1: \documentclass[apj]{emulateapj} \slugcomment{Submitted to the
2: Astrophysical Journal on June 09, 2008}
3: \usepackage{graphicx,amssymb,amsmath,times}
4:
5: \setcounter{page}{1}
6:
7: \def\Journal#1#2#3#4{{#4}, {#1}, {#2}, #3}
8:
9: \newcommand{\lsi}{LS~I~+61$^{\circ}$303}
10: \newcommand{\lomb}{Lomb-Scargle}
11: \newcommand{\eg}{3EG~J0241$+$6103}
12:
13: \begin{document}
14: \title{Periodic very high energy $\gamma$-ray emission from \lsi\ observed with the MAGIC telescope}
15: \shorttitle{Periodic VHE $\gamma$-ray emission from \lsi\ }
16: \shortauthors{J. Albert et~al.}
17:
18:
19: % authors 29.4.2008 Format ApJ
20: %
21: \author{
22: J.~Albert\altaffilmark{a},
23: E.~Aliu\altaffilmark{b},
24: H.~Anderhub\altaffilmark{c},
25: L.~A.~Antonelli\altaffilmark{d},
26: P.~Antoranz\altaffilmark{e},
27: M.~Backes\altaffilmark{f},
28: C.~Baixeras\altaffilmark{g},
29: J.~A.~Barrio\altaffilmark{e},
30: H.~Bartko\altaffilmark{h},
31: D.~Bastieri\altaffilmark{i},
32: J.~K.~Becker\altaffilmark{f},
33: W.~Bednarek\altaffilmark{j},
34: K.~Berger\altaffilmark{a},
35: E.~Bernardini\altaffilmark{k},
36: C.~Bigongiari\altaffilmark{i},
37: A.~Biland\altaffilmark{c},
38: R.~K.~Bock\altaffilmark{h,}\altaffilmark{i},
39: G.~Bonnoli\altaffilmark{l},
40: P.~Bordas\altaffilmark{m},
41: V.~Bosch-Ramon\altaffilmark{m},
42: T.~Bretz\altaffilmark{a},
43: I.~Britvitch\altaffilmark{c},
44: M.~Camara\altaffilmark{e},
45: E.~Carmona\altaffilmark{h},
46: A.~Chilingarian\altaffilmark{n},
47: S.~Commichau\altaffilmark{c},
48: J.~L.~Contreras\altaffilmark{e},
49: J.~Cortina\altaffilmark{b},
50: M.~T.~Costado\altaffilmark{o,}\altaffilmark{p},
51: S.~Covino\altaffilmark{d},
52: V.~Curtef\altaffilmark{f},
53: F.~Dazzi\altaffilmark{i},
54: A.~De Angelis\altaffilmark{q},
55: E.~De Cea del Pozo\altaffilmark{r},
56: R.~de los Reyes\altaffilmark{e},
57: B.~De Lotto\altaffilmark{q},
58: M.~De Maria\altaffilmark{q},
59: F.~De Sabata\altaffilmark{q},
60: C.~Delgado Mendez\altaffilmark{o},
61: A.~Dominguez\altaffilmark{s},
62: D.~Dorner\altaffilmark{a},
63: M.~Doro\altaffilmark{i},
64: M.~Errando\altaffilmark{b},
65: M.~Fagiolini\altaffilmark{l},
66: D.~Ferenc\altaffilmark{t},
67: E.~Fern\'andez\altaffilmark{b},
68: R.~Firpo\altaffilmark{b},
69: M.~V.~Fonseca\altaffilmark{e},
70: L.~Font\altaffilmark{g},
71: N.~Galante\altaffilmark{h},
72: R.~J.~Garc\'{\i}a L\'opez\altaffilmark{o,}\altaffilmark{p},
73: M.~Garczarczyk\altaffilmark{h},
74: M.~Gaug\altaffilmark{o},
75: F.~Goebel\altaffilmark{h},
76: M.~Hayashida\altaffilmark{h},
77: A.~Herrero\altaffilmark{o,}\altaffilmark{p},
78: D.~H\"ohne\altaffilmark{a},
79: J.~Hose\altaffilmark{h},
80: C.~C.~Hsu\altaffilmark{h},
81: S.~Huber\altaffilmark{a},
82: T.~Jogler\altaffilmark{h,}{*},
83: D.~Kranich\altaffilmark{c},
84: A.~La Barbera\altaffilmark{d},
85: A.~Laille\altaffilmark{t},
86: E.~Leonardo\altaffilmark{l},
87: E.~Lindfors\altaffilmark{u},
88: S.~Lombardi\altaffilmark{i},
89: F.~Longo\altaffilmark{q},
90: M.~L\'opez\altaffilmark{i},
91: E.~Lorenz\altaffilmark{c,}\altaffilmark{h},
92: P.~Majumdar\altaffilmark{h},
93: G.~Maneva\altaffilmark{v},
94: N.~Mankuzhiyil\altaffilmark{q},
95: K.~Mannheim\altaffilmark{a},
96: L.~Maraschi\altaffilmark{d},
97: M.~Mariotti\altaffilmark{i},
98: M.~Mart\'{\i}nez\altaffilmark{b},
99: D.~Mazin\altaffilmark{b},
100: M.~Meucci\altaffilmark{l},
101: M.~Meyer\altaffilmark{a},
102: J.~M.~Miranda\altaffilmark{e},
103: R.~Mirzoyan\altaffilmark{h},
104: S.~Mizobuchi\altaffilmark{h},
105: M.~Moles\altaffilmark{s},
106: A.~Moralejo\altaffilmark{b},
107: D.~Nieto\altaffilmark{e},
108: K.~Nilsson\altaffilmark{u},
109: J.~Ninkovic\altaffilmark{h},
110: N.~Otte\altaffilmark{h,}\altaffilmark{w},
111: I.~Oya\altaffilmark{e},
112: M.~Panniello\altaffilmark{o,}\altaffilmark{$\dag$},
113: R.~Paoletti\altaffilmark{l},
114: J.~M.~Paredes\altaffilmark{m},
115: M.~Pasanen\altaffilmark{u},
116: D.~Pascoli\altaffilmark{i},
117: F.~Pauss\altaffilmark{c},
118: R.~G.~Pegna\altaffilmark{l},
119: M.~A.~Perez-Torres\altaffilmark{s},
120: M.~Persic\altaffilmark{q,}\altaffilmark{x},
121: L.~Peruzzo\altaffilmark{i},
122: A.~Piccioli\altaffilmark{l},
123: F.~Prada\altaffilmark{s},
124: E.~Prandini\altaffilmark{i},
125: N.~Puchades\altaffilmark{b},
126: A.~Raymers\altaffilmark{n},
127: W.~Rhode\altaffilmark{f},
128: M.~Rib\'o\altaffilmark{m},
129: J.~Rico\altaffilmark{y,}\altaffilmark{b},
130: M.~Rissi\altaffilmark{c},
131: A.~Robert\altaffilmark{g},
132: S.~R\"ugamer\altaffilmark{a},
133: A.~Saggion\altaffilmark{i},
134: T.~Y.~Saito\altaffilmark{h},
135: M.~Salvati\altaffilmark{d},
136: M.~Sanchez-Conde\altaffilmark{s},
137: P.~Sartori\altaffilmark{i},
138: K.~Satalecka\altaffilmark{k},
139: V.~Scalzotto\altaffilmark{i},
140: V.~Scapin\altaffilmark{q},
141: R.~Schmitt\altaffilmark{z},
142: T.~Schweizer\altaffilmark{h},
143: M.~Shayduk\altaffilmark{h},
144: K.~Shinozaki\altaffilmark{h},
145: S.~N.~Shore\altaffilmark{z},
146: N.~Sidro\altaffilmark{b,}{*},
147: A.~Sierpowska-Bartosik\altaffilmark{r},
148: A.~Sillanp\"a\"a\altaffilmark{u},
149: D.~Sobczynska\altaffilmark{j},
150: F.~Spanier\altaffilmark{a},
151: A.~Stamerra\altaffilmark{l},
152: L.~S.~Stark\altaffilmark{c},
