Working with Vector Search

            // Create a vector index configuration for indexing 3 dimensional vectors embedded
            // in the documents' key named "vector" using 2 centroids.
            var config = new VectorIndexConfiguration("color", 3, 100);

            // Create a vector index named "color_index" using the configuration
            var collection = database.GetCollection("colors");
            collection.CreateIndex("colors_index", config);

            // Register the predictive model named "ColorModel".
            Database.Prediction.RegisterModel("ColorModel", new ColorModel());

            // Create a vector index configuration with an expression using the prediction
            // function to get the vectors from the registered predictive model.
            var expression = "prediction(ColorModel, {\"colorInput\": color}).vector";
            var config = new VectorIndexConfiguration(expression, 3, 100);

            // Create a vector index from the configuration
            var collection = database.GetCollection("colors");
            collection.CreateIndex("colors_index", config);

            // Retrieve the index you wish to update
            var index = collection.GetIndex("index-name");

            // Start an update on it (in this case, limit to 50 entries at a time)
            var updater = index.BeginUpdate(50);

            // If updater is null, that means there are no more entries to process
            while (updater != null) {
                using (updater) {
                    // Otherwise, the updater will contain a list of data that needs embeddings generated
                    int i = 0;
                    foreach (var entry in updater) {
                        // The type of entry will depend on what you have set as your index.
                        // In this example, we will assume it was set to a string property.
                        // Let's also assume that if an embedding is not applicable, this
                        // pseudo function returns null
                        try {
                            var embedding = await GenerateEmbedding((string)entry);
                            if (embedding == null) {
                                // No embedding applicable.  Calling SetVector will null will
                                // cause the underlying document to NOT be indexed
                                updater.SetVector(i, null);
                            } else {
                                // Yes this if/else is unneeded, and only to demonstrate the
                                // effect of setting null in SetVector
                                updater.SetVector(i, embedding);
                            }
                        } catch (Exception) {
                            // Bad connection?  Corrupted over the wire?  Something bad happened
                            // and the embedding cannot be generated at the moment.  So skip
                            // this entry.  The next time BeginUpdate is called, it will be considered again
                            updater.SkipVector(i);
                        }
                    }

                    // This writes the vectors to the index.  Disposing it without calling this
                    // will throw out the results without saving.  You MUST have either set or
                    // skipped all the entries inside of the updater or this call will throw an exception.
                    updater.Finish();
                }

                // Ready for the next batch!
                updater = index.BeginUpdate(50);
            }

            // Combining vector search logic with other logic makes a hybrid query
            var sql = """
                      SELECT color 
                      FROM colors 
                      WHERE APPROX_VECTOR_DISTANCE(vector, $vector) < 0.5 AND group = 'group1'
                      LIMIT 10
                      """;

            var query = database.CreateQuery(sql);

            // Setting the vector ommitted for brevity (see other examples)

            // To avoid an accidental resource intensive exhaustive search of the
            // database, a LIMIT clause it required for non-hybrid vector search
            // queries

            // This will work
            var sql = "SELECT id, color, approx_vector_distance(vector, $vector) " +
                      "FROM _default.colors " +
                      "LIMIT 8";

            var query = database.CreateQuery(sql);

            // This will not
            sql = "SELECT id, color, approx_vector_distance(vector, $vector) " +
                      "FROM _default.colors ";

            query = database.CreateQuery(sql);

            // The following shows using Full Text Search and Vector Search
            // in the same query.  Note that this example doesn't show
            // setting up the full text index 'desc-index'.
            var sql = """
                      SELECT color 
                      FROM colors 
                      WHERE MATCH(desc-index, $text) AND group = 'group1'
                      ORDER BY APPROX_VECTOR_DISTANCE(vector, $vector)
                      """;

            var query = database.CreateQuery(sql);