The Optical Gravitational Lensing Experiment. The OGLE-III Catalog of Variable Stars. X. Enigmatic Class of Double Periodic Variables in the Large Magellanic Cloud
ABSTRACT The tenth part of the OGLE-III Catalog of Variable Stars contains 125 Double Periodic Variables (DPVs) from the Large Magellanic Cloud. DPVs are semi-detached binaries which show additional variability with a period around 33 times longer than the orbital period. The cause of this long cycle is not known and previous studies suggest it involves circumbinary matter. We discuss the properties of the whole sample of the LMC DPVs and put more attention to particularly interesting objects which may be crucial for verifying hypothesis explaining long cycle variability. Secondary eclipses of one of the objects disappear during some orbital cycles and primary eclipses are deeper during long cycle minimum. Comment: 18 pages, 11 figures
arXiv:1009.5376v2 [astro-ph.SR] 26 Oct 2010
Vol. 60 (2010) pp. 179–196
The Optical Gravitational Lensing Experiment.
The OGLE-III Catalog of Variable Stars.
X. Enigmatic Class of Double Periodic Variables
in the Large Magellanic Cloud∗
R. Poleski1, I. Soszy ´ nski1, A. Udalski1, M.K. Szyma ´ nski1,
M.Kubiak1, G.Pietrzy´ nski1,2, Ł.Wyrzykowski3and K.Ulaczyk1
1Warsaw University Observatory, Al. Ujazdowskie 4, 00-478 Warszawa, Poland
2Universidad de Concepción, Departamento de Fisica, Casilla 160-C, Concepción, Chile
3Institute of Astronomy, University of Cambridge, Madingley Road, Cambridge CB3
Received September 15, 2010
The tenth part of the OGLE-III Catalog of Variable Stars contains 125 Double Periodic Variables
(DPVs) from the Large Magellanic Cloud. DPVs are semi-detached binaries which show additional
variability with a period around 33 times longer than the orbital period. The cause of this long cycle
is not known and previous studies suggest it involves circumbinary matter.
We discuss the properties of the whole sample of the LMC DPVs and put more attention to
particularly interesting objects which may be crucial for verifying hypothesis explaining long cycle
variability. Secondary eclipses of one of the objects disappear during some orbital cycles and primary
eclipses are deeper during long cycle minimum.
Key words: Catalogs – binaries: close – Magellanic Clouds
Double Periodic Variables (DPVs) were first recognized as a separate class of
variable stars by Mennickent et al. (2003). Their inspection of the photometry col-
lected during the second phase of the Optical Gravitational Lensing Experiment
(OGLE) revealed a group of 27 Large Magellanic Cloud (LMC) and 3 Small Mag-
ellanic Cloud (SMC)blue stars which simultaneously show two periods which ratio
is close to 35.2. The light curve of the shorter period (P1) in some instances was
∗Based on observations obtained with the 1.3-m Warsaw telescope at the Las Campanas Observa-
tory of the Carnegie Institution of Washington.
characteristic for eclipsing binaries. The longer periods (P2) were in the interval
between 140 and 960 days and the shape of the light curves was sinusoidal in most
cases. It was speculated that the latter periods were caused by the precession of an
elliptical disk around the blue component of the semi-detached binary. Mennickent
et al. (2005) revealed three new DPVs in the LMC and analyzed the spectroscopic
observations done for some of the systems. The multi-epoch spectroscopy of two
DPVs with small amplitudes and sinusoidal light curves of the short period shows
that their brightness changes are caused by the ellipsoidal variation of one of the
components of the binary system. This finding was crucial for assigning short
periods or twice larger values of all DPVs to orbital periods of binary systems.
Mennickent et al. (2005) also had an impression that the longer cycle variability
disappears during the main eclipse, this finding was negated later on. As a conse-
quence, this variability should be caused by a phenomenon taking place in or near
the surface of brighter component. However, any of the known types of stellar vari-
ability is in the agreement with observed features. Another important finding made
by Mennickent et al. (2005) was a period shortening and an amplitude increase in
the long period of one LMC DPV. Buchler et al. (2009) re-investigated MACHO
project photometry for 30 DPVs selected by Mennickent et al. (2003) and revealed
that after prewhitening with the two most prominent frequencies also a sum of these
frequencies is significant†. No further discussion of this combination frequency is
given in the literature.
Mennickent et al. (2008) observed spectroscopically an eclipsing DPV in the
LMC. Their investigation resulted in a model which contains a binary system with
the Roche lobe-filling secondary, hotter primary star with the circumprimary disk,
mass transfer and mass outflow feeding the circumbinary disk. Contrary to the
previous studies, it was inferred that the source of the longer cycle is in the cir-
cumbinary matter as no interference of short and long period was found. Desmet
et al. (2010) used space mission CoRoT photometry and high-resolution ground-
based spectroscopy to investigate Galactic DPV star AU Mon. CoRoT photometry
was obtained during minimum of the longer cycle. Their conclusions regarding the
model of the binary are consistent with previous studies of other DPVs. High qual-
ity space-based photometry allowed finding two additional periods in residual light
curve. Both these periods are approximately a hundred times shorter than the or-
bital period of the binary. Djuraševi´ c et al. (2010) deepened the analysis of CoRoT
photometry and added a hot spot and two bright spots to the models of the optically
and geometrically thick disk. They explained the period-to-period changes in the
CoRoT light curve by a variable contribution of the disk brightness. This changes
cannot explain the long term cycle and Djuraševi´ c et al. (2010) concluded that the
circumbinary matter has to be responsible for the long cycle variability. Another
Galactic DPV – V393 Sco – was spectroscopically observed by Mennickent et al.
