impl(T& x, int n, index_sequence<Is...>) { (vec(get<Is>(x), n), ...); } template <class T, size_t... Is> static auto transpose_impl(const vector<T>& v, index_sequence<Is...>) { tuple<vector<decay_t<tuple_element_t<Is, T>>>...> w; (get<Is>(w).reserve(v.size()), ...); for (const auto& row : v) (get<Is>(w).push_back(get<Is>(row)), ...); return w; } template <class T> static auto transpose(const vector<T>& v) { return transpose_impl(v, make_index_sequence<tuple_size_v<T>>{}); }
};
template <size_t BufSize = 1 << 26> class stdout_writer {
 public: ~stdout_writer() { flush(); } void flush() { fwrite(buf, 1, p - buf, stdout), p = buf; } void write_char(char c) { *p++ = c; } void write() {} void write(char c) { write_char(c); } template <class T> enable_if_t<is_integral_v<T>> write(T x) { if (!x) return write_char('0'); if constexpr (is_signed_v<T>) if (x < 0) write_char('-'), x = -x; static char tmp[16]; char* q = end(tmp); while (x >= 10000) memcpy(q -= 4, digits.data + x % 10000 * 4, 4), x /= 10000; if (x < 10) write_char('0' + x); else if (x < 100) write_char('0' + (uint8_t)x / 10), write_char('0' + (uint8_t)x % 10); else if (x < 1000) memcpy(p, digits.data + x * 4 + 1, 3), p += 3; else memcpy(p, digits.data + x * 4, 4), p += 4; memcpy(p, q, end(tmp) - q), p += end(tmp) - q; } template <class T> void_t<decltype(&T::val)> write(T x) { write(x.val()); } void write(double x) { static char tmp[40]; sprintf(tmp, "%.15f", x); write(tmp); } void write(long double x) { static char tmp[40]; sprintf(tmp, "%.15Lf", x); write(tmp); } void write(const char* s) { while (*s) *p++ = *s++; } void write(const string& s) { memcpy(p, s.c_str(), s.size()), p += s.size(); } template <class T, class U> void write(const pair<T, U>& x) { write(x.first), write_char(' '), write(x.second); } template <class... Ts> void write(const tuple<Ts...>& x) { write_tuple(x, make_index_sequence<sizeof...(Ts)>{}); } template <class... Ts> void write(const Ts&... xs) { ((write(xs), write_char(' ')), ...), --p; } template <class... Ts> void writeln(const Ts&... xs) { write(xs...), write_char('\n'); } template <class... Ts> void operator()(const Ts&... xs) { writeln(xs...); } template <class It> void iter(It first, It last, char sep = ' ') { if (first == last) write_char('\n'); else { while (first != last) write(*first++), write_char(sep); p[-1] = '\n'; } } template <class It> void iter1(It first, It last, char sep = ' ') { if (first == last) write_char('\n'); else { while (first != last) write(1 + *first++), write_char(sep); p[-1] = '\n'; } } template <class T> void vec(const vector<T>& v, char sep = ' ') { iter(all(v), sep); } template <class T> void vec1(const vector<T>& v, char sep = ' ') { iter1(all(v), sep); } void del() { *--p = 0; } void Yes(bool b = true) { writeln(b ? "Yes" : "No"); } void YES(bool b = true) { writeln(b ? "YES" : "NO"); } void Takahashi(bool b = true) { writeln(b ? "Takahashi" : "Aoki"); } private: char buf[BufSize], *p = buf; template <class T, size_t... Is> void write_tuple(const T& x, index_sequence<Is...>) { ((write(get<Is>(x)), write_char(' ')), ...), --p; } struct four_digits { char data[40000]; constexpr four_digits() : data() { for (int i = 0; i < 10000; i++) for (int n = i, j = 4; j--;) data[i * 4 + j] = n % 10 + '0', n /= 10; } } static constexpr digits{}; public:
#define INSTANT(s) void s() { writeln(#s); }
  INSTANT(No) INSTANT(NO) INSTANT(Aoki) INSTANT(possible) INSTANT(Possible) INSTANT(POSSIBLE) INSTANT(impossible) INSTANT(Impossible) INSTANT(IMPOSSIBLE)
#undef INSTANT
};
stdin_reader<> in;
stdout_writer<> out;
#line 3 "lib/num/int/convex.hpp"

// Returns (min f, min argmin f)
template <class T, class F>
auto convex_min(T l, T r, F f) {
  if (l + 1 == r) return pair(f(l), l);

  auto sf = [l, r, &f](T x) {
    return x < r ? f(x) : numeric_limits<T>::max();
  };

  const double PHI = 1.6180339887498948;
  vector<T> fib = {T(1), T(1)};
  fib.reserve(log((r - l + 1) / PHI) + 1);
  while (fib.back() <= r - l) fib.push_back(fib.end()[-1] + fib.end()[-2]);
  int i = fib.size();
  l = l - 1, r = l + fib[--i];
  T x = l + fib[i-2], y = l + fib[i-1];
  auto fx = sf(x), fy = sf(y);
  while (i >= 4) {
    if (fx > fy) {
      l = x;
      tie(x, fx) = pair(y, fy);
      y = l + fib[--i - 1];
      fy = sf(y);
    } else {
      r = y;
      tie(y, fy) = pair(x, fx);
      x = l + fib[--i - 2];
      fx = sf(x);
    }
  }
  return fx > fy ? pair(fy, y) : pair(fx, x);
}
#line 3 "main.cpp"

int main() {
  int n = in, k = in, l = in, h = in;
  int mid = (l + h) / 2;
  auto a = in.vll(n);
  __int128 M = __int128(1) << 68;
  auto [mini, arg_mini] = convex_min(0LL, ll(2e12), [&](ll x) {
    __int128 sum = 0;
    rep(i, n) {
      sum += M * ((max(0LL, x - (h - l) - a[i]) + (k - 1)) / k +
                  (max(0LL, a[i] - x) + (k - 1)) / k) +
             abs(x - (h - mid) - a[i]) * 2;
    }
    return sum;
  });
  out(ll(mini / M));
}