tapsap.preprocess package
Submodules
tapsap.preprocess.baseline_correction module
- tapsap.preprocess.baseline_correction.baseline_correction(flux: ndarray, times: ndarray, baseline_time_range: Optional[list] = None, baseline_amount: Optional[float] = None) dict
Baseline correction of the flux through user input. The user can either provide a list of baseline time ranges ([start, end]), the baseline amount (0.2), or neither where the last 95% of the flux will be taken as the baseline.
- Args:
flux (float ndarray): The outlet flux.
times (float ndarray): An array of time.
baseline_time_range (float list): The points in time in which to take the baseline.
baseline_amount (float): The amount of baseline correction to apply.
- Returns:
corrected_flux, baseline_amount (dict): The baseline corrected flux and the baseline amount.
- Citation:
None
- Implementor:
Ross Kunz
- Link:
None
tapsap.preprocess.baseline_gamma module
- tapsap.preprocess.baseline_gamma.baseline_gamma(flux: ndarray, times: ndarray) dict
Baseline the flux through examining the tail of the gamma distribution.
- Args:
flux (float ndarray): The outlet flux.
times (float ndarray): An array of time.
- Returns:
corrected_flux, baseline_amount (dict): The baseline corrected flux and the baseline amount.
- Citation:
Kunz et al, “A Priori Calibration of Transient Kinetics Data via Machine Learning” (In prep)
- Implementor:
Ross Kunz
- See also:
tapsap.utils.gamma_pdf
- Link:
tapsap.preprocess.calibration_coef module
- tapsap.preprocess.calibration_coef.calibration_coef(flux: ndarray, calibration_amount: Optional[float] = None) dict
Apply calibration correction to the flux through user input. This will result in a value of the flux multiplied by the calibration amount.
- Args:
flux (float ndarray): The outlet flux.
calibration_amount (float): The amount to scale the flux.
- Returns:
corrected_flux, calibration_amount (dict): The calibration corrected flux and the calibration amount.
- Citation:
None
- Implementor:
Ross Kunz
- Link:
None
tapsap.preprocess.calibration_teak module
- tapsap.preprocess.calibration_teak.calibration_teak(flux: ndarray, inert_flux: ndarray, times: ndarray, flux_mass: float, inert_mass: float, huber_loss: bool = False, constraints: bool = True) dict
This function determines the calibration coefficients for the flux based on the inert flux. Note that if the flux is a product, the difference between the inert and the calibrated reactant should be used as the flux value. For example, (Inert - Reactant) >= Product.
- Args:
reactant_flux (float ndarray or list of ndarrays): The outlet flux of the reactant.
product_flux (float ndarray or list of ndarrays): The outlet flux of the product.
inert_flux (float ndarray): The outlet flux of the inert.
times (float ndarray): An array of time.
flux_mass (float): The mass of the flux.
inert_mass (float): The mass of the inert flux.
huber_loss (bool): Use a robust loss function rather than the standard square error loss.
constraints (bool): Apply the molecule constraints. If false, tap_mix performs regular linear regression.
- Returns:
[corrected_flux, calibration_amount] (dict): The calibration corrected flux and the calibration amount.
- Citation:
Kunz et al, “A Priori Calibration of Transient Kinetics Data via Machine Learning” (In prep)
- Implementor:
Ross Kunz
- See also:
tapsap.preprocess.tap_mix
tapsap.moments.isreversible
- Link:
tapsap.preprocess.smooth_flux_gam module
- tapsap.preprocess.smooth_flux_gam.smooth_flux_gam(flux: ndarray, smooth_amount=0.0001) ndarray
Smoothing via a Generalized Additive Model applied to the flux.
- Args:
flux (float array): The outlet flux
- Returns:
smoothed_flux (float array): The smoothed outlet flux
- Citation:
Hastie et al. “Generalized additive models”
Kunz et al, “A Priori Calibration of Transient Kinetics Data via Machine Learning” (In prep)
- Implementor:
Ross Kunz
- Link:
tapsap.preprocess.tap_mix module
- tapsap.preprocess.tap_mix.tap_mix(X: ndarray, y: ndarray, times: ndarray, huber_loss: bool = False, constraints: bool = True, fit_intercept: bool = True, enforce_max: bool = False) dict
Optimization of the calibration coefficient or fragmentation. This method is a convex optimization fitting the response (y) to a single or set of flux. For example, given an inert and reactant, then y (of length n) would be the inert and X would be a matrix of a single flux with a shape of n by 1. This can also be used to extract fragmentation patterns from a single mass measurement.
- Args:
X (float ndarray): A set of flux responses used in describing y.
y (float ndarray): The flux that contains all of X.
times (float ndarray): An array of time.
huber_loss (bool): Use a robust loss function rather than the standard square error loss.
constraints (bool): Apply the molecule constraints. If false, tap_mix performs regular linear regression.
fit_intercept (bool): Fit the intercept within the convex optimization.
enforce_max (bool): Enforce the maximum of the X values must be less than y.
- Returns:
corrected_flux, calibration_amount (dict): The calibration corrected flux and the calibration amount.
- Citation:
Kunz et al, “A Priori Calibration of Transient Kinetics Data via Machine Learning” (In prep)
- Implementor:
Ross Kunz
- Link:
tapsap.preprocess.tap_mix_opt module
- tapsap.preprocess.tap_mix_opt.tap_mix_opt(X: ndarray, y: ndarray) dict
This is an alteration of tap_mix, but with a simple solver. In some cases, the convex solver fails and if it does, this method should be applied. Optimization of the calibration coefficient or fragmentation. For example, given an inert and reactant, then y (of length n) would be the inert and X would be a matrix of a single flux with a shape of n by 1. This can also be used to extract fragmentation patterns from a single mass measurement.
- Args:
X (float ndarray): A set of flux responses used in describing y.
y (float ndarray): The flux that contains all of X.
- Returns:
corrected_flux, calibration_amount (dict): The calibration corrected flux and the calibration amount.
- Citation:
Kunz et al, “A Priori Calibration of Transient Kinetics Data via Machine Learning” (In prep)
- Implementor:
Ross Kunz
- Link: