Profiles

An interface to simulation profiles.

What is a Profile?

Each simulation profile provides performance metrics for a simulation, including CPU time, real time (wall time), and peak memory usage. The ansys.aedt.core.modules.profile.Profiles class is derived from :py:class:collections.abc.Mapping. Information for each profile corresponds to a simulation setup and the parametric variation.

The items in the Profiles class are instances of the :class:ansys.aedt.core.modules.profile.SimulationProfile class.

SimulationProfile

Container for all profile data from a single simulation.

A Profiles instance can be retrieved from the solution setup. Each element in the Profiles instance provides access to the SimulationProfile instance for the variation:

>>> from ansys.aedt.core import Hfss
>>> hfss = Hfss(project="/path/to/solved/OptimTee.aedtz")
>>> this_setup = hfss.get_setup("Setup1")
>>> profiles = this_setup.get_profile()
>>> for key in profiles.keys():
...     print(f"Variation: {key.split('-')[-1]}")
...
Variation:  offset='0in'
Variation:  offset='0.10000000000000001in'
Variation:  offset='0.20000000000000001in'
Variation:  offset='0.29999999999999999in'
Variation:  offset='0.40000000000000002in'
Variation:  offset='0.5in'
Variation:  offset='0.59999999999999998in'
Variation:  offset='0.69999999999999996in'
Variation:  offset='0.80000000000000004in'
Variation:  offset='0.90000000000000002in'
Variation:  offset='1in'

Note

Parameter values that define a “variation” may deviate from expectations due to floating point representation error. The correct variation can be obtained using the .keys() method as shown below.

>>> my_profile = profiles[list(profiles.keys())[3]]
>>> bool(my_profile.adaptive_pass)  # Auto-adaptive mesh refinement used?
True
>>> my_profile.num_adaptive_passes  # How many adaptive passes?
6
>>> my_profile.real_time()  # Real time for all adaptive passes?
datetime.timedelta(seconds=7)
>>> my_profile.real_time(num_passes=5)  # Real time for 5 adaptive passes.
datetime.timedelta(3)
>>> my_profile.has_frequency_sweep
False

Some SimulationProfile attributes are instances of the ProfileStep class. A ProfileStep instance, in turn, may be comprised of additional steps that can be accessed as a Pandas.dataframe instance. The SimulationProfile class allows you to retrieve summary data from all steps with the real_time, max_memory and cpu_time attributes.

For example

>>> len(my_profile.adaptive_pass.process_steps)  # Number of adaptive passes
6
>>> step_name = my_profile.adaptive_pass.process_steps[2]  # Select the 3rd adaptive pass.
>>> print(step_name)
"Adaptive Pass 3"
>>> adapt_data = my_profile.adaptive_pass.steps[
...     step_name
... ]  # Retrieve the ProfileStep instance
>>> adapt_table = adapt_data.table(columns=["real_time", "max_memory"])
>>> print(adapt_table.to_string(index=False))  # Pandas dataframe
Step                real_time   max_memory
Adaptive Refine     00:00:02    73.1 M
Simulation Setup    00:00:01    206 M
Matrix Assembly     00:00:02    263 M
Matrix Solve        00:00:05    870 M
Field Recovery      00:00:02    271 M
Data Transfer       00:00:01    252 M

The attribute values of a SimulationProfile instance depend on the type of simulation. The table() method can be used to retrieve a :class:pandas.DataFrame instance summarizing all simulation steps.

Attribute	Simulation Type	Description
.adaptive_pass.steps[key_name]	HFSS	Solution steps during adaptive refinement.
.frequency_sweeps[sweep_name]	HFSS	Frequency sweep steps.
.transient	Maxwell	Transient solution steps.

Some methods and attributes in the Profiles class are:

• elapsed_time (Measured elapsed time for the entire simulation.)

• real_time() (Sum of real time for all process steps.)

• max_memory() (Peak memory over all processes.)

• num_cores (Number of compute cores.)

• product (Which solver was used to generate the profile.)

• num_adaptive_passes (Number of adaptive passes - only if adaptive refinement is used)

• has_frequency_sweep (True if a frequency sweep was run.)

• is_transient (True if the profile is for a transient solution.)

• os (Operating system)

Main profile classes

The following classes cover simulation-level data, steps, and sweeps.

Profiles	Provide an interface to solver profiles.
SimulationProfile	Container for all profile data from a single simulation.
ProfileStep	A profile step possibly containing nested sub-steps.
ProfileStepSummary	Summary information for a single profile step.
TransientProfile	Profile data for a transient solution.
FrequencySweepProfile	Profile data for a frequency sweep.
AdaptivePass	Information for a single adaptive pass.
MemoryGB	Represent memory with conversion to gigabytes and arithmetic.

Utilities

Helper functions for parsing and presentation.

string_to_time	Convert a string to a datetime.timedelta object.
format_timedelta	Format datetime.timedelta for tables.
step_name_map	Map verbose AEDT step labels to compact names.
merge_dict	Recursively merge two dictionaries using type-aware rules.
get_mesh_process_name	Return the name of the meshing process group if present.