UWM LSC group 6 month research plan
Aug 15, 1998 - Feb 15, 1999

The following are our research goals for the six-month period Aug 15,
1998 - Feb 15, 1999.

A.  BENCHMARKING/BEOWULF:
    The Milwaukee LSC group is now operating one of the most advanced
    beowulf-type parallel computer systems in the world.  We intend to
    continue the development and testing of this system, as we exercise
    it on filtering data from the 40-meter prototype instrument and
    continue to develop time/frequency and regression methods.  We will
    also carry out any further benchmarking of the data analysis filtering
    pipeline as specifically requested by LIGO project.

B.  GRASP:
    Continue to maintain/repair GRASP.  Use GRASP primarily as a
    documented archive for different research groups that wish to make
    their prototype code publicaly available to record progress and
    dead ends in the search for useful techniques.

C.  BINARY INSPIRAL SEARCH:
    The scientific work for this project is now finished.  We have
    completed a search for inspiralling binary systems in the November
    1994 data set from the Caltech/LIGO 40-meter prototype interferometer,
    and have set an observational upper-bound on the rate of galactic
    inpiral mergers.  During the next six months, we hope to sherpherd
    this work through the LIGO/LSC approval process and into public
    presentation and publication.

D.  40-METER LAB.
    Allen will begin an extended visit to Caltech starting in January
    1999.  At that time, he will return to work on specific projects
    as requested by 40-meter lab.  This will include work on a software
    calibration looop if that is desirable at the time.
  
E.  STOCHASTIC BACKGROUND:
    Allen will work with Romano and Flanagan to study the ways in which
    non-Gaussian and non-stationary noise affects a stochastic background
    correlation search.

F.  WAVELET TECHNIQUES: 
    W. Anderson and Balasubramanian will continue to investigate the
    suitability of wavelets for robust detection algorithms. They will
    also consider their use in detector characterization.
    
G.  TIME/FREQUENCY TECHNIQUES:
    W. Anderson and Balasubramanian will continue to investigate the 
    use of time frequency methods for blind and/or robust searches. 
    They will finish implementing hysteresis methods for signal extraction 
    in time-frequency representations shortly. They will also be evaluating 
    Hough transforms and watershed algorithms as potential methods 
    for signal extraction/identification. False alarm and false dismissal 
    rates for each of these methods in conjunction with windowed Fourier 
    transform, Wigner-Ville distribution and Choi-Williams distributions 
    will be calculated and compared. Anderson and Brady will also explore 
    statistical methods for detecting excess power in the time-frequency 
    plane in as an implemetation of the excess power blind search technique 
    (see item H).

H.  BLIND SEARCH TECHNIQUES
    W. Anderson and Brady (in collaboration with J. Creighton and
    Flanagan) will continue to work on the excess power statistic as a
    means of detecting poorly understood signals, for which only time
    duration and frequency bandwidth information is available. Work
    underway to determine an optimal algorithm for implementation
    as a feasible search strategy will be completed shortly. They
    will also extend the method to multiple detectors in an optimal
    manner, and determine a multi-detector, hierarchical version of
    the algorithm which cirmcumvents the need to bring data streams
    from all the detectors together at the first stage.  W. Anderson
    and Balasubramanian will evaluate time-frequency methods as a blind
    search tool.

I.  REGRESSION TECHNIQUES:
    Continue work on techniques to search for and remove correlations
    in the IFO output arising from different monitored channels such as
    the ground acceleration and ambient magnetic fields.  Explore the
    use of compression techniques like Linear Predictive Coding (LPC)
    to reduce the bandwidth of the data stream.

J.  FRAME FORMAT DATA:
    Continue to maintain the translation code which is used to
    translate old-format 1994 data into FRAME form.  Use this data to
    continually monitor and test the functionality of the FRAME
    library.  Continue to provide tapes of this Nov 1994 data to LIGO
    collaborators in the latest FRAME format.

K:  ASIS
    The Milwaukee LSC group will carry out some of the core activities
    of the Astrophysical Source Identification and Signature subgroup
    of the LSC, including chairing the subgroup, maintaining its user
    software archive, and maintaining the ASIS group web pages.

L:  BINARY INSPIRAL WAVEFORMS

    Wiseman will code the newly-derived 5/2 Post-Newtonian Waveforms into
    GRASP and test them.  Brady will work with J. Creighton and Pullin
    to determine the range of applicability of close limit waveforms
    for quasinormal ringdown searches.

M:  DISCRIMINATION AND VETO TECHNIQUES

    Brady and Balasubramanian will design, implement and test a
    multi-template discrimination and veto test, in collaboration with
    T. Creighton and J. Creighton.

N:  CONTINUOUS WAVE SOURCES

    Brady (with T Creighton) will complete the study of hierarchical
    strategies to search for sources of continuous gravitational waves.
    A manuscript which examines the computational requirements,
    and sensitivities that might be achieved, is in the final stages
    of revision.  The manuscript will be completed and submitted for
    publication.