The Parkes High Time Resolution Universe Legacy Survey
HTRU's "Diamond Planet" Gains World-wide International Recognition!
See the HD video describing the discovery here: YouTube Link
The Parkes High Time Resolution Universe Legacy Survey will perform a comprehensive survey of the Galaxy for millisecond and regular pulsars, rotating radio transients, and giant pulses. The survey will also have exceptional sensitivity to short-duration extragalactic radio bursts. It is a collaboration between the Swinburne University of Technology Pulsar Group, the University of Manchester Pulsar Group, the University of Cagliari Pulsar Group and the Australia Telescope National Facility.
Swinburne is responsible for the development of the 13-beam digital filterbank backend and associated data capture system, which is based upon the iBOB board developed by CASPER (Dan Werthimer's Berkeley group) and powered by Xilinx's FPGAs with the spectrometer FPGA development work performed by Peter Mc Mahon (University of Cape Town). Swinburne PhD students will reprogram the boards to enable polarimetry of the bursts. The boards will also be used in Swinburne Engineering undergraduate courses as professional implementations of FPGA design. The UDP data capture solution software PSRDADA will enable widespread application of the board and code in radio astronomy.
The Parkes Multibeam Surveys represent a watershed in pulsar astrophysics. The 13-beam multibeam receiver with its exceptional system temperature combined with Parkes' unrivaled view of the Southern skies allowed the discovery of over 800 radio pulsars, including dozens of millisecond pulsars, the double pulsar, the highest-magnetic field pulsars known, the eccentric relativistic binary white dwarf-pulsar PSR J1141-6545, the binary pulsar with the most massive companion, the most-eccentric binary pulsar, the rotating radio transients, and more than doubled the total population. These pulsars have enabled the most rigorous tests of General Relativity, enabled the measurement of precise pulsar masses via Shapiro delay, and are important for efforts to detect gravitational waves from supermassive black hole binaries. A parallel survey of the Magellanic clouds was later reprocessed to discover the first extragalactic radio burst. Over a dozen PhD students participated in these surveys.
The crucial technology used in these historic surveys for the telescope "backend" were 1-bit analogue filterbanks with 3 MHz resolution. However, the FGPA digital signal processing revolution that is transforming radio astronomy enables the production of inexpensive digital solutions where previously cumbersome and limited analogue solutions were only possible. After discussions with Dan Werthimer, it became clear that CASPER's iBOB board which uses Xilinx FPGAs could be re-programmed to generate two 1024-channel filterbank 8-bit streams across a 400 MHz RF band with 32 μs time resolution. This would deliver unprecedented time and frequency resolution across 400 MHz for pulsar surveys and greatly enhance the volume of the Galaxy that can be searched for millisecond and binary pulsars, especially in the direction of the Galactic Centre, where the density of pulsars is highest. The 8-bit data from the iBOB will enable more rigourous radio frequency interference excision than has been hitherto possible.
While the previous Parkes Multibeam Surveys were state-of-the-art at their conception, the FPGA revolution allows us today to significantly improve on them in time and frequency resolution as well as dynamic range.
The survey will address a wide range of scientific questions, enabled by the discovery of a large variety of pulsars and transient sources:
- What is the Galactic population of neutron stars?
- Do pulsars rotate with sub-millisecond periods?
- How many rotating radio transients exist, compared to pulsars?
- How common are extragalactic bursting radio sources?
- Which new rare and exotic objects exists?
- Are there more Double Pulsars and where are the pulsar-black hole systems?
The subsequent study of the discoveries allows us to exploit these sources for fundamental problems in physics and astrophysics, ranging from the properties of super-dense matter to cosmology and galaxy evolution:
- Is Einstein's general relativity correct under extreme conditions?
- What is the amplitude of gravitational waves created by merging
super-massive black holes in the early Universe?
- How dense can we pack matter?
- Is the rotation speed of neutron stars limited by some braking mechansim?
