Struct ExponentialAverage 

pub struct ExponentialAverage<Val, Alpha, Beta> { /* private fields */ }

An exponential moving average and variance tracker using fixed-point values.

This structure tracks a value’s exponential moving average and its variance using separate filtering coefficients (alpha for positive changes and beta for negative changes, or a single coefficient for both). Note: If V_FRAC is 0 (integer values), the variance update might lose significant precision or decay to zero prematurely because internal calculations (like beta * delta^2) are rounded to the nearest integer. Users should be encouraged to use types with enough fractional bits for the accumulator if precision in the variance is important.

Additionally, because variance calculations involve squared delta terms (\Delta^2), there is a high risk of saturation. It is strongly recommended to use larger integer types (e.g. i64 or i128) as the underlying type if large input ranges or high variance values are expected.

Implementations

impl<V, const V_FRAC: usize, A, const A_FRAC: usize, B, const B_FRAC: usize> ExponentialAverage<Fixed<V, V_FRAC>, Fixed<A, A_FRAC>, Fixed<B, B_FRAC>>

where V: Bounded + NumCast + Copy + PromoteAddSub<V> + Signed, A: Bounded + NumCast + Copy + PromoteAddSub<A>, B: Bounded + NumCast + Copy + PromoteAddSub<B>, V::Output: Bounded + NumCast + Copy, A::Output: Bounded + NumCast + Copy, B::Output: Bounded + NumCast + Copy, Fixed<V, V_FRAC>: Copy + Default + PartialOrd + AddAssign + Sub<Output = Fixed<V::Output, V_FRAC>> + Add<Output = Fixed<V::Output, V_FRAC>>, Fixed<V::Output, V_FRAC>: Copy + PartialOrd, Fixed<A, A_FRAC>: Copy + PartialOrd + Sub<Output = Fixed<A::Output, A_FRAC>>, Fixed<B, B_FRAC>: Copy + PartialOrd + Sub<Output = Fixed<B::Output, B_FRAC>>, Value<V, 0>: ConvertValue<V, V_FRAC>, Value<V::Output, 0>: ConvertValue<V::Output, V_FRAC>, Value<A, 0>: ConvertValue<A, A_FRAC>, Value<B, 0>: ConvertValue<B, B_FRAC>,

pub fn new( value: Fixed<V, V_FRAC>, alpha: Fixed<A, A_FRAC>, beta: Fixed<B, B_FRAC>, ) -> Self

Creates a new ExponentialAverage tracker with separate positive (alpha) and negative (beta) filtering coefficients.

• alpha is used when the new sample is greater than or equal to the current average (positive change).
• beta is used when the new sample is less than the current average (negative change).

Panics

Panics if alpha or beta is not within the range [0, 1].

pub fn add_sample(&mut self, sample: Fixed<V, V_FRAC>)

Adds a new sample to the tracker, updating both the moving average and variance.

pub fn value(&self) -> Fixed<V, V_FRAC>

Returns the current moving average.

pub fn variance(&self) -> Fixed<V, V_FRAC>

Returns the current variance.

impl<V, const V_FRAC: usize, A, const A_FRAC: usize> ExponentialAverage<Fixed<V, V_FRAC>, Fixed<A, A_FRAC>, Fixed<A, A_FRAC>>

where V: Bounded + NumCast + Copy + PromoteAddSub<V> + Signed, A: Bounded + NumCast + Copy + PromoteAddSub<A>, V::Output: Bounded + NumCast + Copy, A::Output: Bounded + NumCast + Copy, Fixed<V, V_FRAC>: Copy + Default + PartialOrd + AddAssign + Sub<Output = Fixed<V::Output, V_FRAC>> + Add<Output = Fixed<V::Output, V_FRAC>>, Fixed<V::Output, V_FRAC>: Copy + PartialOrd, Fixed<A, A_FRAC>: Copy + PartialOrd + Sub<Output = Fixed<A::Output, A_FRAC>>, Value<V, 0>: ConvertValue<V, V_FRAC>, Value<V::Output, 0>: ConvertValue<V::Output, V_FRAC>, Value<A, 0>: ConvertValue<A, A_FRAC>,

pub fn new_single_rate(value: Fixed<V, V_FRAC>, alpha: Fixed<A, A_FRAC>) -> Self

Creates a new ExponentialAverage tracker using a single coefficient for both positive and negative changes.