The Net Advance of Physics: The Nature of Dark Matter, by Kim Griest -- Section 7D.

Event Selection

Most of the stars we monitor are constant within our photometric

errors, but about one half of one percent are variable. The MACHO

database, as repository for the largest survey ever undertaken in the

time domain, is an extremely valuable resource for studies of

variable stars. From our first year LMC data alone we have already

identified about 1500 Cepheid variables, 8000 RR Lyrae, 2200

eclipsing binaries, and 19000 long period variables. Example

lightcurves from each of these classes can be found in reference

[46,47]. We also have many rare variables, and have given the first

conclusive evidence of 1st overtone pulsation in classical Cepheids

[48]. We have also observed what may turn out to be entirely new

types of variable stars [46].

Given that the incidence of stellar variability, systematic error, and

other sources of stellar brightening is much higher than the

incidence of microlensing, how can one hope to discriminate the

signal from the background among the tens of millions of stars we

monitor nightly? Fortunately, there are several very powerful

microlensing signatures which exist:

1. High amplification. Very high amplifications are possible, so

   we can set our threshold high enough to avoid many

   types of systematic error background.

2. Unique shape of lightcurve. Only 5 parameters are needed to

   completely specify the 2-color lightcurve.

3. Achromaticity. Lensing magnification should be equal at all

   wavelengths, unlike brightenings caused by most types of

   stellar variability.

4. Microlensing is rare. The chance of two microlensing events

   occurring on the same star is so small, that any star with more

   than one ``event" can be rejected as a microlensing candidate.

5. Statistical tests: The distribution of peak magnifications

   is known a priori. Microlensing should occur with equal

   likelihood on every type and luminosity of star, unlike known

   types of stellar variability. New microlensing events should be

   discovered each year at a constant rate.

6. Alert possibility. Our alert system is now working and we can

   catch microlensing before the peak and get many

   measurements of high accuracy. Other spectral and

   achromaticity tests can also be performed in follow-up mode.

Using these criteria, as well as others, we have found it possible to

pick out microlensing candidates from variable stars, etc. For

example, starting with about 9.5 million lightcurves from our first

year LMC database, we remove all but 3. These are shown in

Figure 7 ([fig], [captions]).

One of these events is clearly superior in signal/noise to the others,

and we have confidence in the microlensing label. It has ,

and days. The other two, while passing all our cuts, and

certainly consistent with microlensing, are less certain to be actual

microlensing. We should note that our alert system has found a

couple more high signal/noise LMC microlensing events, which are

not included here, since we have performed efficiency calculations

only on the first year data set.

Now, if we had found only these 3 events towards the LMC, we

would not be as confident as we are, that we have seen microlensing.

However, we have many more events towards the galactic bulge,

and some of these are of incredibly high signal/noise. We cannot use

the same selection criteria for the bulge as for the LMC since our

observing schedule towards the bulge is different, and the bulge

stellar population, distance, crowding, and extinction are different,

but using the same statistics, we can make a similar selection

procedure. We find about 43 candidates in our first year data (and

since then a few dozen more in our alert system). Examples of

lightcurves from the bulge are shown in Figure 8 ([fig], [captions]).

Some of these events are truly beautiful, with durations of many

months and magnifications of almost 20. Coupled with the dozen

events from the OGLE collaboration, I think little doubt remains

that microlensing has been seen.

BIBLIOGRAPHY