External API

SCons Extensions

waves.scons_extensions.abaqus_extract(program: str = 'abaqus', **kwargs) → Builder[source]

Abaqus ODB file extraction Builder

This builder executes the odb_extract command line utility against an ODB file in the source list. The ODB file must be the first file in the source list. If there is more than one ODB file in the source list, all but the first file are ignored by odb_extract.

This builder is unique in that no targets are required. The Builder emitter will append the builder managed targets and odb_extract target name constructions automatically. The first target determines the working directory for the emitter targets. If the target(s) must be built in a build subdirectory, e.g. in a parameterized target build, then at least one target must be provided with the build subdirectory, e.g. parameter_set1/target.h5. When in doubt, provide the expected H5 file as a target, e.g. source[0].h5.

The target list may specify an output H5 file name that differs from the ODB file base name as new_name.h5. If the first file in the target list does not contain the *.h5 extension, or if there is no file in the target list, the target list will be prepended with a name matching the ODB file base name and the *.h5 extension.

The builder emitter appends the CSV file created by the abaqus odbreport command as executed by odb_extract unless delete_report_file is set to True.

This builder supports the keyword arguments: output_type, odb_report_args, delete_report_file with behavior as described in the ODB Extract command line interface.

Warning

odb_extract requires Abaqus arguments for odb_report_args in the form of option=value, e.g. step=step_name.

Format of HDF5 file
/                 # Top level group required in all hdf5 files
/<instance name>/ # Groups containing data of each instance found in an odb
    FieldOutputs/      # Group with multiple xarray datasets for each field output
        <field name>/  # Group with datasets containing field output data for a specified set or surface
                       # If no set or surface is specified, the <field name> will be
                       # 'ALL_NODES' or 'ALL_ELEMENTS'
    HistoryOutputs/    # Group with multiple xarray datasets for each history output
        <region name>/ # Group with datasets containing history output data for specified history region name
                       # If no history region name is specified, the <region name> will be 'ALL NODES'
    Mesh/              # Group written from an xarray dataset with all mesh information for this instance
/<instance name>_Assembly/ # Group containing data of assembly instance found in an odb
    Mesh/              # Group written from an xarray dataset with all mesh information for this instance
/odb/             # Catch all group for data found in the odbreport file not already organized by instance
    info/              # Group with datasets that mostly give odb meta-data like name, path, etc.
    jobData/           # Group with datasets that contain additional odb meta-data
    rootAssembly/      # Group with datasets that match odb file organization per Abaqus documentation
    sectionCategories/ # Group with datasets that match odb file organization per Abaqus documentation
/xarray/          # Group with a dataset that lists the location of all data written from xarray datasets
SConstruct
import waves
env = Environment()
env["abaqus"] = waves.scons_extensions.add_program(["abaqus"], env)
env.Append(BUILDERS={"AbaqusExtract": waves.scons_extensions.abaqus_extract()})
env.AbaqusExtract(target=["my_job.h5", "my_job.csv"], source=["my_job.odb"])
Parameters:

program – An absolute path or basename string for the abaqus program

Returns:

Abaqus extract builder

waves.scons_extensions.abaqus_input_scanner() → Scanner[source]

Abaqus input file dependency scanner

Custom SCons scanner that searches for the INPUT= parameter and associated file dependencies inside Abaqus *.inp files.

Returns:

Abaqus input file dependency Scanner

Return type:

SCons.Scanner.Scanner

waves.scons_extensions.abaqus_journal(program: str = 'abaqus', post_action: list = [], **kwargs) → Builder[source]

Abaqus journal file SCons builder

This builder requires that the journal file to execute is the first source in the list. The builder returned by this function accepts all SCons Builder arguments and adds the keyword argument(s):

  • journal_options: The journal file command line options provided as a string.

  • abaqus_options: The Abaqus command line options provided as a string.

