Physicists Turn Pulse Timing into a Quantum Ruler for Weak Values

Weak measurement, the strange quantum protocol in which a system barely touches a meter and is then post-selected, has long been a laboratory curiosity with outsized metrological payoff. Now a new theoretical and experimental study shows that the technique works even when the measuring device is far from ideal. Using an ordinary commercial pulsed diode laser whose light is only partially coherent, researchers have measured the post-selected time shift of a polarization qubit with few-picosecond precision, and they have shown that a single visibility parameter is enough to describe how the imperfect pointer bends the result. The work, published as a combined theory-and-experiment paper, demonstrates that partial coherence is not a nuisance to be engineered away but a controllable feature that can be folded directly into the model of the measurement.

The protocol traces back to Aharonov, Albert, and Vaidman, who introduced weak measurement as a scheme in which a preselected quantum system is weakly coupled to a pointer and then post-selected, revealing an observable quantity known as the weak value. That weak value is generally complex and can fall outside the eigenvalue spectrum of the measured observable, a property that underlies the famous amplification of tiny physical shifts. In optical implementations, the real and imaginary parts of the weak value show up in conjugate pointer quadratures: position and momentum when the pointer is a spatial beam, time and frequency when the pointer is the arrival time of a pulse. Most table-top demonstrations have relied on spatial pointers, inferring the measured quantity from a beam displacement. Temporal and spectral pointers apply the same weak-coupling logic but replace spatial imaging with time-resolved detection, an approach first used to measure group and signal velocities in optical fiber and later combined with spectral interference to resolve delays far below the pulse width.

Temporal pointers are attractive for polarization qubits because they can be built from standard pulsed lasers and fast electronic detection, with no specialized spatial or spectral imaging. The catch is that temporal precision is bounded by pulse coherence, and inexpensive pulsed sources can be only partially coherent on the delay scales relevant to weak measurement with typical mechanisms such as birefringent walk-off. In the new experiment, that partial coherence arises intrinsically from a commercial multimode pulsed source rather than from a deliberately degraded light field. Crucially, the authors emphasize, the imperfect pointer coherence is not a failure mode of the experiment: it shifts the measured signal in a controlled way, and it is therefore included explicitly in the theory rather than treated as noise.

The theoretical core of the paper is a closed-form expression for the post-selected time shift that interpolates between two familiar limits. When the pointer visibility at the measurement delay equals one, the expression recovers the ideal coherent weak-measurement response, reducing to the real part of the weak value in the weak-coupling regime. When the visibility equals zero, it gives the projective result, the Born-rule probability for the measured projector. Intermediate visibilities glide smoothly between these extremes. The derivation defines a single overlap parameter, the real part of the normalized field autocorrelation of the pointer at the imposed delay, which lies between zero and one and depends only on the pointer itself, not on the input polarization state. Because the two interfering components in the experiment originate from the same field split on a polarizing beam splitter, they share the same temporal amplitude and spectrum, so the same visibility applies to every input state at a given delay.

To test this framework, the researchers chose three input-state trajectories corresponding to rotations about the principal axes of the Poincaré sphere, the three-dimensional space in which every pure polarization state is a point. The Stokes parameters serve as coordinates: S1 separates horizontal from vertical polarization, S2 diagonal from antidiagonal, and S3 right-circular from left-circular. The horizontal-diagonal-vertical-antidiagonal trajectory, a rotation about S3, is sensitive to pointer visibility. The horizontal-right-vertical-left trajectory, a rotation about S2, and the diagonal-right-antidiagonal-left trajectory, a rotation about S1, both yield predictions that are independent of visibility wherever the weak value is defined. Those two visibility-independent trajectories therefore act as parameter-free checks of the apparatus and the extraction method, while the third trajectory constrains the visibility itself. This internal consistency across complementary trajectories is the central experimental test of the partial-coherence framework.

The apparatus is deliberately modest. A commercial multimode pulsed diode laser at 640 nanometers produces 10-nanosecond pulses at a 10-megahertz repetition rate, with a near-flat-top temporal profile and a sharp leading edge. A half-wave plate, and for two trajectories a quarter-wave plate, prepares the input polarization state, which is scanned by a motorized rotation stage. A polarizing beam splitter separates the horizontal and vertical components, the horizontal arm is delayed by a mirror translation of about 2.5 centimeters, giving a round-trip path difference near 5 centimeters and a delay of order 170 picoseconds, and the paths recombine before post-selection on the diagonal state with a half-wave plate and linear polarizer. A 2-gigahertz silicon photodiode feeds an oscilloscope triggered on a reference channel picked off upstream of the entire measurement, so trigger jitter is common to the reference and the signal.

The extraction method is a study in careful bookkeeping. Because the pulses are nearly rectangular, the most stable timing marker is the inflection point of the leading edge, located as the maximum of a Savitzky-Golay-smoothed first derivative. Points within 60 percent of that maximum on both sides are retained for a fourth-order polynomial fit, evaluated on a thousand-fold refined grid to define the time for each trace. Equivalent-time sampling builds the waveform up to an effective spacing of one point every 4 picoseconds, with each trace averaged over more than ten thousand waveforms. The horizontal and vertical pole states provide an internal calibration that defines the session-specific delay, typically around 177 picoseconds, which absorbs trigger drifts of several hundred picoseconds between acquisitions. The result is a post-selected normalized pointer shift with a precision of order 0.02, and the finite-coupling correction is a negligible 2 times 10 to the minus 3.

The results confirm the model at every turn. Along the horizontal-right-vertical-left trajectory, the measured shift follows the parameter-free prediction of a cosine-squared response with an RMS residual of about 0.04. Along the diagonal-right-antidiagonal-left trajectory, the response is constant at 0.50 plus or minus 0.01, exactly as predicted when the visibility-dependent terms cancel between numerator and denominator. The visibility-sensitive trajectory is captured by a single-parameter fit that yields a visibility of 0.09 plus or minus 0.05, placing the pointer close to the projective limit, consistent with a multimode source whose coherence time is short compared with the imposed delay. Notably, near the antidiagonal state, where the ideal coherent weak value would diverge, the measured response remains finite and bounded, exactly the behavior the partial-coherence model predicts. The authors read the fitted visibility as an upper bound rather than a resolved nonzero value.

The same recorded traces also give qualitative access to the imaginary part of the weak value through the spectral content of the signal. Computing the power spectral density of each recorded pulse and tracking its centroid frequency reveals a shift that follows the expected sine dependence along the circular trajectory, with maxima at the right- and left-circular states, reaching 0.274 megahertz at its largest. Because the photocurrent spectrum is centered at zero intermediate frequency and folded about it, the method recovers only the magnitude of the imaginary weak value, not its sign, a limitation the authors describe as a self-homodyne detection in the incoherent regime. A self-heterodyne scheme with a frequency shifter could restore the sign. The visibility parameter also compresses the full multimode structure of the pointer into a single number, sufficient at a fixed delay but requiring refinement for scans across the complete multimode beating structure.

The broader implications are striking. A temporal pointer requires only a delay and time- or frequency-resolved detection, both native to integrated photonics, making the scheme in principle more amenable to on-chip implementation than spatial-pointer measurements that rely on free-space propagation. A more coherent pulsed source would push the measurement toward the ideal weak limit, where the measured shift recovers the real part of the weak value directly, and phase-preserving spectral detection would complete the imaginary quadrature, together turning trajectory scans into a direct route to reconstructing the input polarization state. For now, the demonstration stands on its own: weak measurement, one of quantum mechanics’ most counterintuitive tools, has been shown to survive, and even to make sense of, one of the messiest lasers money can buy.

Subject of Research: Partial-coherence weak measurement of polarization qubits using a temporal pointer

Article Title: Partial-coherence weak measurement of polarization qubits with a temporal pointer: Theory and experiment

Article References: Gervais, S. P., & Giner, L. (2026). Partial-coherence weak measurement of polarization qubits with a temporal pointer: Theory and experiment. Results in Optics, Article 101191. https://doi.org/10.1016/j.rio.2026.101191

Image Credits: AI Generated

DOI: 10.1016/j.rio.2026.101191

Keywords: weak measurement, weak value, polarization qubits, temporal pointer, partial coherence, Poincaré sphere, post-selection, pulsed diode laser, quantum metrology, time-shift extraction, multimode laser, integrated photonics