EDP Sciences
Free Access
Issue
A&A
Volume 525, January 2011
Article Number A33
Number of page(s) 7
Section Galactic structure, stellar clusters and populations
DOI https://doi.org/10.1051/0004-6361/201015279
Published online 29 November 2010

© ESO, 2010

1. Introduction

Identification of the Galactic sources emitting γ-rays with energies above 100 MeV is a long-standing problem in astrophysics. With the launch of the AGILE γ-ray telescope in 2007 (Tavani et al. 2009a) and Fermi in 2008 (Atwood et al. 2009), a great deal of progress has been made. In particular, Fermi and AGILE have shown that the vast majority of Galactic γ-ray sources are probably pulsars and pulsar wind nebulae (Abdo et al. 2009a). However, new source classes have also been identified, including microquasars such as Cygnus X-1 (Sabatini et al. 2010), Cygnus X-3 (Tavani et al. 2009c; Abdo et al. 2009c), LS5039 (Abdo et al. 2009d) and LSI + 61°303 (LSI+61 303) (Abdo et al. 2009e), and Wolf-Rayet stars such as Eta Carinae (Tavani et al. 2009b).

As the identification of the remaining unidentified EGRET sources and the newly found AGILE and Fermi-LAT sources continues, the problem of source confusion will become increasingly serious. Although the angular resolution of both AGILE and Fermi are higher than that of EGRET, they also are sensitive to much lower fluxes, Fermi-LAT in particular. As a result, γ-ray error contours will not only continue to contain many plausible source counterparts, but will be increasingly likely to contain multiple real γ-ray sources above the nominal flux threshold.

thumbnail Fig. 1

SNR Gamma Cygni (G78.2+2.1) in galactic coordinates. DRAO Radio telescope, wavelength = 21.1 cm. White contour levels: AGILE-GRID intensity contour levels – related to Fig. 2 – (pixel size 0.1°), starting from 0.00085 in steps of 0.00002 (intensity per pixel); green contour: AGILE-GRID 95% confidence level for E ≥ 100   MeV; black circle: LAT PSR J2021+4026.

Open with DEXTER

In this paper we provide evidence that 1AGL J2022+4032 in the Cygnus region1 (Pittori et al. 2009) may be one of these sources. We present a study of the time variability of the γ-ray flux of 1AGL J2022+4032 based on long term observations of the source by AGILE. This source was first discovered by COS-B as 2CG078+2, and was listed as the unidentified source 3EG J2020+4017 in the 3rd EGRET catalog (Hartman et al. 1999). It is located (Fig. 1) within the wide radio shell of the prominent SNR Gamma Cygni (SNR G78.2+2.1). Searches for X-ray counterparts of the source were performed with a number of X-ray telescopes, including ROSAT (Brazier et al. 1996), ASCA (Uchiyama et al. 2002), INTEGRAL (Bykov et al. 2004), and Chandra (Becker et al. 2004; Weisskopf et al. 2006). Using a blind search technique, Fermi-LAT was finally able to identify this source, 1FGL J2021.5+4026 (Abdo et al. 2009a, 2010a), as γ-ray pulsar LAT PSR J2021+4026 (Abdo et al. 2009b). The γ-ray flux and spectra are consistent with a pulsar origin. However, γ-ray pulsars show no variability over time scales of weeks to months; indeed, significant variability on these time scales would be difficult to reconcile with theoretical models.

Beginning in November 2007, AGILE data from this source have revealed signs of variable γ-ray emission, as reported in ATel #1492 (Longo et al. 2008), #1547 (Giuliani et al. 2008) and #1585 (Chen et al. 2008). The evidence for γ-ray variability combined with its relatively high unpulsed γ-ray fraction could indicate either the presence of an additional γ-ray source coincident with 1FGL J2021.5+4026 or variability of the γ-ray pulsar LAT PSR J2021+4026.

2. Observations and data analysis

thumbnail Fig. 2

Cygnus region in galactic coordinates, γ-ray intensity map for E ≥ 100   MeV. Deep integration AGILE-GRID data (November 2007–August 2009). Pixel size = 0.1° with 3-pixel Gaussian smoothing. Green contours: AGILE-GRID 95% confidence level. Cyan contours: AGILE-GRID statistical + systematic error (0.1°). Black contours: Fermi-LAT (1-year catalog), statistical error only, crosses shown for contours too small to be visible. The Green contours have been calculated with a multi-source likelihood analysis, using four persistent sources (Table 1).

