You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

151 lines
3.8 KiB
C++

#include "kcnf.hpp"
#include <algorithm>
#include <filesystem>
#include <fstream>
#include <iostream>
#include <random>
#include <stdexcept>
#include <unordered_set>
#include <vector>
#ifdef _OPENMP
#include <omp.h>
#endif
namespace fs = std::filesystem;
const fs::path CNF_DIR = "./CNF";
long binom(long n, long k) {
return (k > n) ? 0 :
(k == 0 || k == n) ? 1 :
(k == 1 || k == n - 1) ? n :
(k + k < n) ?
(binom(n - 1, k - 1) * n) / k :
(binom(n - 1, k) * n) / (n - k);
}
// get the (n choose k) possible variable choices at index idx
// example: unrank_combination(5,3,5) -> (1,4,5)
// the full combination list would look like: (1,2,3), (1,2,4), (1,2,5), (1,3,4), (1,3,5), (1,4,5) <- this is the combination on index 5
std::vector<int> unrank_combination(long n, long k, long long idx) {
std::vector<int> comb;
comb.reserve(k);
long start = 1;
for (long i = 0; i < k; i++) {
for (long v = start; v <= n; v++) {
long long count = binom(n - v, k - i - 1);
if (idx < count) {
comb.push_back((int)v);
start = v + 1;
break;
} else {
idx -= count;
}
}
}
return comb;
}
// sample c indices in [0,total] for our c = # of clauses
// this is the heart of the program as we are flattening the problem with our sampled indices and the unrank function!
std::vector<long long> sample_indices(long long total, long c) {
std::unordered_set<long long> set;
set.reserve(c);
std::random_device rd;
std::mt19937 gen(rd());
std::uniform_int_distribution<long long> dist(0, total - 1);
while ((long)set.size() < c) {
set.insert(dist(gen));
}
return std::vector<long long>(set.begin(), set.end());
}
// decode one clause from global index (think of it like a hash function for better understanding!)
// idx = combo_id * 2^k + mask
Clause decode_clause(long n, long k, long long idx) {
long long combo_id = idx >> k;
int mask = idx & ((1 << k) - 1);
std::vector<int> comb = unrank_combination(n, k, combo_id);
Clause cl;
cl.reserve(k);
for (int i = 0; i < k; i++) {
int lit = comb[i];
cl.push_back((mask & (1 << i)) ? lit : -lit);
}
return cl;
}
// parallel iteration to build clauses from indices
std::vector<Clause> build_clauses_parallel(
long n,
long k,
const std::vector<long long>& indices) {
std::vector<Clause> result(indices.size());
#pragma omp parallel for schedule(dynamic)
for (long i = 0; i < (long)indices.size(); i++) {
result[i] = decode_clause(n, k, indices[i]);
}
return result;
}
KCNF KCNF::generate_random(long n, long c, long k) {
KCNF f;
f.n = n;
f.k = k;
// test for logical limits
long long total = binom(n,k) * (1LL << k);
if (c > total) {
throw std::runtime_error("Too many clauses requested.");
}
// test for hardware limits
long long estimated_bytes =
c * (sizeof(Clause) + k * sizeof(int));
if (estimated_bytes > 500 * 1024 * 1024) { // 500MB
throw std::runtime_error("Estimated memory too large");
}
auto indices = sample_indices(total, c);
f.clauses = build_clauses_parallel(n, k, indices);
return f;
}
bool KCNF::write_dimacs(const std::string& filename) const {
fs::create_directories(CNF_DIR);
fs::path path = CNF_DIR / filename;
path = path.lexically_normal();
std::ofstream out(path);
if (!out) {
std::cerr << "Error: Could not open file " << path << "\n";
return false;
}
out << "p cnf " << n << " " << clauses.size() << "\n";
for (const auto& cl : clauses) {
for (int lit : cl) {
out << lit << " ";
}
out << "0\n";
}
return true;
}