Modifiers and shortcuts

mrfmsim adds additional modifiers and shortcuts specific to the experimental simulation to the mmodel modifiers. See modifier and shortcut tutorial for details. The shortcuts are used to modify the Experiment/Model object directly, which consists of applying modifiers to individual nodes or the model/experiment.

List of modifiers and shortcuts for the mrfmsim and mmodel package.

Modifiers/Shorcuts	Module	Description
loop_input	mmodel.modifier	Modify function to iterate one given parameter.
zip_loop_inputs	mmodel.modifier	Modify function to iterate the parameters pairwise.
profile_time	mmodel.modifier	Profile the execution time of a function.
print_inputs	mmodel.modifier	Print the inputs of the function with units.
print_output	mmodel.modifier	Print the outputs of the function with units.
replace_component	mrfmsim.modifier	Replace the component in the model.
numba_jit	mrfmsim.modifier	Modify a node using numba jit.
loop_shortcut	mmodel.shortcut	Loop over a parameter during the experiment execution.
print_shortcut	mmodel.shortcut	Apply print_inputs and print_output shortcuts to individual nodes that print out intermediate variable values during node execution.

modifier module

mrfmsim.modifier.numba_jit(**kwargs)

Numba jit modifier with keyword arguments.

Add metadata to numba.jit. The numba decorator outputs all the parameters make it hard to read. Use the decorator the same way as numba.jit().

mrfmsim.modifier.replace_component(replacement: dict, allow_duplicate=False)

Modify the signature with components.

The modifier modifies the internal model function. The wrapper function is keyword-only. The function by default does not allow duplicated signature. If we want to replace several attributes with a component, the component name cannot exist in the original signature. For example, if we have a function

def func(a, b, obj):

return a + b, obj

and we want to replace a and b with a component, we cannot name the component “obj”.

In rares that we do want to replace a, b with a component,

def func(obj):

return obj.a + obj.b, obj

func(obj=obj)

we would have to use “obj_obj” as the component name. The result function is equivalent to

def func(obj_obj, obj):

return obj_obj.a + obj_obj.b + obj

func(obj_obj=obj, obj=obj)

The behavior is very confusing to users. The solution is to replace the components, but leave the original “obj” signature as it is. In the final signature, the duplicated signatures are combined into one. The solution here is to add a boolean flag allow_duplicate. If the flag is set to True, the function allows duplicated signatures.

The solution, however, leaves another ambiguity. If we indeed want to replace the component with the same name, but use the original name with an attribute:

def func(obj):

return obj.a + obj.b, obj.obj

func(obj=obj)

We have decided that this behavior is not allowed regardless the flag. Because in this case, in an inspection attempt, it is confusing to understand if the obj is the original obj or an attribute to the new obj. In this case, a new object name should be used.

Parameters:

replacement (dict[str]) – in the format of {component_object: [replacement_attribute1, replacement_attribute2, …]}