At least one target must be specified. The first target determines the working directory for the builder’s action, as shown in the action code snippet below. The action changes the working directory to the first target’s parent directory prior to executing the journal file.

The Builder emitter will append the builder managed targets automatically. Appends target[0].abaqus_v6.env and target[0].stdout to the target list.

The emitter will assume all emitted targets build in the current build directory. If the target(s) must be built in a build subdirectory, e.g. in a parameterized target build, then the first target must be provided with the build subdirectory, e.g. parameter_set1/my_target.ext. When in doubt, provide a STDOUT redirect file as a target, e.g. target.stdout.

Abaqus journal builder action
cd ${TARGET.dir.abspath} && abaqus cae -noGui ${SOURCE.abspath} ${abaqus_options} -- ${journal_options} > ${TARGETS[-1].abspath} 2>&1
SConstruct
import waves
env = Environment()
env["abaqus"] = waves.scons_extensions.add_program(["abaqus"], env)
env.Append(BUILDERS={"AbaqusJournal": waves.scons_extensions.abaqus_journal()})
env.AbaqusJournal(target=["my_journal.cae"], source=["my_journal.py"], journal_options="")
Parameters:

  • program (str) – An absolute path or basename string for the abaqus program.

  • post_action (list) – List of shell command string(s) to append to the builder’s action list. Implemented to allow post target modification or introspection, e.g. inspect the Abaqus log for error keywords and throw a non-zero exit code even if Abaqus does not. Builder keyword variables are available for substitution in the post_action action using the ${} syntax. Actions are executed in the first target’s directory as cd ${TARGET.dir.abspath} && ${post_action}

Returns:

Abaqus journal builder

Return type:

SCons.Builder.Builder

waves.scons_extensions.abaqus_solver(program: str = 'abaqus', post_action: list[str] = [], emitter: Literal['standard', 'explicit', 'datacheck'] | None = None, **kwargs) → Builder[source]

Abaqus solver SCons builder

This builder requires that the root input file is the first source in the list. The builder returned by this function accepts all SCons Builder arguments and adds the keyword argument(s):

  • job_name: The job name string. If not specified job_name defaults to the root input file stem. The Builder emitter will append common Abaqus output files as targets automatically from the job_name, e.g. job_name.odb.

  • abaqus_options: The Abaqus command line options provided as a string.

  • suffixes: override the emitter targets with a new list of extensions, e.g. AbaqusSolver(target=[], source=["input.inp"], suffixes=[".odb"]) will emit only one file named job_name.odb.

The first target determines the working directory for the builder’s action, as shown in the action code snippet below. The action changes the working directory to the first target’s parent directory prior to executing the journal file.

This builder is unique in that no targets are required. The Builder emitter will append the builder managed targets automatically. The target list only appends those extensions which are common to Abaqus analysis operations. Some extensions may need to be added explicitly according to the Abaqus simulation solver, type, or options. If you find that SCons isn’t automatically cleaning some Abaqus output files, they are not in the automatically appended target list.

The emitter will assume all emitted targets build in the current build directory. If the target(s) must be built in a build subdirectory, e.g. in a parameterized target build, then the first target must be provided with the build subdirectory, e.g. parameter_set1/job_name.odb. When in doubt, provide a STDOUT redirect file as a target, e.g. target.stdout.

The -interactive option is always appended to the builder action to avoid exiting the Abaqus task before the simulation is complete. The -ask_delete no option is always appended to the builder action to overwrite existing files in programmatic execution, where it is assumed that the Abaqus solver target(s) should be re-built when their source files change.

