sherlockpipe.single_transits.search.MoriartySearch

class sherlockpipe.single_transits.search.MoriartySearch(object_dir, object_id, is_candidate_from_search, candidates_df, transits_mask=[], batch_size=256, threshold=0.5, cache_dir='/home/docs/')

Bases: ToolWithCandidate

Searches for single transit events using the SANTO neural network model.

Extends ToolWithCandidate to perform single transit detection: loads a pre-trained SANTO model, predicts transit probabilities over the light curve, identifies positive regions, fits box-shaped transit models, and estimates orbital periods.

__init__(object_dir, object_id, is_candidate_from_search, candidates_df, transits_mask=[], batch_size=256, threshold=0.5, cache_dir='/home/docs/') → None

Initialize the Moriarty single-transit search.

Parameters

object_dirstr or None

Directory containing the object’s data files.

object_idstr

Identifier for the target object.

is_candidate_from_searchbool

Whether the candidate comes from a SHERLOCK search.

candidates_dfpandas.DataFrame

DataFrame of candidate information for transit masking.

transits_masklist of dict, optional

List of transit masks with keys ‘P’, ‘T0’, ‘D’ to exclude known transits.

batch_sizeint, optional

Batch size for SANTO predictions. Default is 256.

thresholdfloat, optional

Probability threshold for single-transit classification. Default is 0.5.

cache_dirstr, optional

Directory to cache downloaded models. Default is user home.

Methods

__init__(object_dir, object_id, ...[, ...])

Initialize the Moriarty single-transit search.

greater_than_snr_projected(max_period, ...)

Check if projected SNR for a given period exceeds the threshold.

in_transit_mask(t0, duration_mins)

Check if a given time and duration overlaps with any masked transit windows.

is_candidate_aware()

Boolean return to inform whether the candidate to be processed comes from SHERLOCK searches or is user-given.

mask_previous_candidates(time, flux, ...)

Masks all the candidates found in previous runs in the SHERLOCK search.

nearest_epoch_offset(ts, t0, P[, signed])

Devuelve la diferencia temporal entre ts y el t0 periódico más cercano.

object_dir()

Return the object data directory.

plot_lc_preds_spectrum_broken_x(time, flux, ...)

Figura 3x1 con 'broken X' en las dos primeras filas (curva de luz y predicciones) y el espectro completo en la tercera fila.

plot_light_curve_broken_x(time, flux, ...[, ...])

Broken-X light curve.

roche_limit_period(M_star_sun, R_star_sun[, ...])

Approximate orbital period (in days) at the fluid Roche limit for a planet orbiting a star, assuming a fiducial planet density.

run(cpus, **kwargs)

Execute the single-transit search pipeline.

sample_periods_single_transit(M_star_sun, ...)

Estima una distribución de períodos plausibles para un tránsito único, asumiendo órbita circular y aproximación geométrica simple.

Attributes

watson

greater_than_snr_projected(max_period, depth, duration_s, time_array, flux_std, single_snr, t0=0, projected_snr=10)

Check if projected SNR for a given period exceeds the threshold.

For a given orbital period, this function estimates how many transits could occur within the observation window and calculates the projected SNR from combining multiple transits.

Projected SNR = single_snr * sqrt(n_transits)

Returns True if projected_snr > threshold (reject the candidate).

Parameters

max_periodfloat

Maximum orbital period to consider (days)

depth_pptfloat

Transit depth in parts per thousand

duration_hfloat

Transit duration in hours

time_arraynp.ndarray

Time array of observations (days)

flux_stdfloat

Flux standard deviation (ppt)

single_snrfloat

Single-transit SNR

t0float

Transit epoch (days, default 0)

projected_snrfloat

SNR threshold (default 10)

Returns

bool

True if projected SNR exceeds threshold (reject), False otherwise

in_transit_mask(t0, duration_mins)

Check if a given time and duration overlaps with any masked transit windows.

Parameters

t0float

Transit epoch in days.

duration_minsfloat

Transit duration in minutes.

Returns

bool

True if the transit overlaps with any existing transit mask, False otherwise.

is_candidate_aware()

Boolean return to inform whether the candidate to be processed comes from SHERLOCK searches or is user-given.

Returns:

boolean with the value

mask_previous_candidates(time, flux, flux_err, candidate_id)

Masks all the candidates found in previous runs in the SHERLOCK search.

