METRe: Multidisciplinary Metre-Scale Studies

• METRe is a polysemous research term denoting metre-scale phenomena in quantum superposition tests, meteoroid impact analyses, and EHR data harmonization.
• In quantum physics, METRe distinguishes between interference and first-order coherence in metre-scale Bose–Einstein condensate experiments.
• In planetary science and clinical informatics, METRe statistically constrains meteoroid impact conditions and standardizes cross-database EHR extraction respectively.

METRe is a polysemous research term rather than a single unified concept. In the literature provided here, it denotes at least three distinct usages: a quantum-physics dispute over whether atom-interferometric fringes verify metre-scale spatial superpositions (Stamper-Kurn et al., 2016, Kovachy et al., 2016), a planetary-science shorthand for the study of metre-sized meteoroids or metre-scale Earth impactors and the inference of their impact conditions (Zuluaga et al., 2019, Devillepoix et al., 2018), and, in a different capitalization, the biomedical informatics acronym METRE, the “Multidatabase ExTRaction PipEline” for harmonized extraction from MIMIC-IV and eICU (Liao et al., 2023). The common element is the recurrence of “metre-scale” phenomena or datasets, but the underlying technical content, methods, and controversies are domain-specific.

1. Nomenclature and scope

In the quantum-measurement literature, METRe is used for the Metre-scale Verification of superpositions problem: whether a Bose–Einstein-condensate atom interferometer with arm separation of about 54 cm verifies a macroscopic quantum superposition or only demonstrates interference and second-order coherence (Stamper-Kurn et al., 2016). In the planetary-defense literature, “METRe” is not expanded explicitly as a formal acronym in the cited impact-prediction paper; it is described as best read as a shorthand for metre-sized impactors / metre-sized meteoroids and the question of whether their impact conditions can be constrained in advance (Zuluaga et al., 2019). A related observational paper uses METRe as shorthand for the metre-scale Earth impactor population, bodies roughly 1–100 m across with emphasis on the metre-scale end (Devillepoix et al., 2018). In clinical machine learning, by contrast, METRE is explicitly defined as the “Multidatabase ExTRaction PipEline”, a preprocessing framework for critical-care EHRs (Liao et al., 2023).

These usages should not be conflated. The same letter sequence labels a conceptual controversy in quantum mechanics, an impactor population in planetary science, and a data-engineering system in critical care research. A plausible implication is that METRe functions more as a context-dependent label than as a stable cross-disciplinary acronym.

2. METRe in metre-scale quantum superposition verification

The quantum-mechanical METRe problem arises from the experiment of Kovachy et al., which used optical pulses to place a freely falling Bose–Einstein condensate into two spatial trajectories that separated by about 54 cm, then recombined them to produce high-contrast interference fringes with random phase from shot to shot (Stamper-Kurn et al., 2016). The central claim of “Verifying quantum superpositions at metre scales” is that the observed interference is consistent with, but does not prove, that the separated atomic ensembles were in a quantum superposition state (Stamper-Kurn et al., 2016).

The paper’s conceptual core is the distinction between interference and phase-coherent superposition. A coherent two-branch state is written as

$|\psi\rangle = \frac{1}{\sqrt{2}\left(|L\rangle + e^{i\phi}|R\rangle\right),$

with a well-defined relative phase $\phi$, and therefore off-diagonal structure in the one-body density matrix. By contrast, a phase-randomized state is represented as

$$\rho = \frac{1}{2}|L\rangle\langle L| + \frac{1}{2}|R\rangle\langle R|,$$

or more generally as an ensemble over phases, and need not exhibit fixed one-body phase coherence (Stamper-Kurn et al., 2016).

On that basis, the paper distinguishes first-order coherence, associated with

$$\langle \hat{\Psi}^\dagger(\mathbf{r}) \hat{\Psi}(\mathbf{r}') \rangle,$$

from second-order coherence, associated with

$$\langle \hat{\Psi}^\dagger(\mathbf{r}) \hat{\Psi}^\dagger(\mathbf{r}') \hat{\Psi}(\mathbf{r}') \hat{\Psi}(\mathbf{r}) \rangle.$$

Its argument is that a single-shot fringe pattern with random phase demonstrates strong interference visibility and second-order coherence, but does not verify first-order coherence across the full metre-scale path separation (Stamper-Kurn et al., 2016).

This distinction is used to frame the experiment’s relation to collapse and decoherence models. The paper argues that the fringes do not straightforwardly exclude mechanisms such as continuous spontaneous localization (CSL) or gravitationally induced decoherence, because a one-body mixed state can still yield random-phase interference when the clouds overlap (Stamper-Kurn et al., 2016). The conclusion is therefore restrictive rather than dismissive: the interferometer is a major technical achievement, but metre-scale superpositions remain experimentally unverified unless the relative phase is stabilized or referenced by a configuration such as two well-separated interferometers sharing the same pulses (Stamper-Kurn et al., 2016).

