Indirect Weak Measurement

• Indirect weak measurement is a quantum protocol where system observables are inferred via a weakly coupled probe, minimizing state disturbance.
• It employs pre- and post-selection along with unitary evolution to extract weak values through pointer shifts and full counting statistics.
• Experimental implementations in optics and superconducting circuits highlight its applications in state tomography, detector reconstruction, and precision metrology.

Indirect weak measurement is a quantum measurement scheme in which the observable of interest is not read out directly but is inferred through a probe, meter, or ancilla that interacts weakly with the system. In the weak regime, only a tiny amount of information is extracted in each run, so the disturbance is minimal and the measured state is left essentially undisturbed; in the experimental language of orbital-angular-momentum measurements, it “will not cause the measured state collapse” (Zhu et al., 2018). In the standard pre- and post-selected formulation, the measurable signature is governed by the weak value

$$A_w=\frac{\langle f|\hat A|i\rangle}{\langle f|i\rangle},$$

which can be complex and can be accessed through shifts or full counting statistics of a pointer observable rather than through a projective measurement of $\hat A$ itself (Svensson, 2012, Lorenzo, 2011). The formalism underlies optical, superconducting-circuit, interferometric, and tomography protocols, and it also supplies a common language for amplification, partial measurement, and non-demolition-style readout (Qin et al., 2016, Zhu et al., 2018).

1. Formal structure of the indirect weak-measurement scheme

In the ancilla formulation, the system $S$ and meter $M$ are assigned Hilbert spaces $\mathcal H_S$ and $\mathcal H_M$, with initial density matrix

$\rho_0=\rho_S\otimes \rho_M.$

A pre-measurement unitary entangles the two subsystems, after which the meter is read out and, in post-selected protocols, the system is projected onto a chosen final state (Svensson, 2012). The standard von Neumann coupling is written as

$U=\exp(-ig\, S\otimes P),$

or equivalently through an interaction Hamiltonian such as

$H_{\rm int}(t)=\epsilon\, g(t)\, A P_d$

or

$\hat H=\gamma\, \hat A\otimes \hat P,$

depending on the physical realization (Aharonov et al., 2012, Zhu et al., 2018).

Pre-selection and post-selection are central. The system is prepared in an initial state $\hat A$0 or $\hat A$1, interacts weakly with the meter, and is then post-selected in $\hat A$2 or $\hat A$3. To first order in the weak interaction, the post-selected meter state is governed by the weak value. In the density-matrix review, the meter state after post-selection is approximated by

$\hat A$4

with

$\hat A$5

and the meter expectation value obeys relations such as

$\hat A$6

for a conjugate pointer variable $\hat A$7 (Svensson, 2012).

The Two-State-Vector Formalism supplies a standard interpretation of this pre-/post-selected structure. In that framework, weak measurement probes the system between two strong measurements, and the pointer shift reflects the weak value rather than an eigenvalue. The pointer is usually prepared as a broad Gaussian, so any individual event is imprecise, while repeated trials accumulate a statistically resolvable signal (Aharonov et al., 2012).

2. Weak values, output statistics, and generalized formulations

The weak value is the canonical first-order quantity, but indirect weak measurement is not restricted to average pointer shifts. A general full-counting-statistics treatment introduces the characteristic function for an arbitrary Hermitian probe observable $\hat A$8,

$\hat A$9

with the exact representation

$S$0

where

$S$1

This formulation extends weak-measurement analysis from means and variances to all moments of the conditional probe distribution (Lorenzo, 2011).

The same framework introduces the normal weak values

$S$2

which remain finite even when pre- and post-selected states are nearly orthogonal. Within a good approximation, the whole output statistics can be described by a complex parameter, the weak value, and a real one; the latter appears in the literature as the “second” weak value $S$3 (Lorenzo, 2011).

A distinct generalization replaces the usual perturbative derivation by structural conditions on pointer states. For an entangled state

$S$4

the AAV effect follows if the pointer states are approximately indistinguishable,

$S$5

and if the matrix elements of suitable pointer observables satisfy

$S$6

Under those conditions,

$S$7

This shifts the emphasis from small coupling alone to pointer-state overlap as the operative criterion for weakness, and it predicts “weak echo,” in which weak-value behavior can re-emerge at high interaction strengths when pointer states again become nearly indistinguishable (Salvail, 2013).

3. Experimental realizations

A direct optical realization of indirect weak measurement is the measurement of the topological charge $S$8 of orbital-angular-momentum beams. In that experiment, a He-Ne laser and a spatial light modulator prepare the OAM beam, polarization pre-selection is implemented with wave plates, the weak interaction is realized in a polarizing Sagnac interferometer with a PZT-shifted mirror, post-selection uses further wave plates and a PBS, and a CCD records the output intensity profile (Zhu et al., 2018). The transverse centroid coordinates are

$S$9

and the topological charge is inferred from

$M$0

The reported experimental range is from $M$1 to $M$2, with a scaling and threshold correction introduced to compensate for CCD sensitivity limitations (Zhu et al., 2018).

