Einstein@Home
Einstein@Home is a volunteer distributed computing project that needs your help to find Neutron Stars via their electromagnetic and gravitational wave emission.
Wikipedia page
Why Einstein@Home?
During a lunchtime conversation in 1999, Bruce Allen and a friend were discussing an article that they read that day in The Los Angeles Times about SETI@home. The thought occurred that this would be a great way to supply computer cycles to tackle the data analysis problem that they had, but concluded that there would be very little public interest and the topic was dropped.
In 2004, the idea was revisited due to the upcoming event World Year of Physics 2005. The American Physical Society offered publicity and volunteers and after eventually connecting with David Anderson, who spread the excitement of BOINC, Einstein@Home was launched in February of 2005. [1]
Goal
Einstein@Home uses the idle time of computing devices to search for weak astrophysical signals from spinning neutron stars (often called pulsars) using data from the LIGO gravitational-wave detectors, the MeerKAT radio telescope, the Fermi gamma-ray satellite, as well as archival data from the Arecibo radio telescope.
The long-term goal is to make the first direct detections of gravitational-wave emission from spinning neutron stars. Gravitational waves were predicted by Albert Einstein a century ago, and were directly seen for the first time on September 14, 2015. This observation of gravitational waves from a pair of merging black holes opens up a new window on the universe, and ushers in a new era in astronomy.
Methods
Einstein@Home employs the following search methods:
The gravitational wave emitted by a deformed spinning neutron star is very simple. It is almost perfectly monochromatic. This means that it has a single frequency (twice the rotation frequency of the neutron star). This instantaneous frequency decreases slowly over time as the spinning neutron star loses energy through the emission of gravitational (and, if it is a pulsar, electromagnetic) waves. If one were to observe the gravitational-wave emission while floating in space at rest relative to the rotating deformed neutron star, things would be easy. Finding nearly monochromatic gravitational waves in a noisy detector is straightforward: A simple Fourier analysis would quickly reveal the periodicity. But in reality, the actual search is much more complicated and computationally demanding. One of the main reasons: Our detectors are not at rest relative to the neutron star. They sit on the surface of the Earth, which rotates daily and orbits the Sun once a year: The detectors are moving relative to the neutron star. This causes a Doppler shift in the gravitational-wave frequency observed by the detectors. The strength of the Doppler effect depends on time (during a day and within a year) and on the position of the neutron star in the sky. The plot on the right shows a simulation of a continuous gravitational-wave signal received on Earth. You can observe the annual and daily Doppler effect modulations.
To describe a continuous gravitational-wave signal, four different parameters are required: the sky position (two parameters, for example: right ascension and declination), the gravitational-wave frequency (one parameter), and the change of the gravitational-wave frequency over time (one parameter, usually called spin-down). To search for a faint signal in noisy detector data, long stretches of data (covering months of observations) must be analyzed. If the parameters of the signal are unknown, many different possible parameter combinations must be tested: Suppose there's a signal with a certain frequency, spin-down, and position in the sky. This combination of parameters will tell you what the expected signal would look like. Now, check the detector data for the presence of the expected signal using Fourier analysis methods. If nothing is found, try again with a different combination of parameters. Such a search requires a very large number of parameter combinations. This is because, over time, even a tiny offset in one of the parameters would cause the search to potentially miss a signal hidden in the detector noise: Assume a frequency value just a little off from the true one, and the signal will not show up in the analysis. The same holds for offsets in sky position or spin-down. To minimize the chance of missing a hidden signal, the data is very finely combed using a large number of parameter combinations.
Finding the periodic pulsations from gamma-ray pulsars is very difficult – even more so from the very fast millisecond pulsars. On average only 10 photons per day are detected from a typical pulsar by the LAT onboard the Fermi spacecraft. To detect periodicities, years of data must be analyzed, during which the pulsar might rotate tens of billions of times. For each photon one must determine exactly when during a single milliseconds rotation period it was emitted. This requires searching over long data sets with very fine resolution in order not to miss any signals. The computing power required for these “blind searches” – when little to no information about the pulsar is known beforehand – is enormous.
The search is a a "blind search" because we do not know the exact distance, spin frequency, and orbital parameters of the radio pulsar that might be hidden in a data set. A wide range in these parameters must be searched to maximize detection probability.
Interstellar space is filled with clouds of gas and dust. Some of these clouds have temperatures of about 8,000 K and contain free electrons. These clouds will disperse radio waves travelling through them, meaning that higher radio frequencies arrive earlier than lower ones. The more electrons in the gas along the line of sight, the larger this time-delay. Radio telescopes observe a wide band of radio frequencies, so this dispersion has to be corrected for. Since the exact amount of dispersion depends on the unknown distance to the pulsar and the number of electrons along this distance we correct for 628 trial values of dispersion and search each of the resulting data sets independently. This process is called "dedispersion" and done on the Einstein@Home servers.
Since we are ignorant of the orbital parameters of the binary we have to try thousands of possible orbital templates, each corresponding to a different pattern of Doppler spinup and spindown. For each of these templates the data are corrected for the full Doppler effect of the corresponding orbit. This is the first step done on the computers attached to the project. The next step is to test whether there is a radio pulsar present in that data set on that (or a similar) orbit. This is done by using a frequency analysis (Fourier transform) that will recover the spin frequency without smearing.
Because the signals of radio pulsars are not sinusoidal but pulsed, the frequency analysis will show frequency components at the fundamental frequency (the intrinsic spin frequency) and at higher harmonics (integer multiples of the fundamental frequency). Summing these components is a well-known trick in pulsar searches and significantly increases the sensitivity of the search. This summation is the last step done on the users' computers. Finally a list of the most significant candidates is reported back to the Einstein@Home servers and analyzed by the project scientists.
Project team / Sponsors
Bruce Allen. University of Wisconsin Milwaukee, Max Planck Institute for Gravitational Physics (Albert Einstein Institute) Hanover, Germany.
Einstein@Home is a World Year of Physics 2005 and an International Year of Astronomy 2009 project. It is supported by the American Physical Society (APS), the US National Science Foundation (NSF), the Max Planck Society (MPG), and a number of international organizations.
See the list of contributors
Scientific discoveries
einsteinathome.org/science/discoveries