Pulsars I. Pulsars I. The Why and How of Searching for Exotic Pulsars The Why and How of Searching for Exotic Pulsars Jim Cordes, Cornell University Jim Cordes, Cornell University • Why would you want to know about pulsars and • The forefront of neutron star science why would you like to discover more? • Extreme states of matter • Science, the big questions • Gravitational laboratories • How do I find new pulsars? • Probing core collapse supernovae • Galactic structure • How do I estimate pulsar distances? • Issues in pulsar survey optimization • What can I learn about … • Dedispersion • the inside of a NS? • Periodicity searches • the magnetosphere of a NS? • Single pulse searches • the orbit of a binary pulsar? • Simulations of pulsar surveys • the space velocity of a pulsar? • ALFA: A massive pulsar survey at Arecibo • the ISM along the path to a pulsar? • SKA: toward a full Galactic census of pulsars
Pulsars… Neutron Star Astrophysics • Embody physics of the EXTREME – surface speed ~0.1c – 10x nuclear density in center 13 – some have B > Bq = 4.4 x 10 G – Voltage drops ~ 10 12 volts 9 9 11 –FEM = 10 Fg = 10 x 10 FgEarth 6 –Tsurf ~ 10 K • Relativistic plasma physics in action ( γ ~ 10 6) • Probes of turbulent and magnetized ISM Surface quantities: • Precision tools, e.g. B ≈≈≈ 10 12 Gauss - Period of B1937+21 (the fastest millisecond pulsar) ≈≈≈ 11 gNS 10 g⊕⊕⊕ P = 0.0015578064924327 ±0.0000000000000004 s F ≈≈≈ 10 9 g m Orbital eccentricity of J1012+5307: e<0.0000008 EM NS p • Laboratories for extreme states of matter and as clocks for ΦΦΦ≈≈≈ 10 12 volts probing space-time and Galactic material
Pulsar Sounds Radio signals demodulated into audio signals
P f=1/P sound file Pulsar (ms) (Hz) B0329+54 714 1.4 B0950+08 253 3.9 B0833-45 89 11.2 (Vela) B0531+21 33 30.2 (Crab) J0437-4715 5.7 174 B1937+21 1.56 642 J0437 -4715 profile
1 Pulsar Populations: diagram Pulsar Populations: diagram • Canonical • Canonical • P~ 20ms – 5s • P~ 20ms – 5s • B ~ 10 12±1 G • B ~ 10 12±1 G ) )
• Millisecond pulsars (MSPs) -1 • Millisecond pulsars (MSPs) -1 • P ~ 1.5 – 20ms • P ~ 1.5 – 20ms • B ~ 10 8 – 10 9 ms • B ~ 10 8 – 10 9 ms • High field • High field • P ~ 5 – 8 s • P ~ 5 – 8 s • B ~ few x 10 13 G • B ~ few x 10 13 G • Braking index n: • Braking index n: • • logPeriod derivative(ss logPeriod derivative(ss • n=3 magnetic dipole radiation • n=3 magnetic dipole radiation • Death line • Death line • Strong selection effects • Strong selection effects Period (sec) Period (sec)
Manifestations of NS: Spindown • Rotation driven: • “radio” pulsars (radio → γ rays) • magnetic torque • γγ → e+ e- + plasma instability ⇒ coherent radio • Accretion driven: • X-rays • LMXB, HMXB • Magnetic driven: Crustquakes? • Magnetars (AXPs, SGRs) • Spindown … but • Gravitational catastrophes? • Gamma-ray bursts, G.wave sources, hypernovae?
Spinup Quarks to Cosmos: Relevant Questions
• Did Einstein have the last word on Gravity?
• How do cosmic accelerators work and what are they accelerating?
• What are the new states of matter at exceedingly high density and temperature?
• Is a new theory of light and matter needed at the highest energies?
2 First Double Pulsar: J0737-3939
Lyne et al.(2004)
ω • Pb=2.4 hrs, d /dt=17 deg/yr Testing GR: •MA=1.337(5)M , M B=1.250(5)M obs s = ± Now to 0.1% exp .1 000 .0 002 s Kramer et al.(2004)
Double Neutron Star Binary: What are the Big Questions? J0737-0939A,B • Formation and Evolution: • What determines if a NS is born as a magnetar vs a canonical pulsar? • How fast do NS spin at birth? • How fast can recycled pulsars spin? • What is the role of instabilities and gravitational radiation in determining the spin state? • How do momentum thrusts during core collapse affect the resulting spin state and translational motion of the NS? • What processes determine the high space velocities of NS? » Neutrino emission » Matter rocket effects » Electromagnetic rocket effect (Harrison-Tademaru) » Gravitational-wave rocket effect • Are orbital spiral-in events at all related to high-energy bursts? (GRBs? Other transients?)