153: L.~Takalo\altaffilmark{u},
154: F.~Tavecchio\altaffilmark{d},
155: P.~Temnikov\altaffilmark{v},
156: D.~Tescaro\altaffilmark{b},
157: M.~Teshima\altaffilmark{h},
158: M.~Tluczykont\altaffilmark{k},
159: D.~F.~Torres\altaffilmark{y,}\altaffilmark{r},
160: N.~Turini\altaffilmark{l},
161: H.~Vankov\altaffilmark{v},
162: A.~Venturini\altaffilmark{i},
163: V.~Vitale\altaffilmark{q},
164: R.~M.~Wagner\altaffilmark{h},
165: W.~Wittek\altaffilmark{h},
166: V.~Zabalza\altaffilmark{m},
167: F.~Zandanel\altaffilmark{s},
168: R.~Zanin\altaffilmark{b},
169: J.~Zapatero\altaffilmark{g}
170: }
171:
172: \altaffiltext{a} {Universit\"at W\"urzburg, D-97074 W\"urzburg, Germany}
173: \altaffiltext{b} {IFAE, Edifici Cn., Campus UAB, E-08193 Bellaterra, Spain}
174: \altaffiltext{c} {ETH Zurich, CH-8093 Switzerland}
175: \altaffiltext{d} {INAF National Institute for Astrophysics, I-00136 Rome, Italy}
176: \altaffiltext{e} {Universidad Complutense, E-28040 Madrid, Spain}
177: \altaffiltext{f} {Technische Universit\"at Dortmund, D-44221 Dortmund, Germany}
178: \altaffiltext{g} {Universitat Aut\`onoma de Barcelona, E-08193 Bellaterra, Spain}
179: \altaffiltext{h} {Max-Planck-Institut f\"ur Physik, D-80805 M\"unchen, Germany}
180: \altaffiltext{i} {Universit\`a di Padova and INFN, I-35131 Padova, Italy}
181: \altaffiltext{j} {University of \L\'od\'z, PL-90236 Lodz, Poland}
182: \altaffiltext{k} {DESY Deutsches Elektr.-Synchrotron D-15738 Zeuthen}
183: \altaffiltext{l} {Universit\`a di Siena, and INFN Pisa, I-53100 Siena, Italy}
184: \altaffiltext{m} {Universitat de Barcelona (ICC/IEEC), E-08028 Barcelona, Spain}
185: \altaffiltext{n} {Yerevan Physics Institute, AM-375036 Yerevan, Armenia}
186: \altaffiltext{o} {Inst. de Astrofisica de Canarias, E-38200, La Laguna, Tenerife, Spain}
187: \altaffiltext{p} {Depto. de Astrofisica, Universidad, E-38206 La Laguna, Tenerife, Spain}
188: \altaffiltext{q} {Universit\`a di Udine, and INFN Trieste, I-33100 Udine, Italy}
189: \altaffiltext{r} {Institut de Cienci\`es de l'Espai (IEEC-CSIC), E-08193 Bellaterra, Spain}
190: \altaffiltext{s} {Inst. de Astrofisica de Andalucia (CSIC), E-18080 Granada, Spain}
191: \altaffiltext{t} {University of California, Davis, CA-95616-8677, USA}
192: \altaffiltext{u} {Tuorla Observatory, Turku University, FI-21500 Piikki\"o, Finland}
193: \altaffiltext{v} {Inst. for Nucl. Research and Nucl. Energy, BG-1784 Sofia, Bulgaria}
194: \altaffiltext{w} {Humboldt-Universit\"at zu Berlin, D-12489 Berlin, Germany}
195: \altaffiltext{x} {INAF/Osservatorio Astronomico and INFN, I-34143 Trieste, Italy}
196: \altaffiltext{y} {ICREA, E-08010 Barcelona, Spain}
197: \altaffiltext{z} {Universit\`a di Pisa, and INFN Pisa, I-56126 Pisa, Italy}
198: \altaffiltext{\dag} {deceased}
199: \altaffiltext{*} {Corresponding authors: T.~Jogler,
200: jogler@mppmu.mpg.de, N.~Sidro, nsidro@ifae.es}
201:
202: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
203: \begin{abstract}
204: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
205:
206: The MAGIC collaboration has recently reported the discovery of
207: $\gamma$-ray emission from the binary system \lsi\ in the TeV energy
208: region. Here we present new observational results on this source
209: in the energy range between $300\textrm{ GeV}$ and
210: $3\textrm{ TeV}$. In total 112 hours of data were taken between
211: September and December 2006 covering 4 orbital cycles of this
212: object. This large amount of data allowed us to produce an integral
213: flux light curve covering for the first time all orbital phases of \lsi.
214: %
215: In addition, we also obtained a differential energy spectrum for two
216: orbital phase bins covering the phase range $0.5 < \phi<0.6$ and $0.6 <
217: \phi<0.7$.
218: The photon index in the two phase bins is consistent within the errors
219: with an average index $\Gamma=2.6\pm0.2_{stat}\pm0.2_{sys}$.
220: %
221: \lsi\ was found to be variable at TeV energies on timescales of
222: days. These new MAGIC measurements allowed us to search for
223: intra-night variability of the VHE emission; however, no evidence
224: for flux variability on timescales down to 30 minutes was found.
225: %
226: To test for possible periodic structures in the light curve, we
227: apply the formalism developed by Lomb and Scargle to the \lsi\ data
228: taken in 2005 and 2006. We found the \lsi\ data set to be periodic
229: with a period of (26.8$\pm$0.2)~days (with a post-trial chance
230: probability of 10$^{-7}$), close to the orbital period.
231:
232: \end{abstract}
233:
234: \keywords{gamma rays: observations --- stars: individual (\lsi) --- X-ray: binaries}
235:
236: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
237: \section{Introduction}
238: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
239:
240: The $\gamma$-ray binary system \lsi\ is located at a distance of
241: $\sim$2 kpc and is composed of a compact object of unknown nature
242: (neutron star or black hole) orbiting a Be star in a highly
243: eccentric orbit ($e=0.72\pm0.15$ or $e=0.55\pm0.05$
244: following~\cite{Casares:2005wn} and \cite{Grundstrom:2006}
245: respectively).
