Not worth it when it takes less than a second (or two?)

per transit, useful to plot depths with uncertainties over time

Refactor hardcoded self.rp = kwargs.get("rp", (1.42 * R_earth) / R_sun)
Move to the top where all constants are defined. Also:
self.per = 12.9
self.rp = (0.03)
self.a = (23.1)
self.inc = 89.21
self.b = 0.99
self.per = 29
self.rp = (11 * R_earth) / R_sun
self.a = (26.9)

Good for debugging / verification
Good for teaching / explaining the method

Currently clean_array must be called manually. As the runtime is very short and it has no negative side effects, we should call it automatically before each run.

There might be transits shorter than the shortest range in Figure 5.

• Add a parameter to let the user explore the full range
• Same for very long durations?

• Split into separate files (constants, helper functions, core?)
• Re-run black
• hardcoded self.rp

Not my private domain
instead: github? readthedocs.org?

Occasionally, the transit model light curve in the phase-folded plot appears to be too shallow.
Examples: EPICs

• 246389858
• 212006344
• 210968143
• 220621087
• 206154641

Shifting the fitted transit depth down in 1% steps worsens the results.chi2_min monotonically. Therefore, the power spectrum (and SDE) appears to be correct. The error appears to be in the visualization only.

Advantage: Avoid distributing large CSV file in PIP package
Try astroquery for K2 EPIC and K2plr, k2plr

Depending on (standard deviation) and (duration in samples)

Binning the data before running a search increases speed almost quadratically. For very high cadence data, the loss from binning should be small. Add option to bin 2x, 4x, 8x,...

Try https://stackoverflow.com/questions/7343697/how-to-implement-linear-interpolation

These empty return values can not directly be fed into TLS. Extra code necessary to catch/handle on the user side.
Possible solution: Return "none" and handle it internally in TLS (fall back to defaults). Document the behavior.

Is your feature request related to a problem? Please describe.
A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]

Describe the solution you'd like
A clear and concise description of what you want to happen.

Describe alternatives you've considered
A clear and concise description of any alternative solutions or features you've considered.

Additional context
Add any other context or screenshots about the feature request here.

CASE: Light curve is so flat the the transit_depth_min (default 10 ppm) causes no transit to be fit.
Current behaviour: TLS raises a UserError and quits
Desired behaviour: Raise a warning and return a flat power spectrum + all zero values

Allow optional parameter for user to search for very short transits; shorter than previously known (e.g., due to grazing transits)

• New: Balancing speed and detection efficiency
• New: TESS
• New: Command-line interface
• New: Uncertainties

e.g., flares

Currently only available when quadratic limb darkening is used, otherwise returns None

User may provide values leading to an empty period grid.
Then, do not raise a warning and quit.
Instead, provide a default grid, warn, and continue.

k2plr doesn't give them:
import k2plr
star = k2plr.API().k2_star(204099713)
print(star.e_rad)
returns "None"

Vizier has it, but interface not working?
astropy/astroquery#1326

Add CUDA implementation

• numpy warnings (mean of empty slice, etc) --> are currently hidden with numpy.warnings.filterwarnings('ignore')
• fix numpy deprecation warnings
• snr_pink_per_transit
• odd_even_mismatch

Use the ternary algorithm
Required when fitting signal shapes that have no analytical correspondence between max and mean depth (e.g., flares)

Dear Sir or Madam,

I have been using the MAST API in the past month and the examples from https://mast.stsci.edu/api/v0/pyex.html worked great. Recently I noticed that something broke. When I use the service "Mast.Catalogs.Filtered.Tic", I get the error:

{"status" : "ERROR","msg" : "The column prefix 'c' does not match with a table name or alias name used in the query."}

My query is:

mastQuery(
        {
            "service": "Mast.Catalogs.Filtered.Tic",
            "format": "json",
            "params": {
                "columns": "c.*",
                "filters": adv_filters,
            },
        }

Currently, TLS silently swallows unknown parameters.
Better handling should be: Reject, or at least warn

Needed for the cluster search

Compensate for morphological light-curve distortions (temporal smearing effects) due to finite integration time (Kipping 2010). Following his Eq. 40 we can estimate that this is probably not necessary to re-compute at every period, but only occasionally, so that the speed loss is small. Can also be made an optional parameter.

• Determine code coverage https://coverage.readthedocs.io/en/v4.5.x/
• add additional test cases to improve coverage (e.g., of constants)
• include tests for command line (+default config)

If user provides mass, radius with very small values, the period_grid can contain a very large number of periods. Providing unrealistically small mass and radius can happen by accident or from faulty catalog values. Possible solutions:

• Reject unphysical stellar values (e.g., make R_star_min > 0.1)
• Warn if len(period_grid)>X where e.g. X>10^6

When I was trying to pip install, it reports

(python37) [boma@16:48:11@~/transit/tls] pip install tls
Collecting tls
Could not find a version that satisfies the requirement tls (from versions: )
No matching distribution found for tls

I have python 3.7 installed using Conda. Not sure which version is required by TLS.

Searches

• K2 multiples
• Paper: Quad system
• Paper: New multis
• K2 full
• K2 multiples
• K1 full
• TESS

Theory

• Determine effect of TTVs on SDE:
• Is a TTV-smeared transit more transit-like (or more box-like)?
• Are very short transits (outside of Figure 5) detectable?
• See previous estimates about exoplanet detectability around White Dwarfs (Agol 2011) and other exotic places (Imara & Stefano 2017).
• If this is realistic, how should TLS be adapted?
• Simply change the shortest transit duration to 1 cadence?
• How big is the morphological shape deformation with "long" cadences?
• Is the best transit template V-shaped?
• Account for TTVs
• For BLS this has been implemeneted: Quasiperiodic Automated Transit Search (Carter & Agol 2012)
• Is this computationally realistic fo TLS? The major additional effort should be non-constant phase folding?
• Determine detection biases:
• Observational biases for transiting planets (Kipping & Sandford 2016) are partly due to the box shaped transit fit (when using BLS).
• Even when fitting a better transit shape like the TLS template, similar biases can be expected since our template curve cannot be a perfect fit for all transits.
• For example, an eccentric or V-form grazing transit shape is substantially different from a box.
• This causes increased noise, resulting in lower detection efficiency.
• A few real, but rare, transit shapes might be closer to a box than to the reference transit template, resulting in a different set of observational biases.
• Characterizing these can be a natural follow-up work.
• Benchmark machine-learning vs. TLS:
• Speed
• Detection efficiency
• Develop a heuristic of a shallowest transit depth to be fitted, as a function of noise in the data, period, and other factors.
• What is the optimal (most sparse but still accurate) period grid to re-calculate the transit shape to correct for morphological deformations
• PLATOsim