- Why differ different types of neutron stars and how are they formed?
- What are the counterparts to high-energy sources?
- What phenomena create powerful radio bursts?
and many more.
Inspecting a radio map of the Universe (on right), one can clearly identify the band of the Milky Way. Searching in this area of the sky promises to reveal the largest number of pulsars and also the most compact binary systems. However, here we also encounter the largest amount of interstellar medium between the Earth and the pulsar. Above and below the Galactic plane we expect a large number of millisecond pulsars which are very old and therefore had a long time to move away from their birth place. Even further away from the plane, the number of pulsars will become less but we explore a part of the sky that has never been searched for fast transient radio sources, possibly even emerging from extragalactic sources. For maximum efficiency, we will take these differences into account, and will adopt an optimized survey strategy.
In the plane, we point the telescope for 70 min at the various positions, ensuring that with this observing time and our high frequency resolution, we will penetratre the whole Galaxy. At intermediate latitudes, we can afford to observe much shorter, only for 8 min. To cover the huge area of the remaining sky, we will integrate for 4 min, being still very sensitive to bursting radio transients. In all cases, we will record the signals of all 13 receivers very quickly, more than 15,000 times a second - data processing is a huge challenge as described below!
Expected Outcome & Applications
The deep pointings onto the Galactic Plane will discover distant pulsars that were previously hidden from our view. These will include young pulsars, those which are still associated with the remnant of their violent formation in a supernova (see the Crab supernova remnant at left), and also a group of old pulsars in tight binary orbits. In fact, this survey offers the best chance yet to discover a pulsar orbiting a black hole! Such system will be an exciting cosmic laboratory to test Einstein. In total we expect several hundred new pulsars in this area plus many more rotating radio transients sources.
So far rotating radio transients have been found only at low Galactic latitudes, but a search away from the plane may reveal new sources which will allow us to study their true nature. However, the majority of pulsars to be found in this area will be fast rotating millisecond pulsars. They are often found with a white dwarf companion, which may allow us to measure the neutron star masses via Shapiro delay measurements. We will identify those which perform as the best clocks and use them in an experiment to detect low-frequency gravitational waves. The large number of millisecond pulsars that we expect to find will also give us a good understanding how fast neutron stars can really spin.
The vast areas above and below the plane will reveal very old and also very fast pulsars. The problem of how pulsars obtain their high speeds of up to 1000 km/s or more are still poorly understood. But what makes the search in this area of the sky even more exciting is the prospect to find strong bursts from extragalactic explosions or other phenomena, as described above. We know very little today how the radio sky changes on very short time scales and how many flashes of radio light we have missed even with the largest telescopes. This survey will go a long way to explore this astrophysical frontier.
Telescope and Receiver Packages
The 64 metre Parkes radio telescope has been an indispensable instrument for observational pulsar astrophysics. With its large collecting area of 3200 m2 and location in the Southern hemisphere, the telescope's visible sky spans a large fraction of the Galactic plane where density of pulsars is highest.
In 1996 the telescope was equipped with a 21cm multibeam receiver at its prime focus, enhancing its pulsar discovery and research potential by manifold. Designed by Trevor Bird at CSIRO Radiophysics and constructed by the ATNF receiver group, this receiver comprises a hexagonal feed cluster of 13 beams and cryogenically cooled receivers. The receiver is sensitive to radio astronomy signals over a large bandwidth of 300 MHz centred at 1350 MHz and boasts an exceptional system temperature of 20 K. With a half power beam width of 14 arcmin and a spacing on the sky of 28 arcmin, the feed assembly provides a large instantaneous sky coverage. As well as a remarkable survey instrument, the multibeam receiver can be well exploited for efficient transient searches through its discriminatory potential to find and eliminate spurious signals.
The Digital Filterbanks.