Palomar Bow Shocks Bow Shocks H image MSPs Guitar Nebula: Low V • Ordinary pulsar High • P = 0.68 s Edot • B = 2.6 x 10 12 G J2124-3358 B1957+20 (Kulkarni & Hester; Gaensler et al. J0437-47 • = 1.1 Myr HST Gaensler et al WFPC2 (Fruchter et al.) 33.1 -1 • E = I 10 erg s H • D 1.9 kpc (from DM) Mouse Duck • 1600 km s -1 at nominal distance • Will escape the Milky Way 1994 Radius of curvature of bowshock RXJ1856 nose increased from 1994 to 2001, corresponding to a 33% B0740-28 decrease in ISM density Canonical pulsars J0617 The pulsar is emerging from a 2001 High V, low to high Edot region of enhanced density Chatterjee & Cordes 2004
3 What are the Big Questions? What are the Big Questions?
• NS Structure: • NS as Laboratories: • Are neutron stars really neutron stars? • Can departures from General Relativity be identified in the orbits of compact binary pulsars? • What comprises the core of a NS? • Does the Strong-Equivalence Principle hold to high precision in • What is the mass distribution of NS? pulsars with WD or BH companions? • In what regions are the neutrons in a superfluid • NS as Gravitational Wave Detectors: state? • Use pulsars to detect long-period gravitational waves • How large are interior magnetic fields? » Early universe » Mergers of supermassive black holes • Magnetosphere and Emission Physics: » Topological defects (cosmic strings) • What QED processes are relevant for • Pulsars as Probes of Galactic Structure electromagnetic emissions? • What kind of spiral structure does the Galaxy have? • What is the nature of interstellar turbulence?
Forefronts in NS Science Forefronts in NS Science • Finding compact relativistic binary pulsars for • Understanding NS populations and their use as laboratories physical differences • Gravity • Radio pulsars and their progenitors • Relativistic plasma physics in strong B • Magnetars • Radio quiet/Gamma-ray loud objects • Finding spin-stable MSPs for use as • Branching ratios in supernovae gravitational wave detectors ( λ ~ light years) • h ~ σ T-1 (T = data span length) • The physics of NS runaway velocities TOA • Are “neutron stars” neutron stars? • Complete surveys of the transient radio sky • pulsars as prototype coherent radio emission
Currently about 1700 pulsars known Galactic birth rate ~ 1/100 yr × 10 Myr lifetime for canonical pulsars Fulfilling the Promise of NS Physics ⇒10 5 active pulsars × 20% beaming fraction and Astrophysics ⇒ 2×10 4 detectable pulsars + 10% more MSPs + NS-NS, NS-BH etc. • Find more pulsars • Time them with maximal precision • Phase-resolved polarimetry • VLBI them to get high astrometric precision
Step 1: conduct surveys that optimize the detection of faint, pulsed emission that is dispersed and that may or may not be periodic.