Parameters:

  • time – the time array

  • flux – the flux measurements array

  • flux_err – the flux error measurements array

  • candidate_id – the candidate number

Returns:

time, flux and flux_err with previous candidates in-transit data masked

nearest_epoch_offset(ts, t0, P, signed=True)

Devuelve la diferencia temporal entre ts y el t0 periódico más cercano. ts, t0, P en las mismas unidades (p.ej., días BJD_TDB). Acepta escalares o arrays (vectorizado).

object_dir()

Return the object data directory.

Returns

str

Path to the object’s data directory.

plot_lc_preds_spectrum_broken_x(time, flux, predictions, power_x, spectra, zero_epsilon, lcs_dir, target_file, gap_threshold=0.5, max_panels=6, min_points_per_panel=200, marker_size=1, dpi=300)

Figura 3x1 con ‘broken X’ en las dos primeras filas (curva de luz y predicciones) y el espectro completo en la tercera fila.

  • No se colapsa el eje tiempo: se divide en segmentos (columnas) por huecos > gap_threshold.

  • Se muestran hasta ‘max_panels’ segmentos (los mayores), en orden cronológico.

  • La tercera fila (espectro) ocupa todo el ancho (no se divide).

Parámetros clave:

time, flux, predictions: arrays 1D (se usan hasta N puntos = len(predictions)). power_x, spectra: arrays para el espectro (no dependen de time). zero_epsilon: umbral para calcular límites Y robustos del panel de flux.

plot_light_curve_broken_x(time, flux, half_duration_points, t0, cadence, moriarty_dir, object_id, target_fit, out_path, gap_threshold=0.5, max_panels=6, min_points_per_panel=200, color_oot='blue', color_it='red', marker_size=6, draw_t0_line=True)

Broken-X light curve. In-transit = exactly half_duration_points samples on each side of the sample whose time is nearest to t0 (index-based definition).

roche_limit_period(M_star_sun, R_star_sun, rho_p=5500.0)

Approximate orbital period (in days) at the fluid Roche limit for a planet orbiting a star, assuming a fiducial planet density.

Uses the classical fluid Roche limit:

a_R = 2.44 * R_star * (rho_star / rho_p)**(1/3)

and Kepler’s third law:

P = 2π * sqrt(a_R^3 / (G * M_star))

Parameters

M_star_sunfloat

Stellar mass in solar masses.

R_star_sunfloat

Stellar radius in solar radii.

rho_pfloat, optional

Assumed mean planetary density in kg/m^3. Default is 5500 kg/m^3 (Earth-like).

Returns

P_roche_daysfloat

Orbital period at the Roche limit in days.

run(cpus, **kwargs)

Execute the single-transit search pipeline.

Downloads the SANTO model if needed, runs predictions on the light curve, identifies positive regions, fits box-shaped transit models, estimates orbital periods, generates diagnostic plots, and creates a PDF report.

Parameters

cpusint

Number of CPUs to use.

**kwargsdict

Additional keyword arguments (currently unused).

sample_periods_single_transit(M_star_sun, R_star_sun, T14_days, T14_err_days=None, R_p_earth=1.0, n_samples=100000, b_max=0.9, geometric_b_prior=True, random_state=None)

Estima una distribución de períodos plausibles para un tránsito único, asumiendo órbita circular y aproximación geométrica simple.

Parámetros

M_star_sunfloat

Masa estelar en masas solares.

R_star_sunfloat

Radio estelar en radios solares.

T14_daysfloat

Duración del tránsito (1º a 4º contacto) en días.

T14_err_daysfloat or None

Incertidumbre (1 sigma) en la duración, en días. Si None, se toma fija.

R_p_earthfloat

Radio planetario (valor más probable) en radios terrestres.

n_samplesint

Número de muestras Monte Carlo.

b_maxfloat

Máximo valor de parámetro de impacto que permitimos (en unidades de R*).

geometric_b_priorbool

Si True, usa prior geométrico p(b) ∝ b en [0, b_max]. Si False, prior uniforme en b.

random_stateint or None

Semilla para reproducibilidad.

Devuelve

periods_daysnp.ndarray

Array de períodos en días (filtrado a valores físicos).

summarydict

Diccionario con estadísticos (median, p16, p84, etc.).