Struct Statistics 

pub struct Statistics {
    pub computed_mask: StatisticsMask,
    pub nnz: u64,
    pub sparsity_ratio: f64,
    pub value_min: f64,
    pub value_max: f64,
    pub value_abs_max: f64,
    pub value_mean: f64,
    pub value_stddev: f64,
    pub nm_n: u8,
    pub nm_m: u8,
    pub has_nan: bool,
    pub has_inf: bool,
}

Advisory statistics about a tensor’s data buffer.

Statistics reflect the tensor data at write time. A reader MUST NOT rely on any statistic for correctness — they are optimization hints only (algorithm selection, memory pre-allocation, routing decisions).

Each group of fields is guarded by a bit in computed_mask. A reader MUST check the relevant bit via StatisticsMask before using any field.

Note: derives PartialEq but NOT Eq — f64 NaN semantics make Eq unsound.

Examples

use hurray_core::descriptor::{Statistics, StatisticsMask};

let stats = Statistics {
    computed_mask: StatisticsMask(StatisticsMask::NNZ_VALID),
    nnz: 42,
    sparsity_ratio: 0.0,
    value_min: 0.0,
    value_max: 0.0,
    value_abs_max: 0.0,
    value_mean: 0.0,
    value_stddev: 0.0,
    nm_n: 0,
    nm_m: 0,
    has_nan: false,
    has_inf: false,
};
assert!(stats.computed_mask.nnz_valid());
assert_eq!(stats.nnz, 42);

Fields

computed_mask: StatisticsMask

Bitmask indicating which fields contain valid data.

nnz: u64

Number of non-zero elements. Valid when NNZ_VALID is set.

sparsity_ratio: f64

Fraction of zero elements [0.0, 1.0]. Valid when SPARSITY_VALID is set.

value_min: f64

Minimum element value (dequantized). Valid when VALUE_RANGE_VALID is set.

value_max: f64

Maximum element value (dequantized). Valid when VALUE_RANGE_VALID is set.

value_abs_max: f64

Maximum absolute element value. Valid when VALUE_RANGE_VALID is set.

value_mean: f64

Arithmetic mean of all elements. Valid when VALUE_STATS_VALID is set.

value_stddev: f64

Population standard deviation. Valid when VALUE_STATS_VALID is set.

nm_n: u8

N in N:M structured sparsity. Valid when NM_SPARSITY_VALID is set.

nm_m: u8

M in N:M structured sparsity. Valid when NM_SPARSITY_VALID is set.

has_nan: bool

true if at least one NaN element is present. Valid when NAN_INF_VALID is set.

has_inf: bool

true if at least one infinity is present. Valid when NAN_INF_VALID is set.

Trait Implementations

impl Clone for Statistics

fn clone(&self) -> Statistics

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for Statistics

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

Formats the value using the given formatter.

impl PartialEq for Statistics

fn eq(&self, other: &Statistics) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl StructuralPartialEq for Statistics