tardis.montecarlo.montecarlo_numba.formal_integral_cuda module

exception tardis.montecarlo.montecarlo_numba.formal_integral_cuda.BoundsError[source]

Bases: IndexError

Used to check bounds in reverse binary search

class tardis.montecarlo.montecarlo_numba.formal_integral_cuda.CudaFormalIntegrator(geometry, model, plasma, points=1000)[source]

Bases: object

Helper class for performing the formal integral with CUDA.

formal_integral(iT, inu, inu_size, att_S_ul, Jred_lu, Jblue_lu, tau_sobolev, electron_density, N)[source]

Simple wrapper for the CUDA implementation of the formal integral

tardis.montecarlo.montecarlo_numba.formal_integral_cuda.calculate_p_values(R_max, N)[source]

Calculates the p values of N

Parameters
R_maxfloat64
Nint64
Returns
float64
tardis.montecarlo.montecarlo_numba.formal_integral_cuda.calculate_z_cuda(r, p, inv_t)[source]

Calculate distance to p line

Calculate half of the length of the p-line inside a shell of radius r in terms of unit length (c * t_exp). If shell and p-line do not intersect, return 0.

Parameters
rfloat64

radius of the shell

pfloat64

distance of the p-line to the center of the supernova

inv_tfloat64

inverse time_explosio is needed to norm to unit-length

Returns
float64
tardis.montecarlo.montecarlo_numba.formal_integral_cuda.cuda_formal_integral(r_inner, r_outer, time_explosion, line_list_nu, iT, inu, inu_size, att_S_ul, Jred_lu, Jblue_lu, tau_sobolev, electron_density, N, L, pp, exp_tau, I_nu, z, shell_id)[source]

The CUDA version of numba_formal_integral that can run on a NVIDIA GPU.

Parameters
r_innerarray(float64, 1d, C)

self.geometry.r_inner

r_outerarray(float64, 1d, C)

self.geometry.r_outer

time_explosion: float64

self.geometry.time_explosion

line_list_nuarray(float64, 1d, A)

self.plasma.line_list_nu

iTnp.float64
inunp.float64
inu_sizeint64
att_S_ularray(float64, 1d, C)
Jred_luarray(float64, 1d, C)
Jblue_luarray(float64, 1d, C)
tau_sobolevarray(float64, 2d, C)
electron_densityarray(float64, 1d, C)
Nint64
Larray(float64, 1d, C)

This is where the results will be stored

pparray(float64, 1d, C)
exp_tauarray(float64, 1d, C)
I_nu array(floatt64, 2d, C)
zarray(float64, 2d, C)
shell_idarray(int64, 2d, C)
tardis.montecarlo.montecarlo_numba.formal_integral_cuda.intensity_black_body_cuda(nu, temperature)[source]

Calculate the blackbody intensity.

Parameters
nufloat64

frequency

temperaturefloat64

Temperature

Returns
float64
tardis.montecarlo.montecarlo_numba.formal_integral_cuda.line_search_cuda(nu, nu_insert, number_of_lines)[source]

Insert a value in to an array of line frequencies

Parameters
nu(array) line frequencies
nu_insert(int) value of nu key
number_of_lines(int) number of lines in the line list
Returns
int

index of the next line to the red. If the key value is redder than the reddest line returns number_of_lines.

tardis.montecarlo.montecarlo_numba.formal_integral_cuda.populate_z_cuda(r_inner, r_outer, time_explosion, p, oz, oshell_id)[source]

Calculate p line intersections

This function calculates the intersection points of the p-line with each shell

Parameters
r_innerarray(float64, 1d, C)
r_outerarray(float64, 1d, C)
pfloat64

distance of the integration line to the center

ozarray(float64, 1d, C)

will be set with z values. the array is truncated by the value 1.

oshell_idarray(int64, 1d, C)

will be set with the corresponding shell_ids

Returns
int64
tardis.montecarlo.montecarlo_numba.formal_integral_cuda.reverse_binary_search_cuda(x, x_insert, imin, imax)[source]

Find indicies where elements should be inserted to maintain order in an inversely sorted float array.

Find the indices into a sorted array a such that, if the corresponding elements in v were inserted before the indices on the right, the order of a would be preserved.

Parameters
xnp.ndarray(np.float64, 1d, C)
x_insertfloat64
iminint

Lower bound

imaxint

Upper bound

Returns
np.int64

Location of insertion

tardis.montecarlo.montecarlo_numba.formal_integral_cuda.trapezoid_integration_cuda(arr, dx)[source]

Computes the approximation of the trapezoidal integration of the array.

Parameters
arr(array(float64, 1d, C)
dxnp.float64