4 Arecibo + SKA Surveys Pulsar Search Domains
Region/Direction Kind of Pulsar Telescope Arecibo, Young pulsars Effelsburg, GBT, Galactic Plane (< 1 Myr) Jodrell, Parkes, WSRT Young, recycled, Galactic Center GBT, SKA binary, circum-SgrA* Moderate Galactic MSPs, binary, Arecibo, GBT, latitudes runaway Parkes Arecibo, GBT, Globular clusters MSPs, binary Parkes Local Group Young (probably) Arecibo, GBT, Galaxies Giant pulses SKA
A Single Dispersed Pulse from the Crab Pulsar Refractive indices for cold, magnetized plasma
Arecibo WAPP ν 2 ν 2 − ν 2 ν ν 3 nl,r ~ 1 - p / 2 + p B 2 ν ν ν ν >> p ~ 2 kHz >> B ~ 3 Hz ∆ S ~ 160 x Crab Nebula Group velocity ⇒ group delay = (time of arrival) ~ 200 kJy Detectable to ~ 1.5 Mpc t = t DM ± t RM birefringence with Arecibo ν -2 t DM = 4.15 ms DM ν -3 t RM = 0.18 ns RM -3 Dispersion Measure DM = ∫ ds ne pc cm -2 Rotation Measure RM = 0.81 ∫ ds ne B rad m
Refractive indices for cold, magnetized plasma Basic data unit = a dynamic spectrum 2 2 2 3 nl,r ~ 1 - np / 2n m np n B 2n 10 6 – 10 8 samples x 64 µs n >> np ~ 2 kHz n >> n B ~ 3 Hz Fast-dump spectrometers: ⇒ ∆ Group velocity group delay = (time of arrival) • Analog filter banks
• Correlators t = t DM ± t RM P • FFT (hardware) ν -2 • FFT (software) t DM = 4.15 ms DM • Polyphase filter bank ν -3 Frequency t RM = 0.18 ns RM 64 to 64 1024 channels ∫ -3 E.g. WAPP, GBT DM = ds ne pc cm time correlator + spigot card, new PALFA -2 RM = 0.81 ∫ ds ne B rad m correlator
5 P P Frequency Frequency
Time Time Interstellar Scintillation
RFI New Pulsars Known Pulsars
Periodicity Single-pulse search Arrival time Polarization Scintillation Search Monitoring Analysis Studies (matched filtering) (FFT)
Astrophysical effects are typically buried in noise and RFI
Issues In Pulsar Survey Optimization P Frequency • Broad luminosity function for pulsars • Beam luminosity Time • Geometric beaming • Pulse sharpness • Intrinsic pulse width W • Smearing from propagation effects » Dispersion across finite bandwidth (correctable) New Pulsars Known Pulsars » Multipath propagation (scattering in the ISM) • Smearing from orbital acceleration Periodicity Single-pulse search Arrival time Polarization Scintillation Search Monitoring Analysis Studies (matched filtering) • Intermittency of the pulsar signal (FFT) • Nulling, giant pulses, precession, eclipsing • Interstellar scintillation
Issues In Pulsar Survey Optimization Dedispersion Two methods: • Combine the signal over time and frequency while Coherent : maximizing S/N through matched filtering: • operates on the voltage proportional to the electric • Dedispersion : sum over frequency while removing the dispersive field accepted by the antenna, feed and receiver time delays • computationally intensive because it requires • Single pulses : match the shape and width of the pulse sampling at the rate of the total bandwidth • Periodic pulses : match the period as well as the pulse shape and width • “exact” • Orbital motion : match the change in pulse arrival times related to Post-detection : the changing Doppler effect 2 ⇒ Single-pulse searches: • operates on intensity = |voltage| Search vs. (DM, W) • computationally less demanding ⇒ Periodicity searches: • an approximation Search vs. (DM, W, P, [orbital parameters])
6 Basic data unit = a dynamic spectrum Dispersed Pulse Coherently dedispersed pulse
10 6 – 10 8 samples x 64 µs Fast-dump spectrometers: • Analog filter banks
• Correlators ∆∆∆t = 8.3 µµµs DM ννν-3 ∆ν∆ν∆ν P • FFT (hardware) • FFT (software)
• Polyphase filter bank Frequency 64 to 64 1024 channels E.g. WAPP, GBT time correlator + spigot card, new PALFA correlator
Coherent Dedispersion Coherent Dedispersion pioneered by Tim Hankins (1971) pioneered by Tim Hankins (1971) Dispersion delays in the time domain represent a phase Coherent dedispersion works by explicit deconvolution: perturbation of the electric field in the Fourier domain:
