pySYD
option glossary
Below is a complete list of pySYD
parameters in alphabetical order.
-a, --ask
--all, --showall
creates an additional figure that shows all the iterated background models, which will highlight the selected model
-b, --bg, --background
controls the background-fitting procedure – BUT this should never be touched since a majority of the work done in the software happens here and it should not need to be turned off
--basis
which basis to use for the background fitting (i.e.
'a_b'
,'pgran_tau'
,'tau_sigma'
), NOT OPERATIONAL YETdest =
args.basis
type =
str
default =
'tau_sigma'
--bf, --box, --boxfilter
box filter width for plotting the power spectrum TODO: make sure this does not affect any actual measurements and this is just an aesthetic
dest =
args.box_filter
type =
float
default =
1.0
unit = \(\mu \mathrm{Hz}\)
--bin, --binning
interval for the binning of spectrum in \(\mathrm{log(}\mu\mathrm{Hz)}\) this bins equally in logspace
dest =
args.binning
type =
float
default =
0.005
unit = log(\(\mu \mathrm{Hz}\))
--bm, --mode, --bmode
which mode to choose when binning. Choices are ~[
"mean"
,"median"
,"gaussian"
]dest =
args.mode
type =
str
default =
"mean"
--ce, --cm, --color
change the colormap used in the echelle diagram, which is
'binary'
by defaultdest =
args.cmap
type =
str
default =
'binary'
--cv, --value
the clip value to use for the output echelle diagram if and only if
args.clip_ech
isTrue
. If none is provided, it will use a value that is 3x the median value of the folded power spectrumdest =
args.clip_value
type =
float
default =
3.0
unit = fractional psd
--cli
-d, --show, --display
show output figures, which is not recommended if running many stars
--dnu
option to provide the spacing to fold the power spectrum and “whiten” effects due to mixed modes (
pysyd.target.Target.whiten_mixed
), which also requires a lower and upper folded frequency (i.e. <= dnu) via –le and –ue-e, --est, --estimate
turn off the first module that searches and idenities power excess due to solar-like oscillations, which will automatically happen if numax is provided
--ew, --exwidth
the fractional value of the width to use surrounding the power excess, which is computed using a solar scaling relation (and then centered on the estimated \(\nu_{\mathrm{max}}\))
-f, --fft
--file, --list, --todo
the path to the text file that contains the list of stars to process, which is convenient for running many stars
-g, --globe, --global
do not estimate the global asteroseismic parameter numax and dnu. This is helpful for the application to cool dwarfs, where detecting solar-like oscillations is quite difficult but you’d still like to characterize the granulation components.
--gap, --gaps
- what constitutes a time series gap (i.e. how many cadences)
dest =
args.gap
type =
int
default =
20
see also: -x, –stitch, –stitching
-i, --ie, --interpech
turn on the bilinear interpolation of the plotted echelle diagram
dest =
args.interp_ech
type =
bool
default =
False
action =
store_true
see also: –se, –smoothech
--in, --input, --inpdir
- path to the input data
dest =
args.inpdir
type =
str
default =
INPDIR
--infdir
--info, --information
path to the csv containing all the stellar information (although not required)
dest =
args.info
type =
str
default =
star_info.csv
--iw, --indwidth
width of binning for the power spectrum used in the first module TODO: CHECK THIS
dest =
args.ind_width
type =
float
default =
20.0
unit = \(\mu \mathrm{Hz}\)
-k, --kc, --kepcorr
turn on the Kepler short-cadence artefact correction module. if you don’t know what a Kepler short-cadence artefact is, chances are you shouldn’t mess around with this option yet
dest =
args.kepcorr
type =
bool
default =
False
action =
store_true
--laws, --nlaws
force the number of red-noise component(s). fun fact: the older IDL version of
SYD
fixed this number to2
for the Kepler legacy sample – now we have made it customizable all the way down to an individual star!--lb, --lowerb
the lower frequency limit of the power spectrum to use in the background-fitting routine. Please note: unless \(\nu_{\mathrm{max}}\) is known, it is highly recommended that you do not fix this beforehand
--le, --lowere
the lower frequency limit of the folded power spectrum to “whiten” mixed modes before estimating the final value for dnu
--lp, --lowerp
to change the lower frequency limit of the zoomed in power spectrum (i.e. the region with the supposed power excess due to oscillations). Similar to –ew but instead of a fractional value w.r.t. the scaled solar value, you can provide hard boundaries in this case TODO check if it requires and upper bound – pretty sure it doesn’t but should check
--lx, --lowerx
the lower limit of the power spectrum to use in the first module (to estimate numax)
-m, --samples
option to save the samples from the Monte-Carlo sampling (i.e. parameter posteriors) in case you’d like to reproduce your own plots, etc.
