LSC Detector Characterization Working Group
Subgroup on Data Set Simulation
Leader: Sam Finn
Data set simulation immediately brings to mind simulation of astrophysical
sources, which is the natural purview of both the Astrophysical
Source Indentification and Signatures (ASIS) and the Detection
Confidence and Statistical Analysis working groups. But equally important
is the simulation of purely instrumental & environmental effects. Ultimately,
we need to verify that sources can be detected in a background of instrumental
noise; more immediately, we need to verify that algorithms developed to
characterize the detector can indeed detect and correctly classify noise,
including transient disturbances.
Tasks for this subgroup can be classified into three broad areas with
differing time scales for completion:
-
Creating simulation software infrastructure (short-term) --> First
release of SimData software
-
Applying well understood models to create simulated data sets (medium-term)
-
Improving existing models (long-term) for more sophisticated/realistic
simulation (long-term)
Creating simulation software infrastructure:
These tasks include writing generic utility routines for creating random
time series data from power spectra and for superposing different noise
sources / disturbances, including those from astrophysical sources. Ideally,
a fairly general template should be created that can be emulated by writers
of algorithms to mimic their favorite disturbances. More generic utilities
needed include modelling of analog whitening/dewhitening filters, digitization,
and digital filters.
Applying (reasonably) well understood models to create simulated
data sets:
These tasks include modelling of the following effects:
(see August 98 transparencies by D. Sigg
for a more comprehensive list)
-
Violin modes
-
60 Hz line noise & harmonics
-
Servo gain peaking
-
Impulses (e.g., dropped hammer, wire relaxation, BNC connector slippage)
-
ADC malfunctions (e.g., spikes, saturation, stuck bits)
-
Cross-coupling in longitudinal servo controls
Improving existing models:
These tasks are much more open-ended than those above and will require
considerable work. A large simulation effort is already underway by the
LIGO Lab called the End-to-End
model. This package is meant to provide time- and frequency-domain
simulations of a generic recycled interferometer, with a large set of optical,
geometric, rf, and environmental parameters to define the interferometer.
Working group members can contribute to this effort directly or indirectly
by working on specific modelling tasks, such as those listed below:
-
Size and effects of residual higher-order transverse optical modes
-
Size and effects of residual contrast defect between the two arms
-
Cross-coupling between orientation and longitudinal servo controls
-
Sensitivity of various optical pickoff signals
-
Ground & Stack motion
-
Thermal lensing
-
Light scattering and diffraction
-
Internal electronic saturation
General Plan
The above laundry list of tasks is formidable. Some are clearly more urgent
than others, and manpower should be allotted accordingly. The following
general plan seems appropriate:
-
Set priorities and find volunteers
-
Create simulation infrastructure
-
Implement the more straightforward modelling
-
Start looking at improved modelling