Function TestSummaryInvariants

tests/cpp/common/test_quantile.cc:24–66

namespace {
void TestSummaryInvariants(SummaryCase const& c, WQSummaryContainer const& summary,
                           GeneratedColumn const& col) {
  auto entries = summary.Entries();
  auto ref = AggregateReferenceColumn(col);
  auto nonzero_samples = NonZeroWeightCount(col);
  auto budget = SketchSummaryBudget(c.max_bin, c.rows);
  // An empty sketch should remain empty after finalization.
  if (EmptyReference(ref)) {
    ASSERT_TRUE(entries.empty()) << "case=" << c.name;
    return;
  }

  // A numerical summary should remain a strictly increasing support set.
  ASSERT_FALSE(entries.empty()) << "case=" << c.name;
  for (std::size_t i = 1; i < entries.size(); ++i) {
    EXPECT_LT(entries[i - 1].value, entries[i].value) << "case=" << c.name;
  }

  // Large-n anchors should exercise actual compression rather than exact retention.
  if (c.rows > static_cast<std::size_t>(c.max_bin) * 8) {
    ASSERT_LT(summary.Size(), nonzero_samples)
        << "case=" << c.name << " should exercise sketch compression.";
  }

  // The summary query rule should satisfy the target rank bound plus the final prune term.
  auto total = TotalWeight(ref);
  auto max_error = MaxSummaryQueryRankError(summary, ref, c.max_bin);
  auto eps = SketchEpsilon(c.max_bin, c.rows);
  auto bound = (eps + 1.0 / static_cast<double>(budget)) * total;

  EXPECT_LE(max_error, bound) << "case=" << c.name << ", total=" << total << ", budget=" << budget
                              << ", eps=" << eps;

  // If the target bin count can already represent all distinct values, the summary should
  // preserve the exact support instead of approximating it.
  if (UniqueValueCount(ref) <= static_cast<std::size_t>(c.max_bin)) {
    auto exact_values = ExactValues(ref);
    ASSERT_EQ(entries.size(), exact_values.size()) << "case=" << c.name;
    for (std::size_t i = 0; i < exact_values.size(); ++i) {
      EXPECT_FLOAT_EQ(entries[i].value, exact_values[i]) << "case=" << c.name;
    }
  }
}
void AssertSameOnAllWorkers(Context const* ctx, HistogramCuts const& cuts) {
  auto const world = collective::GetWorldSize();
  if (world <= 1) {