Coherent dedispersion involves multiplication of Fourier amplitudes by the inverse function, Comments and Caveats : For the non-uniform ISM, we have • Software implementation with FFTs to accomplish deconvolution (Hankins 1971) • Hardware implementations: real-time FIR filters (e.g. Backer et al. 1990s-present) which is known to high precision for known pulsars. • Resulting time resolution: 1 / (total bandwidth) The algorithm consists of • Requires sampling at Nyquist rate of 2 samples × bandwidth ⇒Computationally demanding Application requires very fast sampling to achieve usable • Actual time resolution often determined by interstellar scattering (multipath) bandwidths. • Most useful for low-DM pulsars and/or high-frequency observations
Micropulses coherently dedispersed Nanostructure in Crab pulsar giant pulses (Hankins1971)
7 Postdetection Dedispersion: Crab 2 -ns resolution Sum intensity over frequency after correcting for dispersion delay
Interstellar scattering from electron density variations Interstellar Scattering Effects
• Angular broadening (seeing) • Pulse broadening • Diffractive interstellar scintillations (DISS) • Refractive interstellar scintillations (RISS) • TOA fluctuations (multiple effects) • Superresolution phenomena: • Pulsar velocities >> ISM, observer velocities stars twinkle, planets don’t • Scattering is strong for frequencies < 5 GHz ⇒⇒⇒ pulsars show DISS, AGNs don’t • Electron density irregularities exist on scales from ~ 100’s km to Galactic scales
Pulse broadening from interstellar scattering:
Arecibo WAPP data, Bhat et al 2004
8 Choose ∆∆∆ t ⇒⇒⇒ maximum DM Dedispersion at a single known DM
Scattering limited Frequency Frequency time time Σ ν
Dispersion limited I(t) time
Dedispersion over a set of DMs Single Pulse Studies & Searches DM Frequency time time
Giant pulse from the Crab pulsar Nano-giant pulses (Hankins et al. 2003) Arecibo WAPP S ~ 160 x Crab Nebula ~ 200 kJy Arecibo Detectable to ~ 1.5 Mpc with Arecibo 5 GHz 2-ns giant pulses from 0.5 GHz bw the Crab: (Hankins et al. 2003) coherent Giant Pulses seen dedispersion from B0540-69 in LMC (Johnston & Romani 2003)
9 Giant pulses from M33 Single pulse searches Arecibo observations (Mclaughlin & Cordes 2003)
A pulsar found through its single - pulse emission, not its periodicity (c.f. Crab Pulsar Periodicity giant pulses). Searches
Algorithm: matched filtering in the DM -t plane.
ALFA’s 7 beams provide powerful discrimination between celestial and RFI transients
Example Periodicity Search Pulsar Periodicity Search Algorithm DM Frequency time time
FFT each DM’s time series |FFT(f)|
1/P 2/P 3/P •••
10 Harmonic Sum
The FFT of periodic pulses is a series of spikes (harmonics) separated by 1/P.
To improve S/N, sum harmonics . This procedure is an approximation to true matched filtering, which would give optimal S/N.
Sum how many harmonics?
The answer depends on the pulse “duty cycle” = (pulse width / P) (unknown a priori )
⇒ need to use trial values of Nh:
Sum over harmonics
Maximize h() with respect to Nh to identify candidate pulsars. Noise and RFI conspire to yield spurious candidates. ∴ Need a high threshold. How high? Minimum detectable flux density for a single harmonic:
Minimum detectable flux density for harmonic sum:
Example Time Series and Power Spectrum for a recent PALFA discovery Example Time Series and Power Spectrum for a recent PALFA discovery (follow-up data set shown) (follow-up data set shown)
DM = 0 pc cm -3 Time Series DM = 217 pc cm -3 DM = 0 pc cm -3 Time Series DM = 217 pc cm -3
Where is the pulsar? Here is the pulsar
Pulse Survey Selection Against Binaries shape NS-NS binary
Effects that Pulse shape broaden pulses reduce the harmonic sum, which is Phase perturbation bad
FFT FFT harmonics
Harmonic Harmonic Sum sum
11 Dealing With Orbital Motion How Far Can We Look? 1/2 1/4 Orbital acceleration yields a time-dependent Dmax = D (S / S min1 ) Nh period, potentially destroying the power of the S = single harmonic threshold = m S /( ∆ν∆ν∆ν T) 1/2 straightforward FFT + HS. min1 sys m = no. of sigma ~ 10 • Long-period binaries : T = data span length << Porb Nh = no. of harmonics that maximize • Do nothing different harmonic sum • Intermediate-period orbits : T < 0.1 Porb →→→ • Acceleration search: compensate the time domain or match filter in Nh 0 for heavily broadened pulses (scattering) the frequency domain according to an acceleration parameter • Adds another search parameter: DM, P, W, a Regimes: • Very short period orbits : T > P (potentially >> P ) orb orb ∝∝∝ -1/2 Luminosity limited Dmax Smin1 • Do conventional FFT but search for orbital sidebands ∝∝∝ -x DM/SM limited Dmax Smin1 , x<1/2