Open with DEXTER

Table 1

γ-ray sources in the Cygnus region.

AGILE has been in orbit since April 2007 and has observed the Cygnus region numerous times, beginning in November 2007. We used these observations to characterize the γ-ray variability of the sources in the Cygnus region using the following procedure. First, we used the software tool , one of the AGILE Scientific Tools, to perform multi-source likelihood analysis on the deep-integration AGILE-GRID data using the filter. This analysis revealed γ-ray emission from four point sources: 1AGL J2022+4032, 1AGL J2021+3652, 1AGL J2032+4102 and a persistent faint source consistent with the position of Cygnus X-3 (Fig. 2). The average fluxes and positions are shown in Table 1. Fermi source 1FGL J2020.0+4049 was not detected in either energy range due to the combination of its proximity to 1AGL J2022+4032 and its low flux.

thumbnail Fig. 3

Flux for E ≥ 100   MeV on six-day time intervals from 1AGL J2022+4032 (left) and 1AGL J2021+3652 (right). The Red line indicates the weighted mean of the 42 six-day fluxes, from which the χ2 calculated was calculated: (118 ± 4) × 10-8   photons   cm-2   s-1 for 1AGL J2022+4032, and (54 ± 3) × 10-8   photons   cm-2   s-1 for 1AGL J2021+3652.

Open with DEXTER

Next, we divided the observations from November 2007 to August 2009 into 42 discrete fixed-length time intervals of  ≈6 days (≈90 orbits) each, analyzing the AGILE γ-ray flux from the position of 1AGL J2022+4032, (l,b) = (78.23,   2.12), for E ≥ 100   MeV and E ≥ 400   MeV, while keeping its position fixed and all nearby sources fixed in flux and position. We performed the same analysis on the nearby γ-ray source 1AGL J2021+3652, 3.5° away from 1AGL J2022+4032, which Halpern et al. (2008) identified as PSR J2021+365, in order to account for the effects of systematic errors. Figure 3 shows the light curves of the two sources.

thumbnail Fig. 4

Flux for E ≥ 400   MeV on six-day time intervals from 1AGL J2022+4032 (left) and 1AGL J2021+3652 (right). The Red line indicates the weighted mean of the 42 individual fluxes, from which the χ2 was calculated: (29 ± 1) × 10-8   photons cm-2 s-1 for 1AGL J2022+4032, and (15 ± 1) × 10-8   photons   cm-2   s-1 for 1AGL J2021+3652.

Open with DEXTER

For E ≥ 400   MeV, only the emission from the three steady sources shown in the lower half of Table 1 were considered; J2033+4050 is not significantly detected for E ≥ 400   MeV. In Fig. 4 we compare the light curves of 1AGL J2022+4032 and 1AGL J2021+3652.

3. Discussion

3.1. Variability

Table 2

Variability analysis with and without 10% systematic errors on fluxes.

We used the method developed by McLaughlin et al. (1996) to test the γ-ray flux variability of 1AGL J2022+4032 with respect to 1AGL J2021+3652. A similar analysis for all the 1AGL sources is in preparation (Verrecchia et al., in prep.). The weighted mean flux is calculated from the fluxes in each 6-day time interval and their corresponding errors, from which the χ2 is derived. Q is the probability that an intrinsically non-variable source (i.e. with constant flux) would produce by random chance a measured value of χ2 greater than or equal to the χ2 observed, and the variability index V is defined as V =  −log Q. A source can be classified as nonvariable if V < 0.5, uncertain if 0.5 ≤ V < 1, or variable if V ≥ 1. The value V = 1 corresponds to a probability of variability Pvar = 1 − Q of 90%. Table 2 shows the value of V for 1AGL J2022+4032 and 1AGL J2021+3652 for E ≥ 100   MeV and E ≥ 400   MeV. For 1AGL J2022+4032 we find V = 2.18 when systematic effects are included (V = 3.88 for statistical only). For 1AGL J2021+3652 the corresponding values are V = 0.30 (V = 0.48).

As a cross-check, we also calculated a complementary variability index, VF, according to the formula used in the Fermi catalogs (Abdo et al. 2009a, 2010a). This index is a simple χ2 where the weights include the systematic error, frel which in our case is 10%, and the number of degrees of freedom is 41. The values, associated probabilities, and fractional excess variabilities above systematic and statistical fluctuations are found in Table 2.