SConstruct
import waves
env = Environment()
env["abaqus"] = waves.scons_extensions.add_program(["abaqus"], env)
env.Append(BUILDERS={
    "AbaqusSolver": waves.scons_extensions.abaqus_solver(),
    "AbaqusStandard": waves.scons_extensions.abaqus_solver(emitter='standard'),
    "AbaqusOld": waves.scons_extensions.abaqus_solver(program="abq2019"),
    "AbaqusPost": waves.scons_extensions.abaqus_solver(post_action="grep -E '\<SUCCESSFULLY' ${job_name}.sta")
})
env.AbaqusSolver(target=[], source=["input.inp"], job_name="my_job", abaqus_options="-cpus 4")
env.AbaqusSolver(target=[], source=["input.inp"], job_name="my_job", suffixes=[".odb"])
Abaqus solver builder action
cd ${TARGET.dir.abspath} && ${program} -job ${job_name} -input ${SOURCE.filebase} ${abaqus_options} -interactive -ask_delete no > ${TARGETS[-1].abspath} 2>&1
Parameters:

  • program – An absolute path or basename string for the abaqus program

  • post_action – List of shell command string(s) to append to the builder’s action list. Implemented to allow post target modification or introspection, e.g. inspect the Abaqus log for error keywords and throw a non-zero exit code even if Abaqus does not. Builder keyword variables are available for substitution in the post_action action using the ${} syntax. Actions are executed in the first target’s directory as cd ${TARGET.dir.abspath} && ${post_action}.

  • emitter –

    emit file extensions based on the value of this variable. Overridden by the suffixes keyword argument that may be provided in the Task definition.

    • ”standard”: [“.odb”, “.dat”, “.msg”, “.com”, “.prt”, “.sta”]

    • ”explicit”: [“.odb”, “.dat”, “.msg”, “.com”, “.prt”, “.sta”]

    • ”datacheck”: [“.odb”, “.dat”, “.msg”, “.com”, “.prt”, “.023”, “.mdl”, “.sim”, “.stt”]

    • default value: [“.odb”, “.dat”, “.msg”, “.com”, “.prt”]

Returns:

Abaqus solver builder

waves.scons_extensions.add_cubit(names: list[str], env) → str[source]

Modifies environment variables with the paths required to import cubit in a Python3 environment.

Returns the absolute path of the first program name found. Appends PATH with first program’s parent directory if a program is found and the directory is not already on PATH. Prepends PYTHONPATH with parent/bin. Prepends LD_LIBRARY_PATH with parent/bin/python3.

Returns None if no program name is found.

Example Cubit environment modification
import waves

env = Environment()
env["cubit"] = waves.scons_extensions.add_cubit(["cubit"], env)
Parameters:

  • names – list of string program names. May include an absolute path.

  • env (SCons.Script.SConscript.SConsEnvironment) – The SCons construction environment object to modify

Returns:

Absolute path of the found program. None if none of the names are found.

waves.scons_extensions.add_program(names: list[str], env) → str[source]

Search for a program from a list of possible program names. Add first found to system PATH.

Returns the absolute path of the first program name found. Appends PATH with first program’s parent directory if a program is found and the directory is not already on PATH. Returns None if no program name is found.

Example search for an executable named “program”
import waves

env = Environment()
env["program"] = waves.scons_extensions.add_program(["program"], env)
Parameters:

  • names – list of string program names. May include an absolute path.

  • env (SCons.Script.SConscript.SConsEnvironment) – The SCons construction environment object to modify

Returns:

Absolute path of the found program. None if none of the names are found.

waves.scons_extensions.alias_list_message(env=None, append: bool = True, keep_local: bool = True) → None[source]

Add the alias list to the project’s help message

See the SCons Help documentation for appending behavior. Adds text to the project help message formatted as

Target Aliases:
    Alias_1
    Alias_2

where the aliases are recovered from SCons.Node.Alias.default_ans.

Parameters:

  • env (SCons.Script.SConscript.SConsEnvironment) – The SCons construction environment object to modify

  • append – append to the env.Help message (default). When False, the env.Help message will be overwritten if env.Help has not been previously called.

