lenstronomy.LensModel.Util package¶
Submodules¶
lenstronomy.LensModel.Util.epl_util module¶
-
brentq_inline
¶ A numba-compatible implementation of brentq (largely copied from scipy.optimize.brentq).
Unfortunately, the scipy verison is not compatible with numba, hence this reimplementation :( :param f: function to optimize :param xa: left bound :param xb: right bound :param xtol: x-coord root tolerance :param rtol: x-coord relative tolerance :param maxiter: maximum num of iterations :param args: additional arguments to pass to function in the form f(x, args) :return:
-
brentq_nojit
(f, xa, xb, xtol=2e-14, rtol=3.552713678800501e-15, maxiter=100, args=())[source]¶ A numba-compatible implementation of brentq (largely copied from scipy.optimize.brentq).
Unfortunately, the scipy verison is not compatible with numba, hence this reimplementation :( :param f: function to optimize :param xa: left bound :param xb: right bound :param xtol: x-coord root tolerance :param rtol: x-coord relative tolerance :param maxiter: maximum num of iterations :param args: additional arguments to pass to function in the form f(x, args) :return:
-
cart_to_pol
[source]¶ Convert from cartesian to polar :param x: x-coordinate :param y: y-coordinate :return: tuple of (r, theta)
-
cdot
[source]¶ Calculates some complex dot-product that simplifies the math :param a: complex number :param b: complex number :return: dot-product.
-
geomlinspace
(a, b, N)[source]¶ Constructs a geomspace from a to b, with a linspace prepended to it from 0 to a, with the same spacing as the geomspace would have at a.
-
min_approx
[source]¶ Get the x-value of the minimum of the parabola through the points (x1,y1), …
Parameters: - x1 – x-coordinate point 1
- x2 – x-coordinate point 2
- x3 – x-coordinate point 3
- y1 – y-coordinate point 1
- y2 – y-coordinate point 2
- y3 – y-coordinate point 3
Returns: x-location of the minimum
-
pol_to_cart
[source]¶ Convert from polar to cartesian :param r: r-coordinate :param th: theta- coordinate :return: tuple of (x,y)
-
ps
[source]¶ A regularized power-law that gets rid of singularities, abs(x)**p*sign(x) :param x: x :param p: p :return:
-
solvequadeq
[source]¶ Solves a quadratic equation.
Care is taken for the numerics, see also https://en.wikipedia.org/wiki/Loss_of_significance :param a: a :param b: b :param c: c :return: tuple of two solutions