3. Response literature and the superposition–coherence controversy

The immediate response, “Response to ‘Verifying quantum superpositions at metre scales’,” rejects the claim that the Kovachy et al. result fails to probe macroscopic superposition (Kovachy et al., 2016). It advances three counterclaims: standard quantum mechanics does not forbid future differential-measurement use of the system; the experiment does test quantum mechanics by constraining modifications that would reduce contrast for widely separated interferometer arms; and, under a standard definition, the observed interference does arise from quantum superposition at the half-metre scale (Kovachy et al., 2016).

The rebuttal to interaction-based or technical objections is quantitative. The response states that the atomic source had a condensate fraction of about 50% and that the source coherence length was only $2 \times 10^6\ \mathrm{m}$, much smaller than the cloud size. It further emphasizes that coherence between the two arms is created by the initial beam splitter pulse, at which point the interaction ratio is

$U/J \sim 10^{-8},$

and the density is low enough that the mean-field shift is only

$\sim 0.1\ \mathrm{Hz}.$

Under these conditions, the authors state that there is no known mechanism in standard quantum mechanics that prevents future use in differential measurements (Kovachy et al., 2016).

The response also narrows the scope of what the experiment constrains. Its claim is not that every conceivable modification of quantum mechanics is excluded, but that the result bounds modifications that would induce decoherence or contrast loss when the interferometer arms are widely separated. If a spurious momentum kick $hq$ occurs midway through the interferometer, the induced phase shift is estimated as

$$\frac{m(v+hq/m)^2}{2} - \frac{mv^2}{2} \sim \hbar qL,$$

with

$\phi$0

The paper notes that even kicks of order

$\phi$1

can generate phase shifts of about $\phi$2 radians and reduce contrast if they are inhomogeneous (Kovachy et al., 2016). At the same time, it explicitly states that modifications which add only overall phase noise are not ruled out.

The dispute therefore turns on the relation between superposition and first-order coherence. The response explicitly states that quantum superposition is a more general concept than first-order coherence, and that, following Feynman and others, interference necessarily results from superposition whether or not there is a determinate phase (Kovachy et al., 2016). The cited analogy is the Pfleegor–Mandel experiment. The controversy is not merely semantic: it concerns what experimental signature should count as verification of a macroscopic spatial superposition.

4. METRe in statistical prediction of metre-sized meteoroid impacts

In planetary science, one METRe usage concerns whether the impact conditions of metre-sized meteoroids can be constrained in advance even though such objects are typically undetectable before atmospheric entry (Zuluaga et al., 2019). The cited method is Gravitational Ray Tracing (GRT), a backward-integration statistical technique that starts from a known impact time and location, generates candidate trajectories over possible impact speeds and incoming directions, integrates them backward through the Solar System gravitational field, and weights them using the observed Near Earth Object (NEO) distribution in orbital-element space (Zuluaga et al., 2019).

The ray probability is written as

$\phi$3

Here $\phi$4 is a flux correction relative to Earth’s motion and $\phi$5 is the local NEO number density in orbital-element space. The flux correction is parameterized as

$\phi$6

with $\phi$7 and $\phi$8, while the density estimate is written

$\phi$9

The corresponding distance metric in $$\rho = \frac{1}{2}|L\rangle\langle L| + \frac{1}{2}|R\rangle\langle R|,$$0 space is based on the Zappalà/Rożek formulation (Zuluaga et al., 2019).

From these ray probabilities, the paper derives marginal and bivariate probability distributions for impact speed, radiant azimuth and elevation, and asymptotic orbital elements. The marginal and radiant sky probabilities are written as

$$\rho = \frac{1}{2}|L\rangle\langle L| + \frac{1}{2}|R\rangle\langle R|,$$1

and

$$\rho = \frac{1}{2}|L\rangle\langle L| + \frac{1}{2}|R\rangle\langle R|,$$2

The method is tested on Chelyabinsk and Viñales. For Chelyabinsk, GRT predicted a likely radiant in the north-east quadrant and a low-inclination, moderately eccentric orbit; the paper reports that the real event was consistent with those constraints, though the observed azimuth was in a lower-probability part of the distribution (Zuluaga et al., 2019). For Viñales, the best-fit impact conditions include a radiant with $$\rho = \frac{1}{2}|L\rangle\langle L| + \frac{1}{2}|R\rangle\langle R|,$$3, $$\rho = \frac{1}{2}|L\rangle\langle L| + \frac{1}{2}|R\rangle\langle R|,$$4, a speed of $$\rho = \frac{1}{2}|L\rangle\langle L| + \frac{1}{2}|R\rangle\langle R|,$$5, and an estimated fireball energy of about 1.4 kt TNT (Zuluaga et al., 2019).

The paper’s conclusion is deliberately qualified. Metre-sized meteoroid impacts are not predictable in the conventional detection sense, but their impact conditions can still be statistically constrained once time and location are known (Zuluaga et al., 2019). This distinction is central to the METRe usage in this domain.