Superconducting-circuit implementations use ancilla-mediated partial measurement. In one cQED realization, a transmon system qubit $M$3 is indirectly measured by an ancilla qubit $M$4, with a bus resonator and dedicated readout resonators. The interaction strength $M$5 tunes the measurement from no interaction at $M$6 to a projective limit at $M$7, while the ancilla readout basis $M$8, $M$9, or $\mathcal H_S$0 sets the effective measurement operator (Groen et al., 2013). For the informative $\mathcal H_S$1 basis, the system Bloch angle updates as

$\mathcal H_S$2

and the ideal information-disturbance tradeoff is summarized by

$\mathcal H_S$3

The same platform was used to demonstrate the correspondence between non-classical weak values and Leggett-Garg inequality violation (Groen et al., 2013).

A different superconducting implementation employs a dispersively coupled transmon qubit and microwave cavity. The measurement strength is controlled by the integration time $\mathcal H_S$4 relative to the characteristic measurement time $\mathcal H_S$5, and the weak measurement is modeled by Kraus operators

$\mathcal H_S$6

Homodyne detection of the cavity quadrature yields outcome bins $\mathcal H_S$7, so that weak and strong measurements of incompatible observables can be compared within the same cQED hardware (Monroe et al., 2020).

4. State, detector, phase, and entanglement inference

Indirect weak measurement has been used as a state-characterization method in circuit QED. In the dispersive regime,

$\mathcal H_S$8

and the homodyne current is modeled as

$\mathcal H_S$9

For finite-strength measurements, the post-selected average is not given by the infinitesimal AAV expression alone but by

$\mathcal H_M$0

By varying the local-oscillator phase, the real and imaginary parts of $\mathcal H_M$1 can be extracted and used to reconstruct the coefficients of the unknown qubit state. The reported analysis is carried out for finite strength weak measurement and beyond the bad-cavity and weak-response limits, and it is described as robust against measurement inefficiency (Qin et al., 2016).

Weak-value techniques have also been extended from state tomography to detector tomography. In the backward direction of the time-symmetric formalism, the key experimentally accessible quantity is the numerator of the weak value,

$\mathcal H_M$2

from which POVM elements can be reconstructed directly. For arbitrary measurement strength $\mathcal H_M$3,

$\mathcal H_M$4

The completeness relation $\mathcal H_M$5 is then used to improve precision through weighted averaging of directly and indirectly inferred matrix elements (Xu et al., 2020).

Several protocols target quantities that are not observables in the projective sense. For arbitrary two-qubit pure states, concurrence is determined from local weak values

$\mathcal H_M$6

through

$\mathcal H_M$7

An optical implementation uses Laguerre-Gaussian pointer modes, with

$\mathcal H_M$8

so that both parts of the weak value are obtained in one fixed setup (Tukiainen et al., 2016).

Weak values of projection operators also encode geometric phase. For

$\mathcal H_M$9

the phase of the weak value equals the phase of the third-order Bargmann invariant,

$\rho_0=\rho_S\otimes \rho_M.$0

Alternating strong and weak measurements can therefore determine Berry’s phase directly; in SIC-POVM settings the modulus is fixed and the geometric information resides entirely in the argument (Gedik, 2021). Related single-particle protocols use cyclic weak measurements to measure transmission coefficients of a multiport beam splitter and to map the wavefunction while the photon remains superposed (Aharonov et al., 2012).

5. Back-action, precision, and uncertainty bounds

The practical range of indirect weak measurement is limited by finite coupling strength. In polarization weak-value experiments with heralded single photons, birefringent crystals of $\rho_0=\rho_S\otimes \rho_M.$1 mm and $\rho_0=\rho_S\otimes \rho_M.$2 mm yielded coupling strengths $\rho_0=\rho_S\otimes \rho_M.$3 and $\rho_0=\rho_S\otimes \rho_M.$4, respectively. For $\rho_0=\rho_S\otimes \rho_M.$5, the linear weak-value approximation holds well for anomalous values in the interval $\rho_0=\rho_S\otimes \rho_M.$6; for $\rho_0=\rho_S\otimes \rho_M.$7, the linear regime narrows to $\rho_0=\rho_S\otimes \rho_M.$8. Beyond those intervals, higher-order terms become significant, the meter response saturates, and arbitrarily large weak values can no longer be extracted reliably (Piacentini et al., 2017). This fixes a coupling-dependent threshold on signal amplification.

