Note
Go to the end to download the full example code.
Predefined Analyses#
In this section, we showcase various predefined analyses available in the log parser. We utilize project files from the following benchmarks: ogs: Constant viscosity (Hydro-Thermal) and for the staggered scheme we use a prj from ogs tests: HT StaggeredCoupling HeatTransportInStationaryFlow
import pandas as pd
from ogstools.logparser import (
analysis_convergence_coupling_iteration,
analysis_convergence_newton_iteration,
analysis_time_step,
fill_ogs_context,
parse_file,
time_step_vs_iterations,
)
from ogstools.logparser.examples import (
const_viscosity_thermal_convection_log,
staggered_log,
)
pd.set_option("display.max_rows", 8) # for visualization only
The preprocessing of logs remains consistent across all examples and thoroughly explained in Advanced topics.
log = const_viscosity_thermal_convection_log
records = parse_file(log)
df_records = pd.DataFrame(records)
df_log = fill_ogs_context(df_records)
Analysis of iterations per time step#
For detailed explanation, refer to: Introduction. (Section: Use predefined analyses)
ogstools.logparser.analysis_time_step
df_ts_it = time_step_vs_iterations(df_log)
df_ts_it
Analysis of computational efficiency by time step#
The resulting table represents performance metrics for different parts of the simulation,
organized by time step. Is utilizes ogstools.logparser.analysis_time_step.
displaying metrics such
as output time [s], step size [s], time step solution time [s], assembly time [s],
Dirichlet time [s], and linear solver time [s].
df_ts = analysis_time_step(df_log)
df_ts = df_ts.loc[0]
# Removing MPI_process (index=0) from result (all are 0) for serial log.
df_ts
Selecting specific metrics (3) and plotting using pandas plot function.
df_ts[["assembly_time", "dirichlet_time", "linear_solver_time"]].plot(
logy=True, grid=True
)
<Axes: xlabel='time_step'>
Analysis of convergence criteria - Newton iterations#
The ogstools.logparser.analysis_convergence_newton_iteration
function allows for the examination of convergence criteria based on
Newton iterations. The resulting table provides convergence metrics for monolithic processes.
For details, refer to the documentation on
<convergence_criterion > defined in in the prj file.
|x| is a norm of a vector of the global component (e.g. pressure, temperature, displacement).
|dx| is the change of a norm of the global component between 2 iteration of non linear solver.
|dx|/|x| is the relative change of a norm of the global component
For this example we had defined in the prj-file:
<convergence_criterion>
<type>DeltaX</type>
<norm_type>NORM2</norm_type>
<abstol>1.e-3</abstol>
</convergence_criterion>
The resulting table contains |x|, |dx| and |dx|/|x| at different time steps, processes and non linear solver iterations.
analysis_convergence_newton_iteration(df_log)
Staggered#
The resulting table provides convergence criteria for staggered coupled processes,
utilizing ogstools.logparser.analysis_convergence_coupling_iteration
Logs are generated from running
ogs benchmark: HeatTransportInStationaryFlow
log = staggered_log
records = parse_file(log)
df_records = pd.DataFrame(records)
df_log = fill_ogs_context(df_records)
# Only for staggered coupled processes !
analysis_convergence_coupling_iteration(df_log)
Total running time of the script: (0 minutes 0.201 seconds)