CASPER's iBOB board is at the heart of the backend. To speed development we chose to do the absolute minimum on the FPGA board, and refine the data and algorithms using C/C++ codes on the servers which host the Gb NICs that are directly connected to the iBOBs. The iBOB produces 2 1024 channel 8-bit spectra every 32 us and these are placed inside relatively simple UDP frames for transport to the host computers. Each iBOB accepts two IFs, the station clock and a 1 PPS. They digitize up to 400 MHz of IF, and for the purposes of our survey will produce 300 MHz of radio astronomy signal between 1200 and 1500 MHz. The 768 radio astronomy channels are selected and placed in a ring buffer. From there various diagnostics can be performed before real-time searches for giant pulses are performed and the data decimated for offline analysis.
Dr Willem van Straten and Andrew Jameson have developed the data capture software for the board PSRDADA that copies the 8-bit data into a ring buffer for subsequent analysis via 10 Gb ethernet. For each of the 13 beams an iBOB is required, and these are each hosted in their own dedicated server with 16 GB of RAM on a dedicated private network with high speed access to the Swinburne supercomputer 800 km away. Whilst in RAM the data can be searched for giant pulses in real time, before being filtered for radio frequency interference and decimated for subsequent analysis. Each of the 13 beams will produce ~4MB s-1 of data which will be transferred to the Swinburne supercomputer for near-real time analysis via the dedicated Gb link. The iBOB has the capability to be reprogrammed to deliver all 4 Stokes parameters that would enable polarimetry of any extragalactic radio bursts discovered to make the first measurements of the intergalactic magnetic field. A PhD project is being offered to pursue this.
The Supercomputing Challenge
Data rates and Storage
These data will form an invaluable petabyte digital record of the sky for future generations of astronomers. When fully operational, the survey will collect more than four terabytes of data per day of observing. Over the full lifetime of the survey we expect that the total data archive will be in excess of 1000 terabytes. This level of data production creates challenges in storing, monitoring and archiving of the data.
Although much of the processing will be carried out in `real time', experience shows that there is great value in preserving data archives for future analysis. In particular this allows us to take advantage of future developments in hardware and software processing techiques that will allow us to do more in-depth analysis than is possible today.
The green machine is a 1600+ core supercomputing cluster that hosts 3 tape robots and when completed in December 2007 was the largest computing cluster in Australia with over 16 Tflops of computational power. This machine will perform near real time analysis of the data. Archives of the data will be sent to collaborators and in total up to 40 Tflops of computational muscle will sift through over 1000 Terabytes of data for the duration of the surveys.
Millisecond pulsars are generated in relativistic binaries and are often accelerated significantly in a survey pointing. Even with the largest supercomputer in Australia we cannot adequately probe the phase space millisecond pulsars populate. Thus, we will revisit the data many times over coming years as supercomputers grow in power to search for
the most extreme binaries.
The Legacy: A Resource for the present and the future
Although radio telescopes continue to get larger and more sensitive, they cannot turn back time to see how a particular pulsar was spinning or how its pulse profile looked in the past. Future scientists will find pulsars below our discovery threshold with new telescopes, and find our digital record invaluable to see the state of the pulsar in our data.
Teaching the next generation of student researchers
This survey is pioneering in its scope and technologies. Student participation is required across many of its components, from the digitiser and FPGAs, to the data capture systems, the supercomputing cluster that will search for transients in real time, the algorithms behind optimising the search codes, to the supercomputing grid that will search the astronomical haystack for the precious needles. Once sources are found, the astrophysical applications will be even more rewarding. Will this survey uncover the first pulsar-black hole binary - or some hitherto unknown class of object?
Moore's law and reprocessing
Every ten years, computer power increases by about two orders of magnitude. Thus although we can only just process the data today, in coming decades scientists will be able to reprocess the data and search much larger phase spaces for even more exotic objects.