246:
247: \lsi\ was found to display periodic variability in the radio,
248: infrared, optical, and X-ray bands (\citealt{1982ApJ...255..210T},
249: \citealt{Paredes:1995}, \citealt{Mendelson:1989} and
250: \citealt{Paredes:1997}, respectively).
251:
252: The orbital period of the system is 26.4960 days
253: long~\citep{Gregory:2002}. The periastron passage, derived from the
254: optical spectra, is found to be at phase $\phi=0.23\pm0.02$
255: in~\cite{Casares:2005wn} and $\phi=0.301\pm0.011$
256: in~\cite{Grundstrom:2006}, adopting a zero-phase at $T_0=\textrm{JD
257: } 2443366.775$.
258:
259: Radio outbursts are observed every orbital cycle at phases varying
260: between 0.45 and 0.95 with a 4.6 years
261: modulation~\citep{Gregory:2002}. Radio imaging techniques have shown
262: extended, radio-emitting structures with angular extensions of
263: $\sim$0.01 to $\sim$0.1 arc-sec, where the radio emission originates
264: in a two-sided, possibly precessing, relativistic jet
265: ($\beta/c=0.6$)~\citep{Massi:2004}. These extended radio structures
266: have led some authors to adopt the microquasar scenario to explain
267: the non-thermal emission in \lsi\ ~\citep[e.g.,][]{Bosch-Ramon:2006,
268: Bednarek:2006}. Recent high resolution VLBA measurements show a
269: complex and changing morphology different from what is expected for
270: a typical microquasar jet (see radio images
271: in~\citealt[e.g.,][]{2006smqw.confE..52D,magiclsi.2006MW}).
272: Furthermore no solid evidence for the presence of an accretion disk
273: (i.e. a thermal X-ray component) has been
274: observed~\citep{2006MNRAS.372.1585C}. This seems to favor a scenario
275: in which the non-thermal emission in \lsi\ is powered by the
276: interaction between a pulsar and the primary star
277: winds~\citep{Maraschi:1981}.
278:
279: At higher energies, \lsi\ was found to be spatially coincident with
280: the EGRET $\gamma$-ray source \eg~\citep{kniffen}. Variable emission at
281: TeV energies was observed with the MAGIC
282: telescope~\citep{Albert:2006vk} and was recently confirmed by
283: VERITAS~\citep{Veritas_lsi}. The system showed the peak TeV
284: $\gamma$-ray flux at phase $\phi\sim0.65$, while no very high-energy
285: emission was detected around the periastron passage.
286:
287: Here we present new MAGIC telescope observations of \lsi. We briefly
288: discuss the observational technique and the data analysis procedure,
289: investigate the very high energy (VHE) $\gamma$-ray spectrum during
290: the high emission phase of the source, and put the results into
291: perspective with previous VHE $\gamma$-ray observations of this
292: system. Finally we analyzed the temporal characteristics of the TeV
293: emission and find a periodic modulation of the signal with the
294: orbital period.
295:
296:
297: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
298: \section{Observations}
299: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
300:
301: The observations were performed from September to December 2006
302: using the MAGIC telescope on the Canary island of La Palma
303: ($28.75^\circ$N, $17.86^\circ$W, 2225~m a.s.l.), from where \lsi\ is
304: observable at zenith distances above $32^\circ$. The telescope
305: operates in the energy band from $50-60$~GeV (trigger threshold at
306: zenith angles less than 30 degrees) up to tens of TeV, with a
307: typical energy resolution of $20-30\%$. The accuracy in
308: reconstructing the direction of the incoming $\gamma$-rays is about
309: $0.1^{\circ}$, depending on the energy. A detailed description of
310: the telescope performance can be found in~\cite{crab:2008}.
311:
312: The data on \lsi\ were taken between 15$^{\textrm{th}}$ of September
313: 2006 and 28$^{\textrm{th}}$ of December 2006 covering 4 orbital
314: periods of the system. In total 120 hours of data were taken at
315: zenith angles between 32 and 55$^{\circ}$, with $\sim97$\% of the
316: data below $44^{\circ}$. After pre-selection of good quality data a
317: total of 112 hours of data remained for the analysis. About 17\% of
318: these were recorded under moderate moonlight conditions. Due to the
319: different observation conditions such as bad weather, too bright
320: moon or too large zenith angle, the data set was not uniform with
321: the orbital phase. In Table~\ref{tab:LC} the observation times of
322: the analyzed data are summarized.
323:
324: The observations were carried out in wobble
325: mode~\citep{Fomin:1994aj}, i.e. by alternately tracking two
326: positions at $0.4^\circ$ offset from the actual source position.
327: This observation mode allows for a reliable background estimate for
328: point like objects such as \lsi.
329:
330: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
331: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
332: \section{Data Analysis}
333: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
334:
335: The data analysis was carried out using the standard MAGIC analysis
336: and reconstruction software~\citep{crab:2008} . The images were
337: cleaned by requiring a minimum number of 10 photoelectrons (core
338: pixels) and 5 photoelectrons (boundary pixels), see
339: e.g.~\citet{Fegan:1997db}. The quality of the data was checked and
340: bad data such as accidental noise triggers or data taken during
341: adverse conditions (very low atmospheric transmission, car light
342: flashes etc.) were rejected. From the remaining events, image
343: parameters were calculated~\citep{Hillas_parameters}.
344:
345: For the $\gamma$/hadron separation a multidimensional classification
346: procedure based on the Random Forest method~\citep{magic:RF,
347: Bock:2004td} was used.
348: For every event a parameter called hadronness ($h$) is derived,
349: based on the values of the events image parameters. The hadronness
350: denotes the probability that an event is a hadronic induced
351: (background) event. The final separation was achieved by a cut in
352: $h$ which was determined by requiring 80\% of the simulated Monte
353: Carlo (MC) $\gamma$-ray events to be kept. In addition to the cut in
354: $h$ a geometrical cut in the squared angular distance of the assumed
355: source position to the shower direction axis ($\theta^2\textrm{
356: cut}$) was performed so that $70\%$ of all simulated MC $\gamma$-ray
357: events from a point-like source are left after the cut. The cut
358: efficiencies were determined by optimizing the significance of a
359: Crab Nebula data sample recorded under the same zenith angle as the
360: \lsi\ data set. The same cut procedure was applied to the final
361: \lsi\ sample. The energy of the primary $\gamma$-ray was
362: reconstructed from the image parameters using also a Random Forest
363: method leading to an assigned estimated energy for each
364: reconstructed $\gamma$-ray event. The differential energy spectrum
365: is unfolded taking into account the full instrumental energy
366: resolution~\citep{magic:unfolding}. For the integral flux
367: calculation of the light curves we used fixed cuts (for all
368: energies) in hadronness and $\theta^2$. In the case of the energy
369: spectrum determination we derived fixed hadronness and $\theta^2$
370: cuts for each energy bin.