We find evidence for variability for E ≥ 100   MeV in the emission from 1AGL J2022+4032 even allowing for systematic errors on the level of 10%. Any systematic effects that would influence the measurement of the flux should also have affected the nearby source 1AGL J2021+3652, for which no corresponding variability is found. However, 1AGL J2021+3652 is only half as bright as 1AGL J2022+4032. Similarly, although we found no evidence for variability in the flux of 1AGL J2022+4032 for E ≥ 400   MeV when systematic errors are taken into account,the average flux is only a quarter that of E ≥ 100   MeV. In both cases, the same intrinsic variability might be rendered undetectable because of reduced photon statistics.

To determine whether the variability of 1AGL J2022+4032 for E ≥ 100   MeV would have been detectable if the source were half as bright (Case 1) or at higher energies (Case 2), we produced simulated observations of hypothetical sources with the same intrinsic variability, where “intrinsic” includes the effects of systematic errors. The meaning of “same” is somewhat ill-defined, since many qualitatively and quantitatively different parent distributions can produce the same variability index, or even the same observed fluxes within errors. In order to simplify the relationship with the calculation of χ2, we assume that the intrinsic flux in each observation is drawn from a Gaussian distribution whose mean is equal to the mean flux of the source in order to find the variance which reproduces the observed value of V = 3.88 for 1AGL J2022+4032. We determined that, given the exposures of each of the 42 time intervals, an intrinsic variability of 26% (square root of variance 33.8 × 10-8   photons   cm-2   s-1 for 131 × 10-8   photons   cm-2   s-1, the mean flux of 1AGL J2022+4032 for E ≥ 100   MeV) produces a distribution in V with a median equal to the observed value, 3.88 (Fig. 5, solid black).

thumbnail Fig. 5

Simulated cumulative distributions (10 000 trials) of the variability index V versus fraction of trials, or probability. Solid black: mean flux 131 × 10-8   photons   cm-2   s-1 and square root of variance 33.8 × 10-8   photons   cm-2   s-1 for E ≥ 100   MeV; dotted red: mean flux 60 × 10-8   photons   cm-2   s-1 and square root of variance 15.5 × 10-8   photons   cm-2   s-1 for E ≥ 100   MeV; dashed green: mean flux 33 × 10-8   photons   cm-2   s-1 and square root of variance 8.5 × 10-8   photons   cm-2   s-1 for E ≥ 400   MeV; dot-dahsed blue: mean flux 17 × 10-8   photons   cm-2   s-1 and square root of variance 4.4 × 10-8   photons   cm-2   s-1 for E ≥ 400   MeV. The vertical lines indicate the observed values of V and the associated probability for the corresponding dataset (Table 2).

Open with DEXTER

For Case 1, we simulated 10 000 series of 42 observations for a source with intrinsic fluxes taken from a Gaussian distribution with mean equal to 60 × 10-8   photons   cm-2   s-1 and square root of variance of 15.5 × 10-8   photons   cm-2   s-1 using the same exposures as 1AGL J2022+4032 for E ≥ 100   MeV (Fig. 5, dotted red). The median value of V is 0.50. V ≤ 0.48 is produced in 49% of trials, while V ≥ 1.0 in 28% of trials. The same level of intrinsic variability would be likely to produce a variability index similar to that of 1AGL J2021+3652, but would not be likely to be classified as variable, if the source were half as bright. For Case 2, we simulated 10 000 series of 42 observations for a source with intrinsic fluxes taken from a Gaussian distribution with mean equal to 33 × 10-8   photons   cm-2   s-1 and square root of variance 8.4 × 10-8   photons   cm-2   s-1, using the exposures of 1AGL J2022+4032 for E ≥ 400   MeV (Fig. 5, dashed green). The median value of V is 1.75. V ≤ 1.11 is produced in 31% of trials, while V ≥ 1.0 in 74% of trials. The same level of flux variability at high energies would be detected more often than not. Equivalent intrinsic variability at the flux level of 1AGL J2021+3652 for E ≥ 400   MeV would be completely undetectable (Fig. 5, dot-dashed blue).

