Any chance to have an estimated % of work completed and work still remaining for the PALFA search
No. BM
What about some statistics about what percentage of the data we are able to analyze. Is the network fast enough to analyze all the data real time? Or is half of it thrown out?
Currently Einstein@home is processing the ARECIBO data slower than we get it, so there's plenty of data left to process. We're working on speeding up the search, but currently we are far from real-time.
Given that the binary pulsar searching is letting data stack up, is there a processing thread that close to keeps up with the data that searches for non-binary (or long binary period) millisecond pulsars? I know in the past things like the NCSA centers (remember those?) were used for this crunching. Picking off some new "low hanging fruit" would certainly make the volunteers happy. I know this implies a new processing thread (and therefore a lot of work), but given the interest in a MSP timing array, finding a few more of those would be a nice benefit.
Or, does the pre-processing thread take care of that? As I understand it, the data for the binary search is already de-dispersed for a target DM. All that's left after that is a (big) FFT and harmonic peak search to find lone pulsars.
Marty
"Better is the enemy of the good." - Voltaire (should be memorized by every requirements lead)
Which is the meaning of the Arecibo Power Spectrum which appears at top right in the graphics? It seems to differ from unit to unit.
Tullio
I don't have authoritative word, but I would presume it's a plot of total energy in the band across various frequency bins. So, high bins are bad in that they generally imply some form of interference in the band (e.g., cheap radios or cell phones that emit out of band, or these days laptop computers are a common culprit given the clock frequencies of these babies exceed 1 GHz).
If you look at the files that get downloaded with each WU, you may see a zaplist_xxx.txt file. Those identify specific frequency bins that were excluded because of too much terrestrial interference.
"Better is the enemy of the good." - Voltaire (should be memorized by every requirements lead)
Which is the meaning of the Arecibo Power Spectrum which appears at top right in the graphics? It seems to differ from unit to unit.
This power spectrum is the result of the computation your machine has done and is refreshed with each new orbital template. It shows an |FFT|^2 of the demodulated time series over the range of 0 Hz to 240 Hz in 40 bins. (Remember that one of the steps in the calculations on your machine is the conversion of the time series into pulsar frame time assuming different binary orbits and orbital orientations to take out Doppler modulation - this step is called demodulation. Check also this info page).
If there's a strong pulsar in the data set it will show up as a peak in at least one of the bins at its spin frequency. There are however some human-made interferences (e.g. radar) that will look similar, so not every peak in this diagram is a pulsar.
By closely looking at the power spectrum one can actually tell how fast the pulsar in the data set is spinning and (in combination with the template information in the lower right corner) how the orbit might look like.
the leftmost and second leftmost bin higher and whiter,
Yes, there is sometimes the radar at 1/12 Hz = 0.083333 Hz, which has many higher harmonics as well. That would show up in these bins.
There are, however, low frequency pulsars as well: re-discoveries all of which would be in the first bins.
Ahh, I gladly stand corrected on the nature of the "Power Spectrum" plot. So it's the "pulsar-like signals" power spectrum. Nice! The re-discovery of the 2.1 ms pulsar I presume would have shown up in one of the more rightward bins since that's over 400 Hz...in fact it may have been right off that scale! The fastest known pulsars spin at about 650 Hz. If I recall my pulsar statistics off the top of my head, more than 90% of pulsars spin in the 0.3 to 5 Hz range. These are the "boring" pulsars few researchers spend much time on. The nearest ones of this ilk were discovered first. Picture Jocelyn Bell, or soon after the Arecibo operators, poring over a strip chart analog recording looking for periodic bumps in the trace. Now we've got thousands of computers doing it.
"Better is the enemy of the good." - Voltaire (should be memorized by every requirements lead)
RE: RE: Any chance to
)
What about some statistics about what percentage of the data we are able to analyze. Is the network fast enough to analyze all the data real time? Or is half of it thrown out?