371:
372: The main contributions to the systematic error of our analysis are
373: the uncertainties in the atmospheric transmission, the reflectivity
374: (including stray-light losses) of the mirror and the light catchers,
375: the photon to photoelectron conversion calibration and the
376: photoelectron collection efficiency in the photomultiplier
377: front-end~\citep{crab:2008}. Also MC uncertainties in the detector
378: simulation and systematic uncertainties from the analysis methods
379: contribute significantly to the overall error.\\
380: All errors in this paper are statistical errors, otherwise it is
381: stated explicitly. In addition there is a $30\%$ systematic
382: uncertainty on flux levels and $0.2$ on the spectral photon index.
383:
384: MAGIC has the capability to operate under moderate moonlight. This
385: permits to increase the duty cycle by up to 28\%, thus considerably
386: improving the sampling of transient sources. In particular, 17\% of
387: the data used in this analysis were recorded under moonlight. The
388: nights which were partly taken under moonlight conditions are
389: labeled with a star in Table\ref{tab:LC}. For these days we estimate
390: an increased systematic error of $\sim 40\%$ instead of the $\sim
391: 30\%$ in the case of the dark night observations. All spectra are
392: derived from data which is only taken under dark night conditions
393: and thus no additional error is present in the obtained parameters.
394:
395:
396: \subsection{Light Curve}
397: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
398:
399: Figure~\ref{fig:lc} presents the gamma-ray flux above 400~GeV
400: measured from the direction of \lsi\ as a function of the orbital
401: phase of the system for the 4 observed orbital cycles. The
402: probability for the distribution of measured fluxes to be a
403: statistical fluctuation of a constant flux (obtained from a $\chi^2$
404: fit to the entire data sample) is $4.4\times10^{-6}$
405: ($\chi^2/\textrm{dof} = 108.9/51$). In all orbital cycles
406: significant detections ($S > 2\sigma$) occurred during the orbital
407: phase bin 0.6--0.7. The highest measured fluxes are dominantly found
408: in this phase bin. Among those nights around phase 0.65, the night
409: MJD 54035.11 shows
410: the maximum flux, with statistical significance of 4.5~$\sigma$.\\
411:
412: At the periastron passage (phase 0.23, according
413: to~\citealt{Casares:2005wn}) the flux level is always below the
414: MAGIC sensitivity and we derive an upper limit with $95\%$
415: confidence level of $4 \times 10^{-12}$~cm$^{-2}$s$^{-1}$ (MJD
416: 53997). If we take for the periastron passage the phase value 0.3 as
417: obtained by~\cite{Grundstrom:2006}, we detect a marginal signal on
418: MJD 53999 with a flux of $F(\textrm{E}>400 \textrm{ GeV})=5.3\pm2.4
419: \times 10^{-12} \textrm{cm}^{-2}\textrm{s}^{-1}$. Since the correct
420: value for the periastron passage is yet debated we can put strong
421: constrains to the emission only in the case of phase 0.23.
422:
423:
424: \begin{figure}[tbp]
425: \centering
426: \includegraphics[width=\linewidth]{f1}
427: \caption{VHE ($E>400\text{GeV}$) gamma-ray flux of \lsi\ as a
428: function of the orbital phase for the four observed orbital cycles
429: (4 upper panels) and averaged for the entire observation time
430: (lowermost panel). In the lower most panel the previous published
431: \citep{Albert:2006vk} averaged fluxes per phasebin
432: are shown in red too. Vertical error bars include $1\sigma$ statistical
433: error.}
434: \label{fig:lc}
435: \end{figure}
436:
437: \begin{table}[tbp]
438: \caption{Observation time, orbital phase, integral flux (above
439: 400~GeV), flux upper limit at the 95\% confidence level (given in
440: case flux significance is $\lesssim 2 \sigma$,
441: ~\citep[following]{Rolke:2004mj}). Nights partly taken under moonlight conditions
442: are labeled with a star.}
443: \vspace*{-0.3cm}
444: \begin{center}
445: \begin{tabular}{lcccc}
446: \hline \hline
447: Middle Time & Obs. Time & Phase & Flux & Upper limit \\
448: (MJD) & (min) & & 10$^{-12}$ & 10$^{-12}$ \\
449: & & & (cm$^{-2}$ s$^{-1}$)& (cm$^{-2}$ s$^{-1}$) \\
450: \hline
451: 53993.18$^{*}$ & 137 & 0.08 & 3.7 $\pm$ 2.3 & 8.5\\
452: 53994.17$^{*}$ & 112 & 0.11 & 0.6 $\pm$ 2.7 & 6.2\\
453: 53995.17$^{*}$ & 157 & 0.15 & $-$2.0 $\pm$ 2.2 & 3.0\\
454: 53997.15 & 229 & 0.23 & 0.3 $\pm$ 1.8 & 4.0\\
455: 53998.15 & 211 & 0.26 & 2.0 $\pm$ 2.0 & 6.0\\
456: 53999.10 & 133 & 0.30 & 5.3 $\pm$ 2.4 & \nodata\\
457: 54001.12 & 82 & 0.38 & $-$3.6 $\pm$ 3.8 & 5.1\\
458: 54002.09 & 188 & 0.41 & 2.4 $\pm$ 2.3 & 7.1\\
459: 54003.08 & 144 & 0.45 & 1.8 $\pm$ 2.7 & 7.2\\
460: 54004.08 & 158 & 0.49 & $-$4.0 $\pm$ 2.5 & 2.5\\
461: 54005.07 & 155 & 0.52 & 3.0 $\pm$ 2.5 & 8.1\\
462: 54006.07 & 162 & 0.56 & 1.8 $\pm$ 2.7 & 7.2\\
463: 54007.08 & 139 & 0.60 & 4.4 $\pm$ 2.8 & 10.2\\
464: 54008.07 & 152 & 0.64 & 8.8 $\pm$ 3.1 & \nodata\\
465: 54009.08 & 147 & 0.68 & 4.4 $\pm$ 2.6 & 9.7\\
466: 54013.24 & 7 & 0.83 & 0.8 $\pm$ 10.7 & 26.7\\
467: \hline
468: 54022.10$^{*}$ & 186 & 0.17 & 1.7 $\pm$ 2.0 & 5.8\\
469: 54023.10$^{*}$ & 269 & 0.20 & $-$2.9 $\pm$ 1.5 & 1.4\\