  • keep_local – Limit help message to the project specific content when True. Only applies to SCons >=4.6.0

waves.scons_extensions.append_env_path(program: str, env) → None[source]

Append SCons contruction environment PATH with the program’s parent directory

Raises a FileNotFoundError if the program absolute path does not exist. Uses the SCons AppendENVPath method. If the program parent directory is already on PATH, the PATH directory order is preserved.

Example environment modification
import waves

env = Environment()
env["program"] = waves.scons_extensions.find_program(["program"], env)
if env["program"]:
    waves.append_env_path(env["program"], env)
Parameters:

  • program – An absolute path for the program to add to SCons construction environment PATH

  • env (SCons.Script.SConscript.SConsEnvironment) – The SCons construction environment object to modify

waves.scons_extensions.catenate_actions(**outer_kwargs)[source]

Decorator factory to apply the catenate_builder_actions to a function that returns an SCons Builder.

Accepts the same keyword arguments as the waves.scons_extensions.catenate_builder_actions()

import SCons.Builder
import waves

@waves.scons_extensions.catenate_actions
def my_builder():
    return SCons.Builder.Builder(action=["echo $SOURCE > $TARGET", "echo $SOURCE >> $TARGET"])
waves.scons_extensions.catenate_builder_actions(builder: Builder, program: str = '', options: str = '') → Builder[source]

Catenate a builder’s arguments and prepend the program and options

${program} ${options} "action one && action two"
Parameters:

  • builder – The SCons builder to modify

  • program – wrapping executable

  • options – options for the wrapping executable

Returns:

modified builder

waves.scons_extensions.conda_environment() → Builder[source]

Create a Conda environment file with conda env export

This builder is intended to help WAVES workflows document the Conda environment used in the current build. At least one target file must be specified for the conda env export --file ${TARGET} output. Additional options to the Conda env export subcommand may be passed as the builder keyword argument conda_env_export_options.

At least one target must be specified. The first target determines the working directory for the builder’s action, as shown in the action code snippet below. The action changes the working directory to the first target’s parent directory prior to creating the Conda environment file.

Conda environment builder action
cd ${TARGET.dir.abspath} && conda env export ${conda_env_export_options} --file ${TARGET.file}

The modsim owner may choose to re-use this builder throughout their project configuration to provide various levels of granularity in the recorded Conda environment state. It’s recommended to include this builder at least once for any workflows that also use the waves.scons_extensions.python_builder(). The builder may be re-used once per build sub-directory to provide more granular build environment reproducibility in the event that sub-builds are run at different times with variations in the active Conda environment. For per-Python script task environment reproducibility, the builder source list can be linked to the output of a waves.scons_extensions.python_builder() task with a target environment file name to match.

The first recommendation, always building the project wide Conda environment file, is demonstrated in the example usage below.

SConstruct
import waves
env = Environment()
env.Append(BUILDERS={"CondaEnvironment": waves.scons_extensions.conda_environment()})
environment_target = env.CondaEnvironment(target=["environment.yaml"])
env.AlwaysBuild(environment_target)
Returns:

Conda environment builder

Return type:

SCons.Builder.Builder

waves.scons_extensions.construct_action_list(actions: list[str], prefix: str = 'cd ${TARGET.dir.abspath} &&', postfix: str = '') → list[str][source]

Return an action list with a common pre/post-fix

Returns the constructed action list with pre/post fix strings as

f"{prefix} {new_action} {postfix}"

where SCons action objects are converted to their string representation. If a string is passed instead of a list, it is first converted to a list. If an empty list is passed, and empty list is returned.

Parameters:

  • actions – List of action strings

  • prefix – Common prefix to prepend to each action

  • postfix – Common postfix to append to each action

Returns:

action list

waves.scons_extensions.copy_substitute(source_list: list, substitution_dictionary: dict | None = None, env: ~SCons.Environment.Base = <SCons.Environment.Base object>, build_subdirectory: str = '.', symlink: bool = False) → NodeList[source]

Copy source list to current variant directory and perform template substitutions on *.in filenames

Warning

This is a Python function and not an SCons builder. It cannot be added to the construction environment BUILDERS list. The function returns a list of targets instead of a Builder object.

