MeerKAT Pulsar Timing Array (MPTA)

• MPTA is a high-cadence pulsar timing program using the MeerKAT interferometer to monitor Southern millisecond pulsars with sub-microsecond precision.
• It employs advanced techniques like coherent dedispersion, subband folding, and rigorous RFI mitigation to achieve high-accuracy pulsar timing.
• MPTA enhances gravitational wave detection and terrestrial time standard calibration by integrating its data with international PTA efforts.

The MeerKAT Pulsar Timing Array (MPTA) is a southern-hemisphere, high-cadence ensemble timing program utilizing the MeerKAT radio interferometer in South Africa to monitor the pulse arrival times of millisecond pulsars (MSPs). Its principal objectives are the detection and characterization of nanohertz-frequency gravitational waves (GWs), calibration of terrestrial time standards using an astrophysical ensemble, and the contribution of precision timing data to the International Pulsar Timing Array (IPTA) network. MPTA provides sub-microsecond timing on 88 southern-sky MSPs leveraging MeerKAT’s sensitivity, large bandwidth, and precise time infrastructure, with data products including sub-banded arrival times, calibrated Stokes profiles, frequency-resolved noise models, and clock-correction waveforms.

1. Scientific Motivation and Array Configuration

MPTA is motivated by two key deficiencies in global pulsar timing: incomplete southern-sky coverage in previous PTAs and the necessity for independent GW background and terrestrial time calibration. MeerKAT, a 64-dish interferometer (13.5 m diameter each), achieves high gain ($G\approx2.0$ K/Jy per polarization), low system temperature ($T_{\rm sys}\approx18$ K), and rapid slew speeds. Its telescope core, $\sim1$ km diameter, minimizes baseline leakage, optimizing pulsar timing sensitivity. Utilizing the L-band receiver (856–1712 MHz, $\sim770$ MHz usable bandwidth), MPTA observes each MSP at a nominal fortnightly cadence, with integrations tailored to individual flux densities (10–30 min/epoch).

Table: Core observational metrics for standard MPTA operations

Metric	Typical Value	Notes
Number of MSPs (2021)	88	$>30$ epochs for 78 pulsars
Cadence	$\sim$14 d	Variable per program
RMS residual, 2.5 yr	median $0.8\;\mu$s	$<1\;\mu$s for 67 MSPs
Bandwidth	$\sim$770 MHz	RFI excision applied

MPTA prioritizes MSPs below declination $+20^\circ$, drawn from ATNF and legacy timing catalogs, with selection favoring low DM ($<100$ cm$^{-3}$ pc), high flux ($S_{1400}>1$ mJy), and orbital parameters suitable for high-precision timing. The observational sample includes isolated and binary MSPs, spanning DM $\sim2$–200 pc cm$^{-3}$ and periods 1.5–8 ms.

2. Data Acquisition, Preprocessing, and Product Structure

The MeerKAT timing backend (PTUSE) processes dual-polarization signals through downconversion, digitization (8-bit, 1 GHz), and channelization (4096 PFB channels). Real-time coherent dedispersion is performed across the full band, followed by subbanding (typically 256 subbands of $\sim$3 MHz each) and time folding into full-Stokes archives (1024 phase bins, 8 s sub-integrations). Radio frequency interference (RFI) mitigation employs both automated median-zapping (e.g., CoastGuard) and manual cleaning for persistent sources.

Archived data products include:

Product Type	Format/Access	Content
TOA lists	.txt/.csv, .tim	Subbanded, band-averaged arrival times
Calibrated archives	PSRFITS	Full polarization, time-frequency profiles
Template profiles (“portraits”)	.std, wideband	Frequency-dependent timing standards
Ephemerides	.par files	Timing solution for each pulsar
Noise models	.yaml/.json	EFAC, EQUAD, red/chromatic noise parameters
Clock correction coefficients	.npy/.txt	Reconstructed timescale offsets

Data are released under CC-BY 4.0 and integrated into both MPTA and IPTA public repositories. Scripts for ingest into TEMPO2 and ENTERPRISE are provided.

