sluyspy.fit module

Fitting functions for the sluyspy package

sluyspy.fit.correlation_matrix_from_variance_covariance_matrix(var_cov)[source]

Compute the normalised correlation matrix from a variance-covariance matrix.

Parameters:

var_cov (float) – 2D NumPy array containing the variance-covariance matrix.

Returns:

2D NumPy array containing the normalised-correlation matrix.

Return type:

(float)

sluyspy.fit.np_polyfit_chi2(xvals, yvals, order, ysigmas=None, verbosity=0)[source]
Wrapper for NumPy polyfit, that returns the fitting coefficients and reduced chi squared.

Print details if desired.

Note: does not compute the uncertainties in the coefficients; use scipy_curvefit_chi2() for that.

Parameters:

  • xvals (float) – X values.

  • yvals (float) – Y values.

  • order (int) – Polynomial order for fit.

  • ysigmas (float) – Uncertainties (errors, standard deviations) in the y values; defaults to None -> all sigmas=1.

  • verbosity (int) – Verbosity: 0: quiet, 1: print red.chi2, 2: print more, 3: print table, 4: print all; defaults to 0.

Returns:

Tuple consisting of (coefs, red_chi2):

  • coefs (float): Array of fitting coefficients, with length (order+1).

  • red_chi2 (float): Reduced chi squared (chi^2 / (n-m)) for the fit.

Return type:

(tuple)

sluyspy.fit.polynomial(x, *coefs)[source]

Return a polynomial y = Sum_i=0^N-1 coef_i x^i.

Parameters:

  • x (float) – X value(s).

  • coefs (float) – Array of polynomial coefficients. The polynomial order is defined by the number of coefficients in the array.

Returns:

Polynomial value.

Return type:

(float)

sluyspy.fit.print_fit_details(fittype, coefs, xvals, yvals, ysigmas, dcoefs=None, var_cov=None, fit_fun=None, yfit=None, verbosity=2, abs_diff=True, rel_diff=True, sigdig=6, coef_names=None, coef_facs=None, rev_coefs=None, yprintfac=1, rel_as_pct=False)[source]

Compute and return the reduced chi^2, and print other fit details if desired.

Calculations are done without fitting, so that this function can be called after a fit. The fit values will be computed from coefs, xvals using fit_fun if specified. In fact, two series of yvals (‘true’ and ‘fit’) can be specified without any fitting ever, by setting fittype=coefs=xvals=None and specifying yfit. This will print statistics on the comparison of the two data sets, similar to those after a fit. If xvals is specified, a bit more detail can be printed.

Parameters:

  • fittype (str) – Type of fit: ‘np_polyfit’, ‘scipy_curvefit’ or None.

  • coefs (float) – Array with fit coefficients.

  • xvals (float) – Array with x values.

  • yvals (float) – Array with y values.

  • ysigmas (float) – Array with y sigmas.

  • dcoefs (float) – Uncertainties in coefficients from scipy_curvefit; optional.

  • var_cov (float) – Variance-covariance matrix in 2D NumPy array; optional.

  • fit_fun (fun) – Fit function used for scipy_curvefit; optional.

  • yfit (float) – “Fit” values for Y for fittype=None; optional.

  • verbosity (int) – Verbosity (0-3); optional - defaults to 2.

  • abs_diff (bool) – Print absolute differences (optional; defaults to True).

  • rel_diff (bool) – Print relative differences (optional; defaults to True).

  • sigdig (int) – Number of significant digits to print (optional, defaults to 6).

  • coef_names (str) – Array of coefficient names for printing (optional).

  • coef_facs (float) – Array of coefficient multiplication factors for printing (optional; defaults to 1). Useful for e.g. printing degrees when radians are fitted.

  • rev_coefs (bool) – Reverse printing order in coefficient table: last coefficient at top. Optional, defaults to True for fittype np_polyfit, to False otherwise.

  • yprintfac (float) – Multiplication factor when printing absolute differences in y (optional, defaults to 1).

  • rel_as_pct (bool) – Print relative difference as percentage rather than fraction (optional, defaults to False).

Returns:

Reduced Chi^2.

Return type:

(float)

Note

  • If fittype = ‘scipy_curvefit’, fit_fun must be specified.

  • If fittype is None, yfit values must be specified, since they will not be computed from xvals and coefs. In that case, coefs=xvals=None can be specified.

sluyspy.fit.scipy_curvefit_chi2(fit_fun, xvals, yvals, coefs0, ysigmas=None, verbosity=0)[source]
Wrapper for SciPy’s curve_fit, that returns the fitting coefficients and reduced chi^2.

Print details if desired.

Parameters:

  • fit_fun (function) – Fitting function, should look like def fit_fun(x, a,b,c): and return y=f(x, a,b,c).

  • xvals (float) – Array containing x values to fit.

  • yvals (float) – Array containing y values to fit.

  • coefs0 (float) – Array with initial guess for fitting coefficients.

  • ysigmas (float) – Array containing y sigmas/uncertainties.

  • verbosity (int) – Verbosity to stdout (0-4).

Returns:

Tuple containing (coefs, dcoefs, red_chi2, var_cov, ier):

  • coefs (float): Array containing the final fitting coefficients.

  • dcoefs (float): Array containing uncertainties on the fitting coefficients.

  • red_chi2 (float): Reduced chi squared (chi2/(n-m)).

  • var_cov (float): 2D array containing the variance-covariance matrix.

  • ier (int): Return value, >0 if the fit succeeded.

Return type:

(tuple)