Struct ColumnStats 

pub struct ColumnStats {
    pub col_idx: usize,
    pub dtype: ScalarType,
    pub null_count: u64,
    pub distinct_estimate: u64,
    pub min_value: Option<i64>,
    pub max_value: Option<i64>,
    pub avg_width: Option<f32>,
}

Per-column statistics for optimizer cost estimation.

Tracks null counts, distinct value estimates, and value ranges for columns. These statistics enable the optimizer to estimate filter selectivity and join cardinalities.

Fields

col_idx: usize

Column index within the relation

dtype: ScalarType

Data type of the column

null_count: u64

Count of null values (for nullable columns)

distinct_estimate: u64

HyperLogLog-style distinct value estimate

min_value: Option<i64>

Minimum value (for orderable types, stored as i64)

max_value: Option<i64>

Maximum value (for orderable types, stored as i64)

avg_width: Option<f32>

Average value length for variable-length types (e.g., symbols)

Implementations

impl ColumnStats

pub fn new(col_idx: usize, dtype: ScalarType) -> Self

Creates new column statistics with default values.

Arguments

• col_idx - The column index within the relation
• dtype - The scalar type of the column

Returns

A new ColumnStats instance with zero counts and no range information.

pub fn update_distinct(&mut self, estimate: u64)

Updates the distinct value estimate.

This should be updated from HyperLogLog or similar cardinality estimation algorithms running on the GPU.

Arguments

• estimate - The new distinct value estimate

pub fn update_range(&mut self, min: i64, max: i64)

Updates the value range for this column.

Arguments

• min - The minimum value (encoded as i64)
• max - The maximum value (encoded as i64)

pub fn update_null_count(&mut self, count: u64)

Updates the null count for this column.

Arguments

• count - The number of null values

pub fn update_avg_width(&mut self, width: f32)

Updates the average width for variable-length columns.

Arguments

• width - The average value width in bytes

pub fn equality_selectivity(&self, total_rows: u64) -> f64

Estimates selectivity for an equality predicate.

Uses the distinct value count to estimate selectivity. If no distinct count is available, returns a default estimate.

Arguments

• total_rows - The total number of rows in the relation

Returns

The estimated selectivity (0.0 to 1.0)

pub fn range_selectivity(&self, low: i64, high: i64) -> f64

Estimates selectivity for a range predicate.

Uses min/max values to estimate what fraction of the range is covered. Returns a default estimate if range statistics are unavailable.

Arguments

• low - The lower bound of the range (inclusive)
• high - The upper bound of the range (inclusive)

Returns

The estimated selectivity (0.0 to 1.0)

pub fn value_size_bytes(&self) -> usize

Returns the storage size per value for this column type.

Trait Implementations

impl Clone for ColumnStats

fn clone(&self) -> ColumnStats

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for ColumnStats

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.