Is the data from arecibo
)
Is the data from arecibo still coming in? The seti@home receiver is down.
Currently Einstein@home is
)
Currently Einstein@home is processing the ARECIBO data slower than we get it, so there's plenty of data left to process. We're working on speeding up the search, but currently we are far from real-time.
BM
BM
Given that the binary pulsar
)
Given that the binary pulsar searching is letting data stack up, is there a processing thread that close to keeps up with the data that searches for non-binary (or long binary period) millisecond pulsars? I know in the past things like the NCSA centers (remember those?) were used for this crunching. Picking off some new "low hanging fruit" would certainly make the volunteers happy. I know this implies a new processing thread (and therefore a lot of work), but given the interest in a MSP timing array, finding a few more of those would be a nice benefit.
Or, does the pre-processing thread take care of that? As I understand it, the data for the binary search is already de-dispersed for a target DM. All that's left after that is a (big) FFT and harmonic peak search to find lone pulsars.
Marty
"Better is the enemy of the good." - Voltaire (should be memorized by every requirements lead)
Which is the meaning of the
)
Which is the meaning of the Arecibo Power Spectrum which appears at top right in the graphics? It seems to differ from unit to unit.
Tullio
RE: Which is the meaning of
)
I don't have authoritative word, but I would presume it's a plot of total energy in the band across various frequency bins. So, high bins are bad in that they generally imply some form of interference in the band (e.g., cheap radios or cell phones that emit out of band, or these days laptop computers are a common culprit given the clock frequencies of these babies exceed 1 GHz).
If you look at the files that get downloaded with each WU, you may see a zaplist_xxx.txt file. Those identify specific frequency bins that were excluded because of too much terrestrial interference.
"Better is the enemy of the good." - Voltaire (should be memorized by every requirements lead)
RE: Which is the meaning of
)
This power spectrum is the result of the computation your machine has done and is refreshed with each new orbital template. It shows an |FFT|^2 of the demodulated time series over the range of 0 Hz to 240 Hz in 40 bins. (Remember that one of the steps in the calculations on your machine is the conversion of the time series into pulsar frame time assuming different binary orbits and orbital orientations to take out Doppler modulation - this step is called demodulation. Check also this info page).
If there's a strong pulsar in the data set it will show up as a peak in at least one of the bins at its spin frequency. There are however some human-made interferences (e.g. radar) that will look similar, so not every peak in this diagram is a pulsar.
By closely looking at the power spectrum one can actually tell how fast the pulsar in the data set is spinning and (in combination with the template information in the lower right corner) how the orbit might look like.
Einstein@Home Project
Thanks to Martin and
)
Thanks to Martin and Benjamin. Most of the times I see the leftmost and second leftmost bin higher and whiter, so I guess they are interference.
RE: the leftmost and second
)
Yes, there is sometimes the radar at 1/12 Hz = 0.083333 Hz, which has many higher harmonics as well. That would show up in these bins.
There are, however, low frequency pulsars as well: re-discoveries all of which would be in the first bins.
Einstein@Home Project
RE: RE: the leftmost and
)
Ahh, I gladly stand corrected on the nature of the "Power Spectrum" plot. So it's the "pulsar-like signals" power spectrum. Nice! The re-discovery of the 2.1 ms pulsar I presume would have shown up in one of the more rightward bins since that's over 400 Hz...in fact it may have been right off that scale! The fastest known pulsars spin at about 650 Hz. If I recall my pulsar statistics off the top of my head, more than 90% of pulsars spin in the 0.3 to 5 Hz range. These are the "boring" pulsars few researchers spend much time on. The nearest ones of this ilk were discovered first. Picture Jocelyn Bell, or soon after the Arecibo operators, poring over a strip chart analog recording looking for periodic bumps in the trace. Now we've got thousands of computers doing it.
"Better is the enemy of the good." - Voltaire (should be memorized by every requirements lead)