conatus.expert.DataInnerProduct

class conatus.expert.DataInnerProduct(kind, n, w=None, F=None, note='')

Bases: object

⟨a, b⟩ = aᴴ W b on flattened snapshots of length n (see module doc).

Parameters:

• kind (str)

• n (int)

• w (Any)

• F (Any)

• note (str)

kind: str

n: int

w: Any = None

F: Any = None

note: str = ''

classmethod quadrature(w, *, note='')

Parameters:

note (str)

Return type:

DataInnerProduct

classmethod gram(W, *, note='')

Parameters:

note (str)

Return type:

DataInnerProduct

classmethod euclidean(n)

Parameters:

n (int)

Return type:

DataInnerProduct

classmethod from_grid(grid, *, components, weights)

Stacked components (e.g. u, v, w) on a conatus.diagnostics.Grid: ⟨a, b⟩ = Σ_c weights[c] ∫ ā_c b_c dV with the grid’s exact quadrature (trapezoid × Fejér). Snapshots are flattened component-major (C order of an array (ncomp, *grid.shape)).

Parameters:

• components (Sequence[str])

• weights (dict)

Return type:

DataInnerProduct

check(X)

Return type:

None

half(X)

F X: coordinates in which the product is Euclidean (X: n × m, NumPy or CuPy).

unhalf(Y)

F⁻¹ Y (inverse of half()).

inner(a, b)

aᴴ W b for vectors or matrices (columns).

explain()

Return type:

str