cooling_cuda.cu File Reference

Functions to calculate cooling rate for a given rho, P, dt. More...

#include <math.h>
#include "../cooling/cool_components.h"
#include "../cooling/cooling_cuda.h"
#include "../cooling/load_cloudy_texture.h"
#include "../global/global.h"
#include "../global/global_cuda.h"
#include "../utils/error_handling.h"
#include "../utils/gpu.hpp"

Include dependency graph for cooling_cuda.cu:

Classes
class	CoolingUpdateExecutor< CoolingRecipe >
	Instances of this class template are callables that serve as callback functions for applying cooling to the grid. More...
struct	CoolRecipeCIE
	Analytic fit to a solar metallicity CIE cooling curve calculated using Cloudy. More...
class	CoolRecipeCloudy
	Uses texture mapping to interpolate Cloudy cooling/heating tables at z = 0 with solar metallicity and an HM05 UV background. More...
class	CoolRecipeCloudyAndPhotoHeating
class	CoolRecipeTIAndCIE
	Analytic cooling/heating recipe that roughly matches the "TI" cooling runs shown in Kim & Ostriker 2015 More...
class	CoolRecipeTIAndCIEAndPhotoHeating
	Analytic cooling/heating recipe that roughly matches the "TI" cooling runs shown in in Kim & Ostriker 2015 And that includes constributions from photoelectric Heating. More...

Functions
template<typename CoolingRecipe >
__global__ void	cooling_kernel (Real *dev_conserved, int nx, int ny, int nz, int n_ghost, int n_fields, Real dt, Real gamma, CoolingRecipe recipe)
std::function< void(Grid3D &)>	configure_cooling_callback (std::string kind, ParameterMap &pmap)

Detailed Description

Functions to calculate cooling rate for a given rho, P, dt.

Nearly all of the functionality implemented in this file follow a common strategy. At this time of writing, there is essentially 1 functions that deviates from the strategy (primordial_cool), which are is left over from earlier implementations.

Interface

In detail, the configure_cooling_callback function produces a std::function, which serves as a callback that performs cooling

• a std::function instance acts a more generalized function-pointer that is able to wrap ordinary functions OR a struct that can act like a function (sometimes known as a "functor" or "callable")
• at this time, the callback will perform cooling, with specialized code based on the cooling recipe, that acts modifies the fields tracked by a grid object. The actual implementation of the callback is opaque to the rest of cholla.

Implementation Strategy

At this time of writing, we implement cooling functionality with some basic template-machinery

• our use of templates allows us to create optimal code for each "cooling recipe", while minimizing duplicated code and avoiding conditional compilation with ifdef statements

Our idea revolves around the concept of a CoolingRecipe.

• we loosely define a CoolingRecipe as any type that implements a __device__ member-function with the Real cool_rate(Real n, Real T) function signature (i.e. it computes the cooling rate per unit volume at a given number density and temperature)
• in principle, this may or may not include the effects of photoelectric heating (it may eventually make more sense to model photoelectric heating separately)

To perform cooling with a given recipe, we an instance of the cooling_recipe to the global function, cooling_kernel. The concrete type of the CoolingRecipe is a template parameter, cooling_kernel, so that invocations of kernels are effectively specialized for each type of recipe.

The CoolingUpdateExecutor class template simply serves as a nice way to package the particular kind of CoolingRecipe (and and cooling-recipe-specific parameters) with the logic for launching cooling_kernel.

Function Documentation

configure_cooling_callback()

std::function< void(Grid3D &)> configure_cooling_callback	(	std::string	kind,
		ParameterMap &	pmap
	)

construct a callback that performs heating/cooling (to approximate the impact of chemistry).

The returned callback can be used as the chemistry_callback