5. METRe as the metre-scale Earth impactor population

A closely related but observationally distinct planetary-science usage treats METRe as the metre-scale Earth impactor population characterized by the Desert Fireball Network (DFN) (Devillepoix et al., 2018). The paper states that Earth is impacted by about 35–40 metre-scale objects per year, yet these objects are too small for efficient NEO telescopic discovery and too rare for conventional ground-based fireball surveys to characterize densely (Devillepoix et al., 2018).

The DFN is described as a continent-scale autonomous camera network in Australia with about 52 observatories and about 3 million km² coverage. Each observatory uses a high-resolution digital camera, a fisheye all-sky lens, long-exposure imaging, and a GNSS-synchronized liquid crystal shutter (Devillepoix et al., 2018). The paper focuses on two DFN events, DN150102_01 / Kalabity and DN170630_01 / Baird Bay, and compares them with US Government (USG) sensor detections and other independently studied events.

The central methodological result is comparative rather than purely descriptive. Ground-based networks are presented as strong for trajectory geometry, velocity evolution, dynamic pressure / fragmentation, pre-atmospheric orbit, and meteorite fall position, whereas orbital sensors provide global collecting area and useful impact energy statistics (Devillepoix et al., 2018). The major limitation of the USG data is the velocity vector. The paper states that reported speeds can be wrong by as much as ~28%, that the radiant direction can be off by several degrees and sometimes up to 90°, and that only 2 out of 9 USG cases would yield reasonably accurate fall positions for recovery work (Devillepoix et al., 2018).

This has implications for source-region inference. The paper argues that a previously discussed Halley-type comet (HTC) source region for some metre-scale impactors is not well supported because two of the three key USG events used for that inference have unreliable velocity solutions (Devillepoix et al., 2018). For Kalabity, the DFN-derived orbit is instead main-belt-like, with $$\rho = \frac{1}{2}|L\rangle\langle L| + \frac{1}{2}|R\rangle\langle R|,$$6, $$\rho = \frac{1}{2}|L\rangle\langle L| + \frac{1}{2}|R\rangle\langle R|,$$7, $$\rho = \frac{1}{2}|L\rangle\langle L| + \frac{1}{2}|R\rangle\langle R|,$$8, and $$\rho = \frac{1}{2}|L\rangle\langle L| + \frac{1}{2}|R\rangle\langle R|,$$9 (Devillepoix et al., 2018). The METRe problem in this usage is therefore observational and statistical: how to measure rare metre-scale impactors accurately enough to determine both hazard and provenance.

6. METRE in critical care informatics

In biomedical informatics, METRE refers to a data-extraction framework rather than a metre-scale physical phenomenon. “A Multidatabase ExTRaction PipEline (METRE) for Facile Cross Validation in Critical Care Research” presents an open-source pipeline that transforms raw EHR data from MIMIC-IV and the eICU Collaborative Research Database into aligned, model-ready structured data frames (Liao et al., 2023).

The architecture begins with cohort selection and then extracts three tables: Static, Vital, and Intervention. The cohort can be constrained by age range, ICU length of stay, missingness thresholds, and condition-specific keywords such as sepsis_3, ARF, shock, COPD, and CHF (Liao et al., 2023). The Static table contains age, gender, ethnicity, comorbidities, and, for MIMIC-IV, an approximate admission year window. The Vital table contains blood gas, lab, urine chemistry, and vital-sign measurements, aggregated by hour with missingness indicators, outlier removal, and imputation. The Intervention table represents procedures and therapies as binary hourly time series, including ventilation, vasopressors, fluid boluses, colloids, CRRT, transfusion types, and antibiotic administration (Liao et al., 2023).

By default on MIMIC-IV, METRE extracts 92 labs and vitals, 16 intervention variables, and 35 time-invariant variables. To align the two databases structurally, it assigns NaNs to 15 variables not found in eICU. Under default cohort settings, it extracts 38,766 ICU stays from MIMIC-IV and 126,448 ICU stays from eICU, for a combined default cohort of 165,254 ICU stays (Liao et al., 2023).

Validation is performed on five clinically relevant prediction tasks derived from Tang et al. (hospital mortality, ARF at 4 h or 12 h, and shock at 4 h or 12 h) using logistic regression, random forest, 1D CNN, and LSTM. The study uses an 80:20 train-test split, 10-fold cross-validation on the training set, and reports AUC and AUPRC with 1,000-bootstrap 95% confidence intervals (Liao et al., 2023). The reported AUROC range across tasks is 0.723–0.888, and the best MIMIC-IV in-hospital mortality result is 0.868 with LSTM. In cross-database evaluation, the abstract reports that the AUC change can be as small as +0.019 or −0.015, although some tasks degrade much more severely; for ARF, when models trained on eICU are tested on MIMIC-IV, AUROC drops are reported as large as 0.322 for random forest and 0.277 for LSTM (Liao et al., 2023).

METRE’s contribution is therefore infrastructural rather than algorithmic. It does not claim to solve generalization across all EHR sources; instead, it harmonizes two specific databases so that a model trained on one can be evaluated directly on the other without transfer learning (Liao et al., 2023). This makes “METRE” a term of workflow standardization, in marked contrast to the quantum and planetary-science uses of “METRe.”