TMTS: Tsinghua University–Ma Huateng Telescope Survey

• TMTS is a high-cadence, wide-field survey system featuring four 40 cm telescopes configured for minute-cadence monitoring of transient and variable phenomena.
• It employs advanced instrumentation including scientific-grade CMOS/CCD detectors and a custom white-light filter to achieve photometric precision as fine as 5 mmag for bright sources.
• TMTS integrates machine learning classification with cross-matches to Gaia and LAMOST data to build high-fidelity catalogs of over 11,000 periodic variables.

The Tsinghua University–Ma Huateng Telescopes for Survey (TMTS) is a high-cadence, wide-field optical survey system located at Xinglong Station (NAOC), designed for time-domain monitoring of the LAMOST spectroscopic footprint. TMTS comprises four identical 40 cm telescopes equipped with scientific-grade detectors and operates primarily in a white-light filter mode. Its key scientific focus is the unbiased, minute-cadence detection and classification of short-period variable stars, transients, and related phenomena across a large area of the northern sky, in synergy with extensive spectroscopic and photometric surveys.

1. System Architecture and Instrumental Design

TMTS consists of four co-mounted optical telescopes (aperture 40 cm each), installed on a single equatorial fork mount. The optical design (prime focus, f/3.75 to f/4.5 depending on configuration) is engineered for wide-field imaging, providing each tube with an instantaneous field of view (FoV) of ~4.5 deg² and a cumulative system FoV of ~18 deg² when operated in sky-mosaic mode. The detector suite comprises scientific CMOS (sCMOS) or front-illuminated/back-illuminated CCDs, each with a pixel scale of approximately 1.3"–2" per pixel, enabling well-sampled point spread functions under the typical Xinglong seeing conditions (1.5"–3"). The system operates almost exclusively in a custom "Luminous" or "L" white-light filter spanning 400–900 nm, maximizing photon throughput for faint time-domain sources. The 3σ, 60-s limiting magnitude in this configuration is m_lim ≈ 19.4 mag (Zhang et al., 2020, Guo et al., 28 Apr 2025).

Table: Primary System Parameters

Parameter	Value
Aperture	40 cm × 4 telescopes
Focal Ratio	f/3.75 (prime focus)
Detector	sCMOS/CCD, ~2" pixel⁻¹
Field of View/tube	4.5 deg²
Total FoV	18 deg²
Filter	White-light (400–900 nm)
Exposure Time	60 s
Cadence	1 min (no dead time)
Limiting Mag (3σ)	19.4 mag (white-light)
Phot. Precision	5 mmag @ m<14

The system’s flexible pointing strategy allows contiguous sky coverage or simultaneous dual-band imaging when equipped with g/r filters (in selected modes) (Zhang et al., 2020).

2. Survey Strategy, Cadence, and Sky Coverage

TMTS targets sky regions corresponding to the LAMOST spectroscopic plates, resulting in optimal overlap for time-domain photometry and instantaneous spectroscopic follow-up. Since 2020, the survey has monitored ≈ 7,000 deg² via continuous observations—each field receives up to 6–12 hr of uninterrupted 1-min cadence imaging per night (Guo et al., 28 Apr 2025). The typical observing strategy cycles through all available fields as per the LAMOST schedule, accumulating dense temporal coverage for millions of sources. Over 1.9 × 10⁷ light curves (≥100 epochs) have been acquired in the first three years (Liu et al., 2024).

A standard field receives up to several hundred consecutive 1-min exposures in a single night, ensuring robust sensitivity to periodicities down to ~10 min and faint variability across the magnitude range m ≃ 9–19.4. The bright-saturation limit, set by nonlinearity and charge bleeding, varies from ~9–10 mag (Guo et al., 28 Apr 2025).

3. Data Reduction Pipeline and Time-Series Analysis

The TMTS photometric data pipeline comprises the following core stages (Zhang et al., 2020, Guo et al., 28 Apr 2025):

• Raw frame preprocessing: bias subtraction, dark correction, flat-field normalization, and cosmetic masking.
• Astrometric calibration: coordinate matching to Gaia DR2/DR3, yielding astrometric residuals of ~0.1"–0.2".
• Source extraction and photometry: aperture photometry (optimized to 1.5×FWHM) and PSF fitting for crowded fields.
• Photometric calibration: zero-points referenced to Pan-STARRS1 grizy bands using cross-matched stars; typical night-to-night zero-point scatter ~0.01–0.02 mag.
• Light curve assembly: epoch-matched time series per object.
• Photometric precision: σ_m ≈ 0.005 mag for m < 14, rising to σ_m ≈ 0.10 mag for m ≲ 18.

Periodicity detection employs the Lomb–Scargle periodogram (LSP), with power thresholds established via empirical noise modeling and robust false-alarm probability analysis (FAP < 10^–3) (Lin et al., 2021, Guo et al., 2024). Candidates are further refined by Fourier analysis, and multi-epoch light curves are cross-matched with external catalogs for validation (Lin et al., 2023, Guo et al., 2024).

4. Machine Learning Classification and Catalog Construction

TMTS leverages ensemble machine learning (Random Forest, XGBoost) for classification of periodic variables. Feature engineering utilizes period, amplitudes, Fourier decomposition, LSP power, and various light-curve shape statistics. A training set built from VSX/manual labels (DSCT, HADS, EA/EB/EW binaries, RS CVn) supports high-fidelity (F₁ > 98%) automated classification (Guo et al., 2024).

