Class CovMatrix

Lightweight wrapper around a flat row-major [n, m, m] TypedArray buffer. Zero-copy construction from consumeData() output — no transpose.

Provides four access patterns:

at(t): zero-copy subarray view of the m×m covariance at time t
series(i, j): copied time series of Cov(i, j) (length n)
get(t, i, j): single scalar element
variance(t, i): diagonal element Var(state_i) at time t

Constructors

constructor

new CovMatrix(data: FloatArray, n: number, m: number): CovMatrix
Parameters
- data: FloatArray
- n: number
- m: number
Returns CovMatrix
- Defined in src/types.ts:219

Properties

`Readonly`data

data: FloatArray

Flat row-major [n, m, m] buffer

`Readonly`m

m: number

State dimension

`Readonly`n

n: number

Number of timesteps

Methods

at

at(t: number): FloatArray
m×m covariance at time t — zero-copy subarray view, length m*m, row-major
Parameters
- t: number
Returns FloatArray
- Defined in src/types.ts:226

get

get(t: number, i: number, j: number): number
Single element: Cov(i, j) at time t
Parameters
- t: number
- i: number
- j: number
Returns number
- Defined in src/types.ts:232

series

series(i: number, j: number): FloatArray
Time series of Cov(i, j) across all t — copied, length n
Parameters
- i: number
- j: number
Returns FloatArray
- Defined in src/types.ts:242

variance

variance(t: number, i: number): number
Var(state_i) at time t — shorthand for get(t, i, i)
Parameters
- t: number
- i: number
Returns number
- Defined in src/types.ts:237