470: 54024.08$^{*}$ & 20 & 0.24 & $-$0.4 $\pm$ 7.0 & 15.2\\
471: 54029.02 & 134 & 0.43 & $-$1.1 $\pm$ 2.5 & 4.1\\
472: 54030.01 & 161 & 0.47 & $-$0.4 $\pm$ 2.3 & 4.2\\
473: 54031.01 & 163 & 0.50 & 5.9 $\pm$ 2.6 & \nodata\\
474: 54032.01 & 139 & 0.54 & 3.4 $\pm$ 2.9 & 9.2\\
475: 54035.11 & 150 & 0.66 & 12.7 $\pm$ 2.9 & \nodata\\
476: 54039.09 & 93 & 0.81 & $-$1.4 $\pm$ 1.2 & 1.7\\
477: \hline
478: 54055.97 & 181 & 0.45 & 4.0 $\pm$ 2.2 & 8.5\\
479: 54056.96 & 223 & 0.48 & $-$0.2 $\pm$ 2.1 & 4.2\\
480: 54057.90 & 66 & 0.52 & 3.3 $\pm$ 3.8 & 11.2\\
481: 54058.90 & 57 & 0.56 & 2.3 $\pm$ 3.3 & 9.2\\
482: 54060.00 & 17 & 0.60 & 16.5 $\pm$ 6.8 & \nodata\\
483: 54061.96 & 221 & 0.67 & 5.9 $\pm$ 2.2 & \nodata\\
484: 54062.96 & 228 & 0.71 & 5.5 $\pm$ 2.1 & \nodata\\
485: 54063.95 & 56 & 0.75 & 3.6 $\pm$ 4.0 & 12.1\\
486: 54065.00$^{*}$ & 71 & 0.79 & 4.5 $\pm$ 3.8 & 12.4\\
487: 54066.02$^{*}$ & 185 & 0.82 & 1.1 $\pm$ 2.4 & 5.9\\
488: 54067.04$^{*}$ & 188 & 0.86 & 0.3 $\pm$ 2.3 & 5.0\\
489: 54068.08$^{*}$ & 77 & 0.90 & $-$1.5 $\pm$ 3.6 & 6.1\\
490: \hline
491: 54081.89 & 17 & 0.42 & $-$0.3 $\pm$ 5.4 & 12.6\\
492: 54082.85 & 77 & 0.46 & 2.9 $\pm$ 3.7 & 10.6\\
493: 54083.88 & 31 & 0.50 & 4.4 $\pm$ 5.3 & 15.9\\
494: 54084.85 & 63 & 0.54 & 1.5 $\pm$ 4.5 & 10.8\\
495: 54085.95 & 111 & 0.58 & 2.4 $\pm$ 1.4 & 5.5\\
496: 54086.95 & 282 & 0.61 & 8.6 $\pm$ 1.8 & \nodata\\
497: 54088.01 & 82 & 0.65 & 9.7 $\pm$ 3.6 & \nodata\\
498: 54088.95 & 83 & 0.69 & 3.4 $\pm$ 2.9 & 9.4\\
499: 54089.89 & 29 & 0.73 & 0.4 $\pm$ 3.7 & 9.0\\
500: 54090.88 & 176 & 0.76 & 3.6 $\pm$ 2.2 & 8.1\\
501: 54091.90 & 140 & 0.80 & 1.9 $\pm$ 2.8 & 7.6\\
502: 54092.92 & 92 & 0.84 & 15.6 $\pm$ 3.8 & \nodata\\
503: 54093.97 & 92 & 0.88 & 7.0 $\pm$ 3.5 & \nodata\\
504: 54095.01$^{*}$ & 57 & 0.92 & 1.1 $\pm$ 1.1 & 4.1\\
505: 54096.02$^{*}$ & 49 & 0.96 & 3.6 $\pm$ 4.4 & 12.8\\
506: \hline
507: \end{tabular}
508: \end{center}
509: \label{tab:LC}
510: \end{table}
511:
512: Summing up all data between phase 0.6 and 0.7, where the maximum
513: flux level is observed, we determine an integral flux above 400 GeV
514: of
515: %
516: \begin{equation*}
517: F(E>400 \mathrm{GeV}) = (7.9 \pm 0.9_\text{stat} \pm
518: 2.4_\text{syst}) \times 10^{-12} \mathrm{cm}^{-2} \mathrm{s}^{-1}.
519: \end{equation*}
520:
521: The above quoted flux corresponds to $7\%$ of the integral Crab
522: nebula flux in the same energy range. The mean flux for all other
523: phase bins can be found in Table~\ref{tab:LCmean}.
524: This is well in agreement with the flux measured by MAGIC in the first
525: campaign~\citep{Albert:2006vk}. The data we presented here have been
526: reanalized with an improved energy estimation.
527:
528: \begin{table}[tbp]
529: \caption{Average flux level above 400~GeV for each orbital phase
530: bin. Flux upper limit at the 95\% confidence level are quoted in
531: case flux significance is $\lesssim 2 \sigma$
532: (following~\cite{Rolke:2004mj}). }
533: \vspace*{-0.3cm}
534: \begin{center}
535: \begin{tabular}{ccc}
536: \hline \hline
537: Phase bin & Flux & Upper Limit \\
538: & (10$^{-12}$cm$^{-2}$ s$^{-1}$)& (10$^{-12}$cm$^{-2}$ s$^{-1}$)\\
539: \hline
540: 0.0--0.1 & 3.7 $\pm$ 2.3 & 8.5 \\
541: 0.1--0.2 & 0.2 $\pm$ 1.2 & 2.7 \\
542: 0.2--0.3 & 0.3 $\pm$ 0.9 & 2.2 \\
543: 0.3--0.4 & -1.2 $\pm$ 2.8 & 4.3 \\
544: 0.4--0.5 & 0.7 $\pm$ 0.8 & 2.4 \\
545: 0.5--0.6 & 3.1 $\pm$ 1.0 & \nodata \\
546: 0.6--0.7 & 7.9 $\pm$ 0.9 & \nodata \\
547: 0.7--0.8 & 4.3 $\pm$ 1.2 & \nodata \\
548: 0.8--0.9 & 2.8 $\pm$ 1.1 & \nodata \\
549: 0.9--1.0 & 0.7 $\pm$ 2.0 & 4.8 \\
550: \hline
551: \end{tabular}
552: \end{center}
553: \label{tab:LCmean}
554: \end{table}
555:
556: One very interesting peculiarity found in the light curve is that a
557: second peak in the flux level is seen in the last observed period at
558: phase 0.84. The flux is $(16\pm4) \times 10^{-12}$~cm$^{-2}$s$^{-1}$
559: which is at a similar level compared to the maximum flux detected in
560: phase $\sim$0.65 (MJD 54035.11). This high flux was not seen in
561: similar phases in any previous cycle, where only upper limits could
562: be set (see Table~\ref{tab:LC}). Six hours after our measurement,
563: the data of the Swift X-ray satellite showed a high flux
564: ($0.25\pm0.01 \textrm{ counts s}^{-1}$)at phase
565: 0.85~\citep{2007A&A...474..575E}. These Swift observations did not
566: cover the same orbital phase in any other orbit. Beside this second
567: peak the main X-ray emission peak is found between the phases
568: 0.5--0.8 ($\sim 0.24\textrm{ counts s}^{-1}$
569: ~\citealt{2007A&A...474..575E}) in exactly the same phase bin where
570: \lsi\ is detected by MAGIC. This is a hint for a correlated
571: X-ray/TeV emission.
572:
573: Our measurement is in agreement with the published VERITAS
574: measurements~\citep{Veritas_lsi}, that \lsi\ is detected at TeV
575: Energies in the phase range 0.5--0.8. The MAGIC and VERITAS data are
576: not strictly simultaneous taken and VERITAS did not observe \lsi\ in
577: December~2006 were the second peak occurred.\\
578: Due to our long observation time and dense sampling of orbital
579: phases we obtained the currently most detailed light curve of \lsi .