Creates an SCons Copy task for each source file. Files are copied to the current variant directory matching the calling SConscript parent directory. Files with the name convention *.in are also given an SCons Substfile Builder, which will perform template substitution with the provided dictionary in-place in the current variant directory and remove the .in suffix.

To avoid dependency cycles, the source file(s) should be passed by absolute path.

SConstruct
import pathlib
import waves
env = Environment()
current_directory = pathlib.Path(Dir(".").abspath)
source_list = [
    "#/subdir3/file_three.ext",              # File found with respect to project root directory using SCons notation
    current_directory / file_one.ext,        # File found in current SConscript directory
    current_directory / "subdir2/file_two",  # File found below current SConscript directory
    current_directory / "file_four.ext.in"   # File with substitutions matching substitution dictionary keys
]
substitution_dictionary = {
    "@variable_one@": "value_one"
}
waves.scons_extensions.copy_substitute(source_list, substitution_dictionary, env)
Parameters:

  • source_list – List of pathlike objects or strings. Will be converted to list of pathlib.Path objects.

  • substitution_dictionary – key: value pairs for template substitution. The keys must contain the optional template characters if present, e.g. @variable@. The template character, e.g. @, can be anything that works in the SCons Substfile builder.

  • env – An SCons construction environment to use when defining the targets.

  • build_subdirectory – build subdirectory relative path prepended to target files

  • symlink – Whether symbolic links are created as new symbolic links. If true, symbolic links are shallow copies as a new symbolic link. If false, symbolic links are copied as a new file (dereferenced).

Returns:

SCons NodeList of Copy and Substfile target nodes

waves.scons_extensions.default_targets_message(env=None, append: bool = True, keep_local: bool = True) → None[source]

Add a default targets list to the project’s help message

See the SCons Help documentation for appending behavior. Adds text to the project help message formatted as

Default Targets:
    Default_Target_1
    Default_Target_2

where the targets are recovered from SCons.Script.DEFAULT_TARGETS.

Parameters:

  • env (SCons.Script.SConscript.SConsEnvironment) – The SCons construction environment object to modify

  • append – append to the env.Help message (default). When False, the env.Help message will be overwritten if env.Help has not been previously called.

  • keep_local – Limit help message to the project specific content when True. Only applies to SCons >=4.6.0

waves.scons_extensions.find_program(names: list[str], env) → str[source]

Search for a program from a list of possible program names.

Returns the absolute path of the first program name found. If path parts contain spaces, the part will be wrapped in double quotes.

Parameters:

  • names – list of string program names. May include an absolute path.

  • env (SCons.Script.SConscript.SConsEnvironment) – The SCons construction environment object to modify

Returns:

Absolute path of the found program. None if none of the names are found.

waves.scons_extensions.matlab_script(program: str = 'matlab', post_action: list[str] = [], **kwargs) → Builder[source]

Matlab script SCons builder

Warning

Experimental implementation is subject to change

This builder requires that the Matlab script is the first source in the list. The builder returned by this function accepts all SCons Builder arguments and adds the keyword argument(s):

  • script_options: The Matlab function interface options in Matlab syntax and provided as a string.

  • matlab_options: The Matlab command line options provided as a string.

The parent directory absolute path is added to the Matlab path variable prior to execution. All required Matlab files should be co-located in the same source directory.

At least one target must be specified. The first target determines the working directory for the builder’s action, as shown in the action code snippet below. The action changes the working directory to the first target’s parent directory prior to executing the python script.

The Builder emitter will append the builder managed targets automatically. Appends target[0].matlab.env and ``target[0].stdout to the target list.