Metrological analyses reach a related conclusion. For weak-measurement-based parameter estimation, the total Fisher information is written as

$\rho_0=\rho_S\otimes \rho_M.$9

In the two generic protocols analyzed, post-selection does not enhance measurement precision once all information is counted, and weak measurement does not offer better precision than strong measurement on the same footing. The one affirmative result concerns phase-space interactions, for which the post-selection statistics alone can attain Heisenberg scaling,

$U=\exp(-ig\, S\otimes P),$0

whereas configuration-space interactions remain at standard-quantum-limit scaling (Zhang et al., 2013).

The uncertainty structure of indirect measurement has also been reformulated in thermodynamic terms. With noise operator

$U=\exp(-ig\, S\otimes P),$1

and survival activity

$U=\exp(-ig\, S\otimes P),$2

the variance decomposition

$U=\exp(-ig\, S\otimes P),$3

leads to

$U=\exp(-ig\, S\otimes P),$4

The result expresses a reciprocal tradeoff between added measurement noise and survival activity, and it constrains the noise term that enters universal noise-disturbance relations (Mihashi et al., 2022).

A complementary entropic formulation was tested experimentally in a superconducting transmon. For a strong measurement $U=\exp(-ig\, S\otimes P),$5 and a weak-then-strong sequence $U=\exp(-ig\, S\otimes P),$6 followed by $U=\exp(-ig\, S\otimes P),$7,

$U=\exp(-ig\, S\otimes P),$8

The weak value lowers the uncertainty bound, and the experiment interprets this as a quantitative measure of a weak measurement’s ability to reconcile incompatible operations through backaction (Monroe et al., 2020).

6. Interpretation, rigorous limits, and broader extensions

The operational content of indirect weak measurement is not in dispute: the weak value controls conditional pointer shifts and related statistics. Its ontological status is disputed. One review states explicitly that if anything is “magic” in weak measurement plus post-selection, it is the interpretation of the weak value, and that whether it is a bona fide property of the system is left unanswered (Svensson, 2012). A more critical analysis argues that weak measurements attempt to answer “Which way?” without destroying interference, and that bizarre weak values indicate the failure of a measurement where the uncertainty principle says it must fail, rather than any additional insight into physical reality (Sokolovski, 2013).

Rigorous mathematical treatments of repeated indirect measurement replace interpretive language by trajectory limits. For a finite-dimensional system with non-degenerate observable

$U=\exp(-ig\, S\otimes P),$9

and jump operators

$H_{\rm int}(t)=\epsilon\, g(t)\, A P_d$0

the unperturbed non-demolition case drives the state toward an eigenstate of $H_{\rm int}(t)=\epsilon\, g(t)\, A P_d$1. When a small Hamiltonian perturbation does not commute with $H_{\rm int}(t)=\epsilon\, g(t)\, A P_d$2, the system exhibits jumps between different eigenstates, and after rescaling time by $H_{\rm int}(t)=\epsilon\, g(t)\, A P_d$3 the maximum-likelihood estimate converges in law to a Markov jump process on $H_{\rm int}(t)=\epsilon\, g(t)\, A P_d$4 with explicitly derived transition rates (Ballesteros et al., 2017). This gives a controlled limit in which indirect weak monitoring generates emergent classical jump dynamics.

Several extensions broaden the notion of indirect weak measurement itself. The instantaneous indirect measurement method infers the expectation value of one operator from the instantaneous change of another operator, relative to a known reference state, and is described as neither significantly affecting the wave function nor causing wave function collapse (Lu et al., 2022). In holographic CFT, weak measurement is modeled as a soft projection

$H_{\rm int}(t)=\epsilon\, g(t)\, A P_d$5

and is dual to an interface brane separating geometries with different effective central charges satisfying

$H_{\rm int}(t)=\epsilon\, g(t)\, A P_d$6

This realizes weak measurement as a continuous interpolation between no measurement and projective measurement, with corresponding effects on entanglement entropy, boundary entropy, and phase structure (Sun et al., 2023).

A further unification places direct and indirect weak measurements in a single framework with total Hamiltonian

$H_{\rm int}(t)=\epsilon\, g(t)\, A P_d$7

so that direct measurement appears as a special case of indirect measurement. In that setting, probe shifts obey

$H_{\rm int}(t)=\epsilon\, g(t)\, A P_d$8

while direct measurement is recovered from transition-probability changes in the system alone. The same paper connects indirect weak measurement to Rabi and Ramsey resonance and compares their precision within a common weak-value-amplification language (Ueda et al., 28 Sep 2025). This suggests that the indirect weak-measurement formalism is not only a measurement protocol but also a unifying framework for partial readout, resonance metrology, tomography, and dynamical state inference.