†Note that Buchler et al. (2009) used different frequencies designations in the text and in their
(2010). They favor equatorial mass loss through the Lagrangian point L3and argue
against the polar jets as a source of the longer cycle. The velocity of the mass lost
through L3was estimated to be about 300 km/s.
Finding variability with a period of a few hundred days and amplitude of about
0.1 mag superimposed on the variability caused by a binary system with much
shorter period is possible when the long-term uniformly obtained photometry is
available. Thatis whyalmost the whole photometry ofDPVsanalyzed in thepapers
mentioned above came from the OGLE or the MACHO project – two microlensing
surveys monitoring the LMC and the SMC.In this paper we use OGLE-III photom-
etry to search for DPVs in the LMC. We increase the number of published stars of
this type almost by a factor of four. Our search yields not only the higher number of
DPVs in the LMC, which allows better statistical description of this group of stars,
but also we present several objects with unexpected features. They may be crucial
in understanding the cause of long term cycle, even though, we do not provide any
self-consistent model explaining all observed features of the long term cycle.
In the next sections we discuss observations and the procedure of selecting
DPVs. Section 4 presents the catalog itself and is followed by a discussion of
properties of the whole sample as well as selected objects. We end with a summary
and future prospects. To avoid confusion, we denote phases of the shorter (orbital)
and longer cycle as φ1and φ2, respectively. In the case of the orbital cycle φ1= 0
corresponds to the minimum light (primary eclipse in eclipsing systems), while
for the longer cycle φ2= 0 corresponds to the maximum light, which is more
distinctive in most cases. Similarly, frequencies f1 and f2 are equal 1/P1 and
The OGLE-III project observed the LMC between June 2001 and May 2009
and covered around 40 square degrees. The observations were conducted with the
1.3 m Warsaw Telescope situated at Las Campanas Observatory, which is oper-
ated by Carnegie Institution of Washington. The telescope was equipped with the
eight chip mosaic camera with the total dimension 8k×8k pixels. The pixel size of
15 µm gives the pixel scale of 0.′′26 and the total field of view around 35′×35′.
Thedetailed description of the instrumentation can be found in Udalski (2003). The
photometry was performed using Difference Image Analysis technique (Alard and
Lupton 1998, Alard 2000, Wo´ zniak 2000) which works very well in crowded fields.
Udalski et al. (2008a) give full description of the photometric and astrometric data
reduction process. The DIA photometric uncertainties are known to be underesti-
mated and thus were corrected using the method described by Wyrzykowski et al.
Typically, there are 500 photometric measurements per star obtained during
OGLE-III observations. Around 90% of them were secured using I filter. The
rest was taken in the V-band. For stars in the central parts of the LMC we con-
nected OGLE-III and OGLE-II photometry (Szyma´ nski 2005) obtained from 1997
to 2000. To keep the data in the same photometric system we added to the OGLE-II
magnitudes the difference between median OGLE-III and OGLE-II magnitudes. In
some cases with a different sampling of the short or long cycles, additional cor-
rection found manually was added. Similar procedure was carried out for objects
present in two or three OGLE-III fields in the overlapping parts of adjacent fields.
Our final photometry contains up to 1500 I- and 270 V-band measurements. One
may find B-band mean magnitudes for stars in central parts of the LMC in Udalski
et al. (2000).
3.Selection of DPVs
We have searched for characteristic period ratio in the results of the massive
period search done for OGLE-III photometry of all stars observed in the LMC by
the OGLE-III survey (Soszy´ nski et al. 2008). More intense search was performed
for stars in the region of the color–magnitude diagram where DPVs are expected
(V −I <0.6 mag and V <19 mag; Mennickent et al. 2005). For these blue objects
we used prewhitening with a variety of a different number of Fourier harmonics.
In all cases the stars with the period ratios in a broad region around expected value
of 35, even two times larger or smaller, were visually examined. Altogether 113
DPVs were selected in this way.
We suspected that our searches still may not reveal all the DPVs because au-
tomatic search for eclipsing binary periods may give spurious results (Derekas et
al. 2007). Fourier fitting procedure with fallacious number of harmonics may also
cause some artifacts in prewhitened light curve hampering searches for the addi-
tional longer period. Because of that we decided to visually inspect light curves
of automatically selected eclipsing binaries candidates with periods longer than
one day.‡26000 blue candidates and 7700 candidates without color information
in OGLE-III data were inspected and stars showing light curves typical for DPVs
phased with the orbital period were selected and studied in more detail. Additional
eight stars were added to the list of DPVs. Another three DPVs were found by
authors during other variability searches.
Our list of DPVs in the LMC was cross-matched with the stars found by Men-
nickent et al. (2003) and Mennickent et al. (2005). One of their stars was not
detected by our procedure. Its designation is OGLE-LMC-DPV-114 (MACHO ID:
76.9844.110). This star does not have the V-band magnitude in the OGLE-III pho-
tometric maps (Udalski et al. 2008b) and it was not among the eclipsing binary
candidates inspected visually. We decided to add it to our catalog. We note that
our selection process was less efficient for stars with orbital periods shorter than
one day as many artifacts appear in this period range. Also stars with exceptionally
large P2values or ones with P2≈ 1 yr could have been overlooked.
‡The catalog of Magellanic Clouds eclipsing binaries with periods longer than one day is under
construction by Graczyk et al. (in preparation).
Fig. 1. Relation between the orbital and long period. Uncertainties of periods are smaller than size
of points. Dotted line represents the relation P2= 33.13P1. Gray vertical line shows the periods of
OGLE-LMC-DPV-065 during the last 18 years. See Section 5.7 for discussion.
0 0.51 1.52
Fig. 2. Exemplary non-sinusoidal light curves of long cycles. Values of P2and catalog numbers in
the catalog are given in each panel.