3.2. Analysis of pulsed and unpulsed components

We attempted to perform an analysis of the on- and off-peak components of the 1AGL J2022+4032 AGILE data with respect to the Fermi γ-ray ephemeris of LAT PSR J2021+4026 in order to test whether the apparent flux variability is due to a source other than the pulsar. However, as shown in Table 3 of the Fermi Pulsar Catalog (Abdo et al. 2010c), LAT PSR J2021+4026 is one of only two Fermi pulsars for which the γ-ray phase profile is difficult to separate into on-peak and off-peak phases, both because the γ-ray peaks of the pulsar are very broad and because the γ-ray emission has a high unpulsed fraction (Fig. 6). We attempted a number of cuts on the AGILE data. Varying the off-peak exposure fraction from 10% to 45% yielded a monotonic variation in flux with no obvious plateau. The counts maps show evidence of the pulsar contribution for all but the lowest off-peak exposure fraction, for which the statistical significance over two years of data is insufficient to perform a variability analysis. For these reasons, a credible analysis of the on-peak and off-peak variability based on the AGILE data was not feasible.

thumbnail Fig. 6

γ-ray phase profile of PSR J2021+4026 as seen by AGILE (top) and Fermi (bottom). The AGILE profile is for photons with E ≥ 100 MeV while the Fermi profile, taken from Abdo et al. (2009b), is for E ≥ 300  MeV.

Open with DEXTER

4. Discussion of possible counterparts

In Fig. 7, we show the X-ray sources listed in Weisskopf et al. (2006) as possible counterparts for 3EG J2020+4017. The AGILE contour of the persistent source is consistent with the position of the X-ray source [WSC2006] S21, which has been associated with the LAT PSR J2021+4026 whose pulsations were discovered by Fermi. Trepl et al. (2010) searched the XMM-Newton archival data and found 2XMM J202131.0+402645, a point source coincident with S21. However, re-analyzing the Chandra data, they found that [WSC2006] S25, a strong point source within the Fermi 0FGL error box although outside the Fermi 1FGL error box, showed evidence of variability during the Chandra observation. In addition, S25 is not visible in the XMM data, indicating long-term X-ray variability. If S25 is also a variable γ-ray source, it would be within the AGILE source location accuracy of  ~1° for the 1–2 week observation durations, and could be responsible for the variability apparently observed by AGILE below 400 MeV. However, S25 has an infrared counterpart and could be a normal star.

The Fermi 11-month source catalog (Abdo et al. 2010a) detected a nearby source, 1FGL J2020.0+4049, associated with TeV source VER J2019+407, that is within the one-week error circle of AGILE. It was not detected by AGILE, probably because of its hard spectrum with index −2.12 ± 0.08. Its Fermi light curve is consistent with no variability at a level of 33%; however, its measured flux for E > 100 MeV decreased from (22 ± 6) × 10-8   photons   cm-2   s-1 in the first month to below 4.3 × 10-8   photons   cm-2   s-1 by the seventh month. An intrinsic flux variability of 34 × 10-8   photons   cm-2   s-1 over the longer AGILE observation period would explain the apparent flux variability of 1AGL J2022+4032.

Detection of simultaneous flaring in γ-ray and, e.g., X-rays would provide a definitive identification. However, soft X-ray instruments have fields of view too narrow to make triggered observations practical. We note that a special class of microquasars may not necessarily produce simultaneous hard X-ray emission (Romero & Vila 2009, hereafter RV09).

4.1. A variable pulsar

It is possible that the pulsar LAT PSR J2021+4026 itself has variable γ-ray emission for the energy range below 400 MeV. This scenario, if confirmed, would open a new field of investigation of γ-ray pulsars. However, in the absence of other information supporting this hypothesis pulsar γ-ray variability of LAT PSR J2021+4026 is unlikely. This conclusion is based on both previous observational evidence from EGRET and from general theoretical considerations. In particular, the spin period and time derivative (Abdo et al. 2009b) of LAT PSR J2021+4026 (P = 265 ms, = 54.8 × 10-15) assign a unremarkable position to this pulsar in both the P −  diagram and in the pulsar γ-ray luminosity vs. period (or Goldreich-Julian current diagram). Instabilities in the γ-ray pulsed emission of LAT PSR J2021+4026 might be associated with radio and/or X-ray signal changes during the period of the detected γ-ray variability. Future monitoring of LAT PSR J2021+4026 can contribute to test this fascinating hypothesis.

thumbnail Fig. 7

Possible X-ray counterparts from Weisskopf et al. (2006). Green contour level: AGILE-GRID 95% C.L. (only statistical) for 1AGL J2022+4032; cyan circle: AGILE-GRID errorbox (statistical + systematic error); red circle: EGRET errorbox for 3EG J2020+4017; black circle: Fermi errorbox for 1FGL J2021.5+4026.

