hash_sparse_vector.cpp

SGSparseVector<float64_t> hash_vector(SGSparseVector<float64_t> vec, int32_t dim,
	uint32_t seed)
{
	SGVector<float64_t> h_vec(dim);
	h_vec.zero();

	for (index_t i=0; i<vec.num_feat_entries; i++)
	{
		uint32_t hash = CHash::MurmurHash3((uint8_t* ) &vec.features[i].feat_index, sizeof (index_t),
						seed);
		h_vec[hash % dim] += vec.features[i].entry;
	}

	int32_t num_nnz_features = 0;
	for (index_t i=0; i<dim; i++)
	{
		if (h_vec[i]!=0)
			num_nnz_features++;
	}

	SGSparseVector<float64_t> sv(num_nnz_features);

	int32_t sparse_index = 0;
	for (index_t i=0; i<dim; i++)
	{
		if (h_vec[i]!=0)
		{
			sv.features[sparse_index].entry = h_vec[i];
			sv.features[sparse_index++].feat_index = i;
		}
	}

	return sv;
}