Dmax vs. Flux Density Threshold Implications: • Optimal integration time: go no deeper than the luminosity limited regime
Scattering limited • Fast-dump spectrometers: need enough channels so that search is not DM limited
Dispersion Luminosity limited limited • Better to cover more solid angle than to integrate longer on a given direction (as long as all solid angles contain pulsars)
AO at S,L,P bands Add slides showing sensitivity curves for Arecibo
12 Survey Selection Against Binaries NS-NS binary Hardware for Pulsar Science
Pulse shape Predetection Samplers and Analyzers: • ASP, GASP (Arecibo & Green Bank) » Real-time dedispersion and folding Phase perturbation • New Mexico Tech burst sampler » Off-line dedispersion • Generic baseband samplers (c.f. radar samplers) FFT harmonics Postdetection Samplers: • WAPP (Arecibo), SPIGOT (GBT) (correlators) » Searching and timing machines Harmonic Sum • New PALFA spectrometer (polyphase filter bank) » Primarily a search machine
Software for Pulsar Searching
• Many proprietary packages • Sigproc/Seek package • PRESTO • Cornell Code • Berkeley Code • PALFA: the PALFA Consortium is testing and consolidating codes to produce a new “standard” pulsar search package
The power of ALFA: I( ννν, t, θθθ ) j=1,7 j Massive ALFA Pulsar Surveys
10 3 new pulsars – Reach edge of Galactic population for much of pulsar luminosity function – High sensitivity to millisecond pulsars
–Dmax = 2 to 3 times greater than for Parkes MB Sensitivity to transient sources Commensal SETI Search (Wertheimer UCB) Data management: – Keep all raw data (~ 1 Petabyte after 5 years) at the Cornell Theory Center (CISE grant: $1.8M) – Database of raw data, data products, end products – Web based tools for Linux-Windows interface (mysql ↔ ServerSql) – VO linkage (in future)
13 ALFA pulsar surveys will be the deepest Example of the SKA as a surveys of the Galaxy until the SKA is built: Pulsar-Search Machine ~10 4 pulsar detections with the SKA (assuming all-sky capability) Blue: known pulsars • rare NS-NS, NS-BH binaries for (prior to Parkes MB) probing strong-field gravity • millisecond pulsars < 1.5 ms • MSPs suitable for gravitational wave Red: Parkes MB detection • Galactic tomography of electron density and magnetic field Green: PALFA • Spiral-arm definition simulated pulsars
Blue points : SKA simulation Black points : known pulsars
SKA: What is It? Pulsar Distances
• An array telescope that combines complete sampling of the time, frequency and spatial domains Type Number Comments with a ×20-100 increase in collecting area (~ 1 km 2) over existing telescopes. • Frequency range 0.1 – 25 GHz (nominal) Parallaxes: 1 mas @ 1 kpc • Limited gains from reducing receiver noise or Interferometry ~13 increasing bandwidth once the EVLA is finished 1.6 µs @ 1 kpc • Innovative design needed to reduce cost timing ~ 5 HST, point spread function • 10 6 meter 2 ⇒ ~ €1,000 per meter 2 optical ~ 1 2 • c.f. existing arrays ~ €10,000 per meter SNRs 8 false associations • An international project from the start ×20 ×50 Associations • International funding GCs 16 • Cost goal ~ € 1 billion • 17-country international consortium LMC,SMC ~8 • Executive, engineering, science, siting, simulation 74 bright pulsars, galactic groups HI absorption • Timeline for construction extends to 2020 rotation model • Can be phased for different frequency ranges all radio pulsars ISM perturbations • Can do science as you go DM + ne model • The existing US radio astronomy portfolio is the (~ 1400) foundation on which to build the SKA
NE2001: Galactic Distribution of Free Electrons + Fluctuations
Paper I = the model (astro-ph/0207156) Paper II = methodology & particular lines of sight (astro- ph/0301598) Based on ~ 1500 lines of sight to pulsars and extragalactic objects Code + driver files + papers: www.astro.cornell.edu/~cordes/NE2001
14 But … if you want a good distance, PSR B0919+06 Very Long S. Chatterjee et al. (2001) measure the parallax ! µµµ = 88.5 ±±± 0.13 mas/yr Baseline Array πππ = 0.83 ±±± 0.13 mas D = 1.2kpc e.g. Arecibo + GBT + VLA ϕϕϕ V = 505 km/s + VLBA
will be a powerful parallax and proper motion machine
Proper motion PSR B1929+10 and parallax Chatterjee using the VLBA et al. 2003 (Brisken et al. 2001)
15