Open with DEXTER

4.2. Background blazar

Another possibility is that a blazar behind the Galactic plane is contributing variable γ-ray emission to the measured flux of 1AGL J2022+4032. There are a number of known blazars in the Cygnus region, but they are well outside the typical AGILE one-week error circle. Most optical and radio AGN catalogs avoid the Galactic plane because of the high concentration of Galactic sources, heavy extinction, and/or diffuse radio emission (Abdo et al. 2010b). Each of these difficulties is especially acute at the position of 1AGL J2022+4032, which is located within the Gamma-Cygni supernova remnant shell (see Fig. 1). Trepl et al. (2010) found multiple areas of concentrated radio emission within the Fermi error box of 1FGL J2021.5+4026 as well as evidence of variable X-ray emission from nearby source [WSC2006] S25. In addition, the γ-ray flux could be well below the detection threshold of AGILE (which, in the presence of a known, bright γ-ray source is quite high) and still contribute to the overall γ-ray variability within the one-week AGILE error box. Knowing that the intrinsic distribution of blazars should be isotropic, we can use the AGN associations above  |b|  > 10° in the First AGN Catalog (Abdo et al. 2010b, hereafter 1LAC) to estimate the probability of finding at least one blazar within the  ≈1° error box of a one-week observation with AGILE. The probability of finding a blazar similar to the 599 1LAC associations is  ≈0.05. A more conservative estimate using only the 281 1LAC associated FSRQs, yielding a probability of  ≈0.02 of finding at least one FSRQ, would better represent the need for γ-ray variability and the non-detection by Fermi. In either case, the probability of chance coincidence is quite low.

4.3. An X-ray quiet microquasar

Taking into account both the AGILE-GRID emission above 100 MeV and the Super-AGILE upper limit in the 15–60 keV range (~60 mCrab), we consider the possibility that the detected γ-ray variability is caused by transient activity of an X-ray quiet microquasar. RV09 analyzed several Galactic sources with variable emission in the γ-ray energy range and showing a ratio Lγ/LX ≫ 1. They proposed that this kind of emission (shortly variable, X-ray quiet, ...) can be produced by proton-dominated jets in a special class of Galactic microquasars. The bulk of the emission at γ-ray energies is produced by hadronic jets emitted from an accreting source. The model of RV09 predicts a γ-ray luminosity for this process on the order of  ≈1034 erg/s. Assuming the presence of a γ-ray source (an X-ray quiet microquasar) within the error box of LAT PSR J2021+4026, we find that that it is required to be at a distance of  ≈300 pc from the Earth, i.e., closer than the pulsar (1–2 kpc).

The probability of finding this particular type of microquasar within the error box of 1AGL J2022+4032 is difficult to quantify. Nevertheless, because X-ray binaries are concentrated in the star-forming regions in the Galactic plane, and high-mass X-ray binaries particularly along tangents of spiral arms such as the Cygnus region (Liu 2007, 2006), the likelihood that there is an appropriate microquasar within the error box is much higher than that of blazars, which are isotropically distributed. Similar reasoning applies to such possible source types such as massive stellar winds (Tavani et al. 2009b) and novae (Abdo et al. 2010d).

5. Conclusions

The apparent γ-ray flux variability of 1AGL J2022+4032 in the 100–400 MeV range as seen by AGILE leads us to consider two possible explanations. One is that the Fermi pulsar, LAT PSR J2021+4026 itself has a variable γ-ray flux. This behavior would be highly unusual given the properties of γ-ray (and radio) pulsars with similar characteristics. Based on the large γ-ray unpulsed fraction of LAT PSR J2021+4026 as seen by its folded light curve (Abdo et al. 2009b), we believe that it is more likely that another variable γ-ray source within the error box of LAT PSR J2021+4026 contributes to the total emission of 1AGL J2022+4032. 1AGL J2022+4032 was not detected in the hard X-rays during the AGILE observation periods, with upper limits in Super-AGILE varying from 10 to 50–60 mCrab in the 15–60 keV range. Therefore, if the variable source is a microquasar rather than an unidentified low-frequency blazar, it would be an X-ray-quiet γ-ray variety, possibly of the type proposed by RV09.

We note that during the period in which 1AGL J2022+4032 was observed by Fermi (starting in August 2008), its flux in AGILE-GRID was slightly more stable. In fact, using only the AGILE fluxes for E ≥ 100   MeV from the 28 time intervals during the Fermi observations, we find that the variability V was 1.65 without systematic errors and 0.96 with systematic errors, while in the period before Fermi the variability V was 1.94 without systematic errors and 1.23 with systematic errors. Future observations of 1AGL J2022+4032 by both AGILE and Fermi will reveal whether this fascinating source continues to show evidence over the long term.