The periodic-variable catalog (2020–2022) comprises:

• 11,638 periodic variables: 4,876 DSCT, 628 HADS, 5,698 EW binaries, 117 EA, 84 EB, and 226 RS CVn candidates.
• Notable high-purity yields: ~88% of DSCT and ~12% of HADS are new discoveries.
• Cross-matching with Gaia DR2/DR3 and LAMOST DR7 provides parallaxes, spectral parameters, and metallicities for population studies (Guo et al., 28 Apr 2025).

Comprehensive catalogs include source ID (e.g., TMTS JHHMMSS.ss±DDMMSS.s), celestial coordinates, period, amplitude, classification probabilities, and multi-band photometry (PS1 grizy, 2MASS JHK_s, WISE W1–W3) (Guo et al., 28 Apr 2025).

5. Scientific Results: Variable Star Populations and Astrophysical Probes

TMTS’s minute-cadence, wide-field strategy has produced transformative samples in several domains:

• Delta Scuti stars (DSCT/HADS): TMTS cataloged 4,876 low-amplitude and 628 high-amplitude DSCT (P < 0.3 days), facilitating new empirical period–luminosity (P–L) relations across 11 bands and markedly improved distance/color excess constraints via Bayesian simultaneous fitting (Guo et al., 28 Apr 2025).
• Eclipsing and contact binaries: Thousands of Algol (EA), β Lyrae (EB), W UMa (EW), and RS CVn systems.
• Cataclysmic variables (CVs): 64 CVs identified—including DNe, NLs, IPs, polars—with minute-cadence photometry enabling detection of superhumps, QPOs, spin modulations, and eclipse morphology (Liu et al., 2024).
• White dwarfs, subdwarfs, and exotic objects: Discovery of blue large-amplitude pulsators (BLAPs), ZZ Ceti, and sdBV stars sampling the short-period regime (Lin et al., 2023).

The survey enables:

• Construction of multi-band P–L relations, using Gaia parallax and 3D dust priors:
  • $M_{i,j} = M_{0,j} + \alpha_j \log_{10}(P_i/P_0)$
  • Simultaneous fitting allows refined estimates of distance modulus $\mu_i$ and extinction $E(B-V)_i$, yielding reduced uncertainties in posterior distributions (Guo et al., 28 Apr 2025).
• Investigation of metallicity effects (β_j [Fe/H]_i) on the P–L–Z relations, with preliminary indications that metallicity correlates with reduced scatter at longer wavelengths, though significance is below 3σ (Guo et al., 28 Apr 2025).

In the CV population, detailed light-curve morphology reveals clear distinctions in $H\alpha$ emission strength and width across the period gap—short-period systems exhibit stronger and broader lines, consistent with emission from more compact disks (Liu et al., 2024).

6. Performance Metrics and Diagnostic Benchmarks

TMTS achieves consistent photometric and astrometric specifications:

• Photometric uncertainties: σ_m ≈ 0.005 mag (@ m ≲ 14), σ_m ≲ 0.10 mag (@ m ≲ 18).
• Limiting magnitude: m_lim ≈ 19.4 mag in Luminous filter (3σ, 60 s).
• Astrometric accuracy: ~0.1" compared to Gaia DR3.
• Period recovery: δP/P < 10⁻⁵ for P ≲ 6 hr, with LSPmax > 10σ ensuring negligible false-alarm probability (Guo et al., 28 Apr 2025).
• Survey completeness: ≳95% detection efficiency for P < 2 hr and L ≲ 16 mag (Lin et al., 2023).

Data pipeline latency (for transient/triggers) is a few minutes, supporting near-real-time alerts (Zhang et al., 2020).

7. Scientific and Programmatic Implications

TMTS is a uniquely capable platform for systematically probing rapid variability:

• It fills the discovery parameter space for sub-hour, low-amplitude pulsators and compact binaries missed by prior day-cadence or low-temporal-resolution surveys.
• The sample of >5,000 DSCT/HADS and >5,800 binaries advances constraints on binary evolution, tidal instability models, pulsation physics, and extinction/distance mapping (Guo et al., 28 Apr 2025, Guo et al., 2024).
• Integration with Gaia astrometry and LAMOST spectroscopy enables three-dimensional mapping of stellar populations and Galactic structure, delivering cross-validated multi-parameter catalogs.
• The survey provides a basis for LISA/TianQin verification sources and electromagnetic follow-up of gravitational wave progenitors (Lin et al., 2021).

Synergy with LAMOST ensures opportunities for rapid spectroscopic follow-up, population studies across the full HR diagram, and direct feedback between photometric variability and spectroscopically derived stellar parameters (Zhang et al., 2020).

Ongoing extensions target expanded sky coverage, longer-term baseline monitoring to capture secular period evolution, and further improvements in machine classification for rarer or more complex variable classes (Liu et al., 2024).

---

Key references:

• "Investigating the Period-Luminosity Relations of delta Scuti Stars: A Pathway to Distance and 3-D Dust Map Inference" (Guo et al., 28 Apr 2025)
• "Minute-Cadence Observations of the LAMOST Fields with the TMTS IV: Catalog of Cataclysmic Variables..." (Liu et al., 2024)
• "Minute-Cadence Observations of the LAMOST Fields with the TMTS V: Machine Learning Classification..." (Guo et al., 2024)
• "Minute-cadence Observations of the LAMOST Fields with the TMTS: I. Methodology..." (Lin et al., 2021)
• "The Tsinghua University-Ma Huateng Telescopes for Survey: Overview and Performance of the System" (Zhang et al., 2020)