580:
581:
582: \subsection{Spectral studies}
583: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
584:
585: As seen from the light curve (Fig.~\ref{fig:lc}) \lsi\ is a very
586: variable source which shows high flux levels only at some orbital
587: phases. For the phases $0.5 < \phi < 0.6$ and $0.6 < \phi < 0.7$,
588: where we measured significantly high flux levels we were able to
589: determine differential energy spectra. In both cases the obtained
590: energy spectra are compatible with pure power laws. In the case of
591: the phase bin $0.6 < \phi < 0.7$ a power law fit gives:
592: \begin{equation*}
593: \frac{\mathrm{d}F}{\mathrm{d}E} =
594: \frac{(2.6\pm0.3_\text{stat}\pm0.8_\text{syst}) \cdot 10^{-12}}
595: {\mathrm{TeV}\,\mathrm{cm}^2\,\mathrm{s}} \left(\frac{E}{1\,
596: \mathrm{TeV}}\right)^{-2.6\pm0.2_\text{stat}\pm0.2_\text{syst}} \, ,
597: \end{equation*}
598: with a reduced $\chi^2/\textrm{dof} = 5.22/5$. The spectral fit
599: parameters agree excellent with the previous reported ones by
600: MAGIC~\citep{Albert:2006vk}.
601:
602: In addition we derived the differential energy spectra for the two
603: nights with a signal of~$ > 4.5\sigma$ significance, which are part
604: of the same phase bin $0.6 < \phi < 0.7$. Both spectra are also well
605: described by a pure power law (see table~\ref{tab:spectra}). No
606: evidence for spectral variations has been found.
607:
608: In case of phase bin $0.5 < \phi < 0.6$ we obtained:
609: \begin{equation*}
610: \frac{\mathrm{d}F}{\mathrm{d}E} =
611: \frac{(1.2\pm0.4_\text{stat}\pm0.3_\text{syst}) \cdot 10^{-12}}
612: {\mathrm{TeV}\,\mathrm{cm}^2\,\mathrm{s}} \left(\frac{E}{1\,
613: \mathrm{TeV}}\right)^{-2.7\pm0.4_\text{stat}\pm0.2_\text{syst}} \, ,
614: \end{equation*}
615: with a reduced $\chi^2/\textrm{dof}=1.42/4$, showing that the
616: spectral shape is well compatible with a simple power law.
617:
618: The energy spectra of the two phase bins together with the power law
619: fits are shown in Fig.~\ref{fig:spec05}. The corresponding fit
620: parameters and their errors are also shown in
621: Table~\ref{tab:spectra}.
622:
623: \begin{table}[tbp]
624: \caption{Spectral fitting parameters} \vspace*{-0.3cm}
625: \begin{center}
626: \begin{tabular}{ccc}
627: \hline \hline
628: Phase / MJD & Flux & Spectral Photon Index \\
629: & (10$^{-12} $TeV$^{-1}$cm$^{-2}$ s$^{-1}$)& \\
630: \hline
631: $0.6-0.7$ & $2.6\pm0.3$ & $2.6\pm0.2$\\
632: $0.5-0.6$ & $1.2 \pm 0.4$ &$2.7 \pm 0.4$\\
633: \hline
634: 54035 &$3.6\pm1.1$ & $2.7\pm0.5$\\
635: 54086 &$3.2\pm0.6$ & $2.6\pm0.3$\\
636: \hline
637: \end{tabular}
638: \end{center}
639: \label{tab:spectra}
640: \end{table}
641:
642: \begin{figure}[tbp]
643: \centering
644: \includegraphics[width=\linewidth]{f2}
645: \caption{Shown is the spectrum of phase bin $0.5 < \phi < 0.6$
646: (green), phase bin $0.6 < \phi < 0.7$ (red) and the spectrum
647: obtained from previous MAGIC measurements (blue
648: dashed)~\cite{Albert:2006vk}. The spectra from phase bin $0.6 < \phi
649: < 0.7$ and the previous MAGIC measurements can be well described by
650: a simple power law with photon index $2.6\pm0.2$. The
651: spectral slope of phase bin $0.5 < \phi < 0.6$ is compatible with these
652: results within the errors.}
653: \label{fig:spec05}
654: \end{figure}
655:
656: The spectral indices of the fitted power laws, for both phase bins
657: and the single night spectra, are compatible within their errors
658: indicating that no significant spectral changes happened between the
659: different phase bins and between the different orbital cycles. In
660: the phase bins $0.0 < \phi < 0.5$ and $0.7 < \phi < 1.0$ the
661: $\gamma$-ray flux is too low to derive meaningful differential
662: energy spectra.
663:
664: Another possibility to search for spectral variation is by means of
665: the hardness ratio $HR$ which we define as the ratio of the integral
666: flux between 400~GeV and 900~GeV and above 900~GeV. The $HR$
667: plotted against the total integral flux above 400~GeV for each night
668: with a signal above $2\sigma$ significance is shown in
669: Fig.~\ref{fig:HR}. The requirement of the $2\sigma$~significance of
670: the signal is to minimize systematic effects on the calculation of
671: the correlation coefficient. We do not find any clear correlation
672: between the $HR$ and the flux level. Thus we do not find any change
673: in the spectral behavior in nights where \lsi\ is detected at modest
674: significance.
675:
676: \begin{figure}[tbp]
677: \centering
678: \includegraphics[width=\linewidth]{f3}
679: \caption{The $HR$, defined as $F(E>900\textrm{GeV}) /F(400<E<900\textrm{ GeV})$, vs. the integral flux
680: $F(E>400\textrm{ GeV})$. Each point
681: is one night with a reconstructed signal of at least
682: $2\sigma$~significance. There is no clear correlation
683: between the $HR$ and the flux level.}
684: \label{fig:HR}
685: \end{figure}
686:
687: The spectral studies on \lsi\ exhibits that the spectrum is soft
688: (compared to other galactic sources) during all phases in which
689: \lsi\ is detected at TeV energies. While the flux level changes on
690: timescales of days and reaches a maximum flux (detected above
691: $3\sigma$) of $15.6\times 10^{-12}$~cm$^{-2}$s$^{-1}$ the source
692: shows a spectral photon index of $2.6\pm0.2$, compatible with being
693: constant.
694:
695: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
696: \section{Timing analysis}
697: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
698:
699: \subsection{Search for intra-night variability}
700: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
701:
702: \lsi\ was found to be variable on timescales of days in the previous
703: observational campaign by MAGIC~\citep{Albert:2006vk}. A still open
704: question is whether \lsi\ also shows variability on shorter time
705: scales. We investigated the data for all nights with a significant
706: flux level ($F(E>400 \mathrm{GeV})>4\times
707: 10^{-12}$~cm$^{-2}$s$^{-1}$) with respect to intra-night variability
708: on time scales ranging from 30 to 75 minutes, in steps of 15
709: minutes. This yields 16 suitable nights. Among those, the longest
710: one was MJD 54086.95 (phase 0.61). Its intranight light curve is
711: shown in Fig.~\ref{fig:intranight_lc}, where each bin has
712: $\sim1$~hour width. The light curve is fitted with a constant flux
713: level with probability of 90\% ($\chi^2/\textrm{dof}=1.08/4$). For
714: the rest of the nights and tested time scales, the post-trial
715: probabilities of being chance fluctuations of a constant flux are
716: all above $32\%$.