3. Timing Analysis, Ephemerides, and Noise Model Construction

TOAs are determined by Fourier-domain template matching across subbands using high signal-to-noise standard templates (summed over $>50$ epochs per MSP). For each epoch $i$, the residual is computed as $r_i = t_{\rm obs,\,i} - t_{\rm model,\,i}$, where $t_{\rm model}$ is the predicted phase from the timing solution (including spin, astrometry, binary Keplerian and post-Keplerian parameters, and DM).

Noise is modeled as:

• White noise: EFAC (scaling factor on TOA errors), EQUAD (additive term, quadrature) per MSP.
• Red noise: power-law process $P(f) = A^2 (f/f_{\rm ref})^{-\gamma}$ for spin irregularities and DM variations.
• Template-fitting: Frequency-dependent “portraits” correct for intrinsic pulse-shape evolution and minimize profile-alignment systematics.

After subtracting modeled white and red noise, 67 MSPs reach frequency-averaged residuals $<1\;\mu$s, and 30 MSPs reach $\lesssim200$ ns over the 2.5-year span. Ephemerides are refined using TEMPO2, seeded from IPTA DR2 solutions.

4. Station Clock Monitoring and Correction via Pulsar Timing

A distinctive achievement of MPTA is the recovery of the station clock waveform directly from ensemble pulsar residuals, providing an independent time standard. The clock correction $\delta t_{\rm clk}(t)$ is parameterized as a sum over Fourier basis functions:

$$\delta t_{\rm clk}(t) = \sum_{k=1}^N [ a_k \sin(2\pi k t / T_{\rm span}) + b_k \cos(2\pi k t / T_{\rm span}) ]$$

The fitted clock correction agrees with independent GNSS clocks at the 20–50 ns level on month-to-year timescales. This enables direct tracking and correction of station time drifts, reducing the risk of time-transfer systematics in GW searches.

5. Early Gravitational-Wave Constraints and Preliminary Outcomes

The first MPTA data release yields competitive sensitivity to a stochastic GW background. Cross-correlation of timing residuals from 67 high-precision MSPs produces a 95% confidence upper limit on characteristic strain:

$$A_{\rm GWB} < 1.2 \times 10^{-15}\;\;\;\mathrm{at}\;f=1\,\mathrm{yr^{-1}}$$

With corresponding upper bound on energy density:

$$\Omega_{\rm gw}(f) < 3 \times 10^{-10}\;\left(f/\mathrm{yr}^{-1}\right)^{5/3}$$

These limits approach those already published by other major PTA consortia, supporting MPTA’s role in the global search for a nanohertz GWB. The high-precision southern coverage reduces degeneracies in stochastic-background and time-standard inference, bridging gaps in the international PTA sky.

A plausible implication is that with planned array expansion (toward 120 MSPs, increased cadence) and S-band receiver installation (for DM mitigation), MPTA is positioned for further sensitivity increases and a possible first definitive GW detection as the timing baseline grows.

6. Outlook, Integration, and Data Accessibility

MPTA’s open-data philosophy ensures that high-fidelity timing, polarization, noise-model, and clock-correction archives are available to the international community. The data set underpins analyses ranging from ensemble GW background studies (via Hellings–Downs angular correlations) to single-source GW searches and population-level neutron star studies. MPTA’s independent time-scale correction and robust noise modeling facilitate joint analyses with PPTA, NANOGrav, EPTA, and future large-scale facilities. The procedure explicitly addresses systematics (RFI, time reference, DM/red noise) and is supported by scripts and documentation for pipeline reproducibility.

As the southern sky’s most sensitive PTA, MPTA decisively strengthens international GW search efforts. Its precision (sub-microsecond for the majority of MSPs), array size, and systematic treatment establish a technical standard for PTA operation and data dissemination.