1

Throughout this paper we refer to the Cygnus region, although physically this direction lies along the tangents of at least three different Galactic spiral arms at different distances, from the local  ≈1 kpc Orion spur, to the Perseus arm, to the Outer arm at  > 8 kpc.

References

All Tables

Table 1

γ-ray sources in the Cygnus region.

Table 2

Variability analysis with and without 10% systematic errors on fluxes.

All Figures

thumbnail Fig. 1

SNR Gamma Cygni (G78.2+2.1) in galactic coordinates. DRAO Radio telescope, wavelength = 21.1 cm. White contour levels: AGILE-GRID intensity contour levels – related to Fig. 2 – (pixel size 0.1°), starting from 0.00085 in steps of 0.00002 (intensity per pixel); green contour: AGILE-GRID 95% confidence level for E ≥ 100   MeV; black circle: LAT PSR J2021+4026.

Open with DEXTER
In the text
thumbnail Fig. 2

Cygnus region in galactic coordinates, γ-ray intensity map for E ≥ 100   MeV. Deep integration AGILE-GRID data (November 2007–August 2009). Pixel size = 0.1° with 3-pixel Gaussian smoothing. Green contours: AGILE-GRID 95% confidence level. Cyan contours: AGILE-GRID statistical + systematic error (0.1°). Black contours: Fermi-LAT (1-year catalog), statistical error only, crosses shown for contours too small to be visible. The Green contours have been calculated with a multi-source likelihood analysis, using four persistent sources (Table 1).

Open with DEXTER
In the text
thumbnail Fig. 3

Flux for E ≥ 100   MeV on six-day time intervals from 1AGL J2022+4032 (left) and 1AGL J2021+3652 (right). The Red line indicates the weighted mean of the 42 six-day fluxes, from which the χ2 calculated was calculated: (118 ± 4) × 10-8   photons   cm-2   s-1 for 1AGL J2022+4032, and (54 ± 3) × 10-8   photons   cm-2   s-1 for 1AGL J2021+3652.

Open with DEXTER
In the text
thumbnail Fig. 4

Flux for E ≥ 400   MeV on six-day time intervals from 1AGL J2022+4032 (left) and 1AGL J2021+3652 (right). The Red line indicates the weighted mean of the 42 individual fluxes, from which the χ2 was calculated: (29 ± 1) × 10-8   photons cm-2 s-1 for 1AGL J2022+4032, and (15 ± 1) × 10-8   photons   cm-2   s-1 for 1AGL J2021+3652.

Open with DEXTER
In the text
thumbnail Fig. 5

Simulated cumulative distributions (10 000 trials) of the variability index V versus fraction of trials, or probability. Solid black: mean flux 131 × 10-8   photons   cm-2   s-1 and square root of variance 33.8 × 10-8   photons   cm-2   s-1 for E ≥ 100   MeV; dotted red: mean flux 60 × 10-8   photons   cm-2   s-1 and square root of variance 15.5 × 10-8   photons   cm-2   s-1 for E ≥ 100   MeV; dashed green: mean flux 33 × 10-8   photons   cm-2   s-1 and square root of variance 8.5 × 10-8   photons   cm-2   s-1 for E ≥ 400   MeV; dot-dahsed blue: mean flux 17 × 10-8   photons   cm-2   s-1 and square root of variance 4.4 × 10-8   photons   cm-2   s-1 for E ≥ 400   MeV. The vertical lines indicate the observed values of V and the associated probability for the corresponding dataset (Table 2).

Open with DEXTER
In the text
thumbnail Fig. 6

γ-ray phase profile of PSR J2021+4026 as seen by AGILE (top) and Fermi (bottom). The AGILE profile is for photons with E ≥ 100 MeV while the Fermi profile, taken from Abdo et al. (2009b), is for E ≥ 300  MeV.

Open with DEXTER
In the text
thumbnail Fig. 7

Possible X-ray counterparts from Weisskopf et al. (2006). Green contour level: AGILE-GRID 95% C.L. (only statistical) for 1AGL J2022+4032; cyan circle: AGILE-GRID errorbox (statistical + systematic error); red circle: EGRET errorbox for 3EG J2020+4017; black circle: Fermi errorbox for 1FGL J2021.5+4026.

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.