717:
718: We conclude that the VHE fluxes are constant on timescales of $30\,
719: -\,75$ minutes within the MAGIC sensitivity.
720:
721: \begin{figure}[tbp]
722: \centering
723: \includegraphics[width=\linewidth]{f4}
724: \caption{Intra-night integral flux behavior for the longest observed
725: night 18$^{\textrm{th}}$ December 2006,
726: MJD 54086.95 and orbital phase 0.61. The
727: light curve is fitted to a constant flux level with a probability of
728: 90\% ($\chi^2/\textrm{dof}=1.08/4$). }
729: \label{fig:intranight_lc}
730: \end{figure}
731:
732:
733: \subsection{Search for periodicity}
734: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
735:
736: The emission of the binary system changes periodically in radio,
737: optical and X-rays, and the modulation is associated with the orbital
738: period of the binary system. At higher energies, EGRET measurements
739: showed hints for variable $\gamma$-ray emission~\citep{tavani:1998},
740: although no periodicity could be established with these data. One of
741: the aims of the MAGIC long observational campaign on \lsi\ was to
742: search for periodic VHE $\gamma$-ray emission.
743:
744: In order to maximize the sensitivity and accuracy of the timing
745: analysis we used the data presented in this work together with the
746: data taken in the first campaign~\citep{Albert:2006vk}, with
747: observation time of 54 h and covering 6 orbital cycles.
748:
749: The periodicity analysis was carried out using the formalism
750: developed by Lomb and Scargle~\citep{Lomb:1976, Scargle:1982}.
751: %
752: This formalism allows to analyze unevenly sampled data while still
753: keeping the simple exponential probability density function (PDF,
754: $P(z>z_{0}) = e^{-z_{0}}$)
755: for Gaussian White Noise (GWN) as valid for the classical Fourier
756: analysis of evenly sampled data. A remaining problem caused by
757: the uneven data sampling is that the independent Fourier spacing is
758: broken, i.e. even a single frequency component can result in a complex
759: power spectrum with a large number of aliasing peaks.
760: An other problem is that the mean and variance values that enter the
761: \lomb\ periodogram have to be estimated from the data themselves.
762:
763: A practical method to determine the chance probabilities is the
764: following (see e.g.~\cite{Frescura:2007}):
765: \begin{enumerate}
766: \item A large number of random data series is constructed with a
767: Monte Carlo simulation of random fluxes while keeping the sampling
768: times fixed.
769: \item For each random series, we construct a periodogram, sampling it
770: for a pre-selected group of frequencies.
771: \item For each frequency, we compare the periodogram derived from the
772: real data set with the probability density function (PDF) obtained
773: from the simulated random series, in order to empirically determine
774: the (pre-trial) chance probability.
775: \item The overall (post-trial) chance probability is computed according
776: to the following generalization: for each simulated data series we
777: inspect the corresponding periodogram, identify the highest Fourier
778: power that occurs at any of the pre-selected frequencies, and use
779: this value to construct the post-trial PDF.
780: It should be noted that this constructed PDF is based on the null
781: hypothesis of GWN.
782: \end{enumerate}
783: Integration of the post-trial PDF gives the complementary Cumulative
784: Distribution Function (cCDF) which is used to determine the
785: (post-trial) chance probability for a given Fourier power value.
786:
787: In Fig.~\ref{fig:LS_density_function} we show the empirical
788: post-trial cCDF of the \lomb\ power, estimated via Monte Carlo
789: simulation of random fluxes. The expected cCDF above a spectral peak
790: $z_0$ is $F(z>z_0) = 1 - (1- e^{-z_0} )^M$, where $M$ is the number
791: of independent frequencies. By fitting the PDF for \lsi, we obtain a
792: probability of 75\% ($\chi^2/\textrm{dof}=263.9/279$) and a number of
793: independent frequencies of $M=550.8\pm0.6$. This result is used to
794: estimate the chance probability of the \lomb\ powers.
795:
796: \begin{figure}[tbp]
797: \centering
798: \includegraphics[width=\linewidth]{f5}
799: \caption{Post-trial complementary Cumulative Distribution Function for the
800: \lomb\ power derived from $10^{6}$ random time series.
801: The expected cCDF is also indicated (solid line). This function is
802: used to estimate the chance probability for powers above 20.}
803: \label{fig:LS_density_function}
804: \end{figure}
805:
806: In Fig.~\ref{fig:LS_periodogram} (middle panel) we show the
807: \lomb\ periodogram for an almost (up to detector related effects)
808: independent background sample, obtained simultaneously with the \lsi\
809: data (see below for the time intervals).
810: The highest obtained power is 7.5, which yields a probability of
811: $0.3$. Thus we obtain no significant probability peaks for any of the
812: scanned frequencies.
813:
814: \begin{figure}[tbp]
815: \centering
816: \includegraphics[width=\linewidth]{f6}
817: \caption{\lomb\ periodogram over the combined
818: 2005 and 2006 campaigns of \lsi\ data (upper panel) and simultaneous
819: background data (middle panel).
820: In the lower panel we show the periodograms after subtraction of a
821: sinusoidal signal (see Fig.~\ref{fig:sinusfit}) at the orbital
822: period (yellow line) and a sinusoidal plus a Gaussian wave form
823: (blue line).
824: Vertical dashed line corresponds to the orbital frequency.
825: Inset: zoom around the highest peak, which corresponds to the
826: orbital frequency ($0.0377$d$^{-1}$).
827: Its post-trial probability is nearly $10^{-7}$ (see
828: Fig.~\ref{fig:LS_density_function}). The IFS is also shown.}
829: \label{fig:LS_periodogram}
830: \end{figure}
831:
832: We apply the \lomb\ test to the \lsi\ data and obtain the periodogram
833: shown in Fig.~\ref{fig:LS_periodogram} (upper panel).
834: The periodogram is performed with the \lsi\ integral flux above 400
835: GeV, measured in a time interval $[t_i-\frac{\Delta t}{2},
836: t_i+\frac{\Delta t}{2}]$, for $\Delta t=15$~minutes and $i=0,\dots 717$
837: data points. The overall time range is $442$ days, which yields an
838: independent Fourier spacing (IFS) of $\nu_{IFS}=1/T=0.0023$~d$^{-1}$.
839: We scanned the frequency range $0.0023$-$0.25$~d$^{-1}$ with a
840: an oversampling factor of 5.
841:
842: A maximum peak in the \lomb\ periodogram is clearly seen
843: at frequency $\nu=0.0373$d$^{-1}$, for which we obtain a power of
844: $\sim22$, corresponding to a post-trial chance probability of
845: $2\times10^{-7}$.
846:
847: Several less prominent but significant peaks are also detected for other
848: frequencies (e.g. 0.041 d$^{-1}$ with probability $\leq 10^{-5}$).