The Virtual Observatory
One unique aspect of this survey will be its large sky coverage, meaning that we will be able to instantly perform searches for transients and pulsars in any area of the Parkes sky. This will provide a unique service to the Virtual Observatory, allowing astronomers to get access to time-series data on their target objects without having to use any physical telescope. This also gives a useful historical census for future projects, allowing them to compare their future data with those collected today.
Data will be archived by the Australia Telescope National Facility and made available to anyone after the normal proprietary period of 18 months. People wanting early access to the data should not hesitate to contact us.
The SKA context
The completed survey promises not only to be the most defining survey in history, but it wil also provide a benchmark for all pulsar work to be done in the future, including the astonishing science that can be achieved with the Square-Kilometre-Array (SKA). The SKA is a global endevour of, currently, 19 countries to build the largest telescope the Earth has ever seen. This unique radio-telescope will have a total collecting area of 1 square kilometre or 300 times the size of the Parkes telescope. The results of our survey will give us an excellent understanding of what to expect with an SKA survey, and will represent an important and exciting stepping stone towards the SKA. When the SKA becomes operational, anticipated for about 2020, it will follow up on many of our new discoveries.
Student training and opportunities!
PhD student places are available at all of the host institutions for participation in this survey. Students with the following experience are keenly sought:
Institutional Contact Personnel:
Swinburne: Email Professor Matthew Bailes (Professor, Swinburne Centre for Astrophysics and Supercomputing)
Manchester: Email Professor Michael Kramer (Professor of Astronomy)
Cagliari: Email Dr Andrea Possenti (Research Fellow)
ATNF: Email Dr Simon Johnston (Senior Research Scientist)
- December 2006: Survey conceived at the Texas Symposium by Simon Johnston, Matthew Bailes, Michael Kramer and Andrea Possenti.
- 15 March 2008: Survey awarded test time on the Parkes 64m radio telescope in semester 2 2008.
- 12 May 2008: iBOB successfully installed at the Parkes 64m telescope.
- 26 May 2008: iBOB sees "first light" from the binary millisecond pulsar PSR J0437-4715.
- 6 June 2008: Xilinx generously donates FPGAs for the iBOB boards.
- 15 Nov 2008: First observations in "survey" mode.
- 1 Dec 2008: First new pulsar discovered!
- 18 Feb 2009: Over 40 TB of data now obtained.
- 12 June 2009: Possible Magnetar discovery.
- 26 June 2009: 120 TB of data obtained, first millisecond pulsar discovered.
- 13 Sep 2009: Magnetar confirmed!
- 1 Oct 2009: First eclipsing millisecond pulsar discovered.
- 12 Dec 2009: 2 hour orbit millisecond pulsar discovered!
- 12 Feb 2010: First moderately recycled millisecond pulsar discovered. (27 milliseconds).
- 26 Feb 2010: Single pulse hunting commences.
- 10 March 2010: 7th millisecond pulsar discovered, 38 pulsars in total, medium latitude survey only 50% processed!
- 12 March 2010: Keck time applied for to image binary pulsar companion.
- June 2010: First high-latitude MSP discovered.
- Sep 2010: MSP with DM = 260 discovered!
- Oct 2010: Keck time used to image binary pulsar field.
- Jan 2011: Last Medium-latitude survey pointing completed. Over 300 TB of data recorded.
- Jan 2011: Papers describing the survey, magnetar, Single pulse discoveries and five millisecond pulsars now accepted.
- Feb 2011: 20th recycled pulsar discovered, including 17 with P<10 ms. Processing of medlat survey still progressing.
- Feb 2011: Apparently super-young single-pulse source hovering above the Galaxy?!
- Mar 2011: Timing of all 20 recycled pulsars reveals new long-period "black widow" system and first massive companion to our MSPs.
- June 2011: 2-hour binary pulsar paper submitted.
- June 2011: Single pulse "RRAT" discoveries now number 16 from < half the data.
- September 2011: Diamond Pulsar Planet paper appears in Science and becomes an internet sensation!