--mc, --iter, --mciter
number of Monte-Carlo-like iterations. This is
1
by default, since you should always check the data and output figures before running the sampling algorithm. But for purposes of generating uncertainties,n=200
is typically sufficient.dest =
args.mc_iter
type =
int
default =
1
--metric
which model metric to use for the best-fit background model, current choices are ~[
'bic'
,'aic'
] but still being developed and testeddest =
args.metric
type =
str
default =
'bic'
-n, --notch
use notching technique to reduce effects from mixes modes (pretty sure this is not full functional yet, creates weird effects for higher SNR cases)
--notebook
similar to –cli, this should not need to be touched and is primarily for internal workings and how to retrieve parameters
--nox, --nacross
specifies the number of bins (i.e. the resolution) to use for the x-axis of the echelle diagram – fixing this number if complicated because it depends on both the resolution of the power spectrum as well as the characteristic frequency separation. This is another example where, if you don’t know what this means, you probably should not change it.
--noy, --ndown, --norders
specifies the number of bins (or radial orders) to use on the y-axis of the echelle diagram NEW: option to shift the entire figure by n orders - the first part of the string is the number of orders to plot and the +/- n is the number orders to shift the ED by
--npb
option for echelle diagram to use information from the spacing and frequency resolution to calculate a better grid resolution (npb == number per bin)
--nt, --nthread, --nthreads
the number of processes to run in parallel. If nothing is provided when you run in
pysyd.parallel
mode, the software will use themultiprocessing
package to determine the number of CPUs on the operating system and then adjust accordingly. In short: this probably does not need to be changeddest =
args.n_threads
type =
int
default =
0
--numax
brute force method to bypass the first module and provide an initial starting value for \(\rm \nu_{max}\)
Asserts len(args.numax) == len(args.targets)
* dest =args.numax
* type =float
* nargs ='*'
* default =None
* unit = \(\mu \mathrm{Hz}\)-o, --overwrite
newer option to overwrite existing files with the same name/path since it will now add extensions with numbers to avoid overwriting these files
--of, --over, --oversample
the oversampling factor of the provided power spectrum. Default is
0
, which means it is calculated from the time series data. Note: this needs to be provided if there is no time series data!--out, --output, --outdir
path to save results to
dest =
args.outdir
type =
str
default =
'OUTDIR'
--peak, --peaks, --npeaks
the number of peaks to identify in the autocorrelation function
dest =
args.n_peaks
type =
int
default =
5
--rms, --nrms
the number of points used to estimate the amplitudes of individual background (red-noise) components Note: this should only rarely need to be touched
dest =
args.n_rms
type =
int
default =
20
-s, --save
turn off the automatic saving of output figures and files
--se, --smoothech
option to smooth the echelle diagram output using a box filter of this width
--sm, --smpar
the value of the smoothing parameter to estimate the smoothed numax (that is really confusing) note: typical values range from
1
-4
but this is fixed based on years of trial & error--sp, --smoothps
the box filter width used for smoothing of the power spectrum. The default is
2.5
but will switch to0.5
for more evolved stars (if \(\rm \nu_{max}\) < 500 \(\mu \mathrm{Hz}\))dest =
args.smooth_ps
type =
float
default =
2.5
unit = \(\mu \mathrm{Hz}\)
--star, --stars
list of stars to process. Default is
None
, which will read in the star list fromargs.file
instead--step, --steps
the step width for the collapsed autocorrelation function w.r.t. the fraction of the boxsize. Please note: this should not be adjusted
dest =
args.step
type =
float
default =
0.25
unit = fractional \(\mu \mathrm{Hz}\)
--sw, --smoothwidth
the width of the box filter that is used to smooth the power spectrum
Warning
All parameters are optimized for most star types but some may need adjusting.
An example is the smoothing width (--sw
), which is 20 muHz by default, but
may need to be adjusted based on the nyquist frequency and frequency resolution
of the input power spectrum.
--thresh, --threshold
the fractional value of the autocorrelation function’s full width at half maximum (which is important in this scenario because it is used to determine \(\Delta\nu\))
dest =
args.threshold
type =
float
default =
1.0
unit = fractional \(\mu \mathrm{Hz}\)
--trials, --ntrials
the number of trials used to estimate numax in the first module – can be bypassed if –numax is provided.
dest =
args.n_trials
type =
int
default =
3
--ub, --upperb
the upper limit of the power spectrum used in the background-fitting module Please note: unless \(\nu_{\mathrm{max}}\) is known, it is highly recommended that you do not fix this beforehand
--ue, --uppere
the upper frequency limit of the folded power spectrum used to “whiten” mixed modes before determining the correct \(\Delta\nu\)
--up, --upperp
the upper frequency limit used for the zoomed in power spectrum. In other words, this is an option to use a different upper bound than the one determined automatically
--ux, --upperx
the upper frequency limit of the power spectrum to use in the first module
-v, --verbose
turn on the verbose output (also not recommended when running many stars, and definitely not when in parallel mode) Check this but I think it will be disabled automatically if the parallel mode is
True
-w, --wn, --fixwn
-x, --stitch, --stitching
-y, --hey
plugin for Daniel Hey’s interactive echelle package but is not currently implemented TODO