849: Those peaks are related to the signal, since they are not
850: present in the contemporaneous background sample
851: (Fig.~\ref{fig:LS_periodogram}, middle panel). These are aliasing
852: peaks of the orbital period of \lsi\ caused by the various gaps in the
853: data set.
854:
855: The observational bias due to the moon cycle cannot be the responsible
856: for the observed peak since this period should otherwise be also
857: present in the background periodogram.
858:
859: \begin{figure}[tbp]
860: \centering
861: \includegraphics[width=\linewidth]{f7}
862: \caption{\lsi\ $\gamma$-ray flux above 400~GeV
863: obtained from the first and second campaigns, folded
864: with the orbital frequency in bins of 0.05 in phase.
865: The black curve is a fit to a sinusoidal
866: signal. We also fitted a sinusoidal signal plus a Gaussian component
867: (blue dotted line), which adjusts better to the data
868: (fit parameters are given in the inset). The vertical dashed lines mark the
869: two measurements of the periastron passage (according
870: to~\cite{Casares:2005wn} and \cite{Grundstrom:2006}).}
871: \label{fig:sinusfit}
872: \end{figure}
873:
874: The data folded with the peak frequency ($\nu=0.0377$d$^{-1}$) is
875: presented in Fig.~\ref{fig:sinusfit}, where a sinusoidal fit is
876: performed ($\chi^2/\textrm{dof}=123.6/16$). Subtracting the obtained
877: sinusoid from the data, we produce the periodogram shown in
878: Fig.~\ref{fig:LS_periodogram} (lower panel, yellow line). The peak
879: associated to the orbital frequency has been removed as expected.
880: Also the satellite peaks are reduced, but the fact that some of the
881: other peaks do not achieve a level consistent with the background
882: test indicates that the signal in the \lsi\ data is not purely
883: sinusoidal.
884:
885: To reduce these remaining powers, we fitted the data set with a more
886: complex signal. Motivated by the data shape, we fitted the data set
887: with a sinusoidal function plus a Gaussian signal contribution
888: ($\chi^2/\textrm{dof}=58.1/13$), as shown in Fig.~\ref{fig:sinusfit} (blue
889: dotted line). The corresponding periodogram subtracting this
890: function to the \lsi\ data set is given in
891: Fig.~\ref{fig:LS_periodogram} (lower panel, blue line). The orbital
892: frequency peak has been removed and some of the periodogram peaks
893: are much more reduced than in the purely sinusoidal subtraction,
894: giving a better agreement with the background periodogram level.
895:
896: We performed a Monte Carlo simulation to evaluate the error in the
897: frequency estimation without any signal shape assumption: we
898: simulate light curves where the number of $\gamma$-ray and background
899: candidates are selected randomly from Poisson distributions with a mean
900: equal to the actually measured distributions of events, arriving in
901: every given time interval.
902: The periodogram is calculated for $10^3$ of those randomly generated
903: series, and the distribution of the resulting peak power frequencies
904: is fitted with a Gaussian function, yielding an error of
905: $0.0003$~d$^{-1}$. An accurate peak frequency determination is done by
906: scanning more frequencies (increasing the oversampling factor)
907: around the frequency which has maximum probability in the
908: periodogram, and is found to be (0.0373$\pm$0.0003)~d$^{-1}$,
909: corresponding to a period of (26.8$\pm$0.2)~days.
910:
911: In Fig.~\ref{fig:periods_compare} we show the period obtained with
912: MAGIC data compared to the measurements in other wavelengths. The
913: most accurate measure of the orbital period is
914: (26.4960$\pm$0.0028)~days, reported in radio by~\cite{Gregory:2002}.
915: We also show period measurements reported in near IR and optical
916: V-band~\citep{Paredes:1994}, optical
917: wavelengths~\citep{Mendelson:1989}, photometry in the I-B and I
918: bands~\citep{Mendelson:1994}, H$\alpha$
919: measurements~\citep{Zamanov:1999}, and soft X-ray measurements
920: from~\cite{Wen:2006} and \cite{Paredes:1997}. The period obtained
921: with MAGIC data is compatible with the orbital radio measurement
922: within 1.5$\sigma$.
923:
924: \begin{figure}[tbp]
925: \centering
926: \includegraphics[width=\linewidth]{f8}
927: \caption{\lsi\ period measurements in different wavelengths. Blue
928: band indicates a 3$\sigma$ region around the radio measurement. The
929: $\gamma$-ray period (in red) is compatible within 1.5$\sigma$ with
930: it.}
931: \label{fig:periods_compare}
932: \end{figure}
933:
934:
935: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
936: \section{Conclusion}
937: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
938: We find that \lsi\ is a periodic $\gamma$-ray binary with an orbital
939: period of 26.8$\pm$0.2~days (and chance probability $\sim 10^{-7}$),
940: compatible with the optical, radio and X-ray period. This result
941: implies that the flux modulation is tied to the orbital period. The
942: high state in VHE $\gamma$-rays occurs in the same phases as the
943: X-ray high state. This is especially interesting since we found a
944: additional hint for X-ray/$\gamma$-ray variability correlation in
945: the orbital phase 0.85. A strictly simultaneous multi-wavelength
946: campaign is needed to investigate this correlation in more detail.
947:
948: We looked for possible intranight variability and found the flux
949: consistent with being constant within errors in 30-75 minutes
950: time-scales.
951:
952: We produce energy spectra for two phase bins $0.5 < \phi < 0.6$ and
953: $0.6 < \phi < 0.7$ and averaged flux values for several phase bins.
954: There is clear evidence for a significant change in the VHE
955: $\gamma$-ray flux level between different phase bins of \lsi. The
956: spectral photon index does not show this dependence on the phase.
957: All derived spectral photon indices are compatible with $2.6\pm0.2$
958: , obtained from the most significant phase bin of \lsi.
959:
960: We can put constraints to the emission at the periastron passage and
961: conclude that the system is detected in $\gamma$-rays only in the
962: phases $0.5-0.9$. Since significant emission is only detected in an
963: orbital sector off the phases at which the maximum gamma ray flux
964: should occur under photon-photon absorption (see fig. 5
965: in~\citealt{2006A&A...451....9D}), the latter can hardly be the only
966: source of variability in the emission.
967:
968: Thorough multiwavelength observations will allow us to probe the
969: intrinsic variability of the non-thermal emission from \lsi\ along
970: the orbit and can proof possible correlations between the X-ray and
971: TeV energy bands. This is a necessary step for understanding the
972: source nature, and the physics underlying the VHE radiation.
973:
974:
975: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
976: \section*{Acknowledgements}
977: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
978:
979: We thank the IAC for the excellent working conditions at the ORM.
980: The support of the German BMBF, MPG and the YIP of the Helmholtz
981: Gemeinschaft, the Italian INFN, the Spanish MEC, the ETH Research
982: Grant TH 34/04 3 and the Polish MNiI Grant 1P03D01028 is gratefully
983: acknowledged.
984:
985:
986: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
987: %% bibliography
988: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
989:
990: \bibliographystyle{astron}
991: \bibliography{biblio}
992:
993:
994:
995: \end{document}
996: