ProbeKind#

class t3toolbox.backend.fitting.ProbeKind#

Bases: _WwSample, ProbeOutputKind

The vector-valued probe (one free mode per measurement), optionally per-mode weighted.

Mode weighting is the order-0 case of the same residual-weight machinery as the derivative kinds: the objective is 1/2 sum_i |omega_i z_i|^2, so omega enters sumsq (x omega) and transpose (x omega^2) only. Plain probe has no order axis, so the weight is a 1-D (d,) per-mode vector.

name = 'probe'#
weight: t3toolbox.backend.common.typ.Optional[t3toolbox.backend.common.typ.Any] = None#

Methods#

__post_init__()

precompute(frame, ww)

The reusable frame sweep at frame for sample (computed once per local model).

forward(v, ww, frame, sweep)

The bare Jacobian J v (no gauge).

transpose(r, ww, frame, sweep)

The bare transpose J^T r, summed over the sample stack W; raw (dU, dG).

point_forward(x_cores, ww)

The point operation S(x) -- what the residual is measured against.

take(sample, data, idx)

Gather a flat W subset of (sample, data) -- the default minibatch draw's layout hook.