Method

NumCosmoXcorKernelget_eval_vectorized_full

Declaration [src]

NcXcorKernelIntegrand*
nc_xcor_kernel_get_eval_vectorized_full (
  NcXcorKernel* xclk,
  NcHICosmo* cosmo,
  gint lmin,
  gint lmax,
  NcmSBesselIntegrator* sbi,
  NcXcorKernelClosure closure_type
)

Description [src]

Same as nc_xcor_kernel_get_eval_vectorized(), but integrates with sbi instead of the kernel’s integrator property.

A NcmSBesselIntegratorLevin holds reusable state tied to one multipole range, so a caller computing several ell blocks does better keeping one integrator per block and passing it in here than letting every kernel carry its own. Passing the integrator rather than storing it also keeps xclk free of per-call state, so one kernel can be evaluated for several blocks at once as long as each gets its own sbi.

sbi is unused below the kernel’s l-limber threshold, where no spherical Bessel integral is performed.