The emitter will assume all emitted targets build in the current build directory. If the target(s) must be built in a build subdirectory, e.g. in a parameterized target build, then the first target must be provided with the build subdirectory, e.g. parameter_set1/my_target.ext. When in doubt, provide a STDOUT redirect file as a target, e.g. target.stdout.

Matlab script builder action
cd ${TARGET.dir.abspath} && {program} ${matlab_options} -batch "path(path, '${SOURCE.dir.abspath}'); ${SOURCE.filebase}(${script_options})" > ${TARGETS[-1].abspath} 2>&1
Parameters:

  • program – An absolute path or basename string for the Matlab program.

  • post_action – List of shell command string(s) to append to the builder’s action list. Implemented to allow post target modification or introspection, e.g. inspect a log for error keywords and throw a non-zero exit code even if Matlab does not. Builder keyword variables are available for substitution in the post_action action using the ${} syntax. Actions are executed in the first target’s directory as cd ${TARGET.dir.abspath} && ${post_action}

Returns:

Matlab script builder

waves.scons_extensions.print_build_failures(print_stdout: bool = True) → None[source]

On exit, query the SCons reported build failures and print the associated node’s STDOUT file, if it exists

Parameters:

print_stdout – Boolean to set the exit behavior. If False, don’t modify the exit behavior.

waves.scons_extensions.project_help_message(env=None, append: bool = True, keep_local: bool = True) → None[source]

Add default targets and alias lists to project help message

See the SCons Help documentation for appending behavior. Thin wrapper around

Parameters:

  • env (SCons.Script.SConscript.SConsEnvironment) – The SCons construction environment object to modify

  • append – append to the env.Help message (default). When False, the env.Help message will be overwritten if env.Help has not been previously called.

  • keep_local – Limit help message to the project specific content when True. Only applies to SCons >=4.6.0

waves.scons_extensions.python_script(post_action: list[str] = []) → Builder[source]

Python script SCons builder

This builder requires that the python script to execute is the first source in the list. The builder returned by this function accepts all SCons Builder arguments and adds the keyword argument(s):

  • script_options: The Python script command line arguments provided as a string.

  • python_options: The Python command line arguments provided as a string.

At least one target must be specified. The first target determines the working directory for the builder’s action, as shown in the action code snippet below. The action changes the working directory to the first target’s parent directory prior to executing the python script.

The Builder emitter will append the builder managed targets automatically. Appends target[0].stdout to the target list.

The emitter will assume all emitted targets build in the current build directory. If the target(s) must be built in a build subdirectory, e.g. in a parameterized target build, then the first target must be provided with the build subdirectory, e.g. parameter_set1/my_target.ext. When in doubt, provide a STDOUT redirect file as a target, e.g. target.stdout.

Python script builder action
cd ${TARGET.dir.abspath} && python ${python_options} ${SOURCE.abspath} ${script_options} > ${TARGETS[-1].abspath} 2>&1
SConstruct
import waves
env = Environment()
env.Append(BUILDERS={"PythonScript": waves.scons_extensions.python_script()})
env.PythonScript(target=["my_output.stdout"], source=["my_script.py"], python_options="", script_options="")
Parameters:

post_action (list) – List of shell command string(s) to append to the builder’s action list. Implemented to allow post target modification or introspection, e.g. inspect a log for error keywords and throw a non-zero exit code even if Python does not. Builder keyword variables are available for substitution in the post_action action using the ${} syntax. Actions are executed in the first target’s directory as cd ${TARGET.dir.abspath} && ${post_action}

Returns:

Python script builder

Return type:

SCons.Builder.Builder

waves.scons_extensions.quinoa_solver(charmrun: str = 'charmrun', inciter: str = 'inciter', charmrun_options: str = '+p1', inciter_options: str = '', prefix_command: str = '', post_action: list[str] = []) → Builder[source]

Quinoa solver SCons builder

This builder requires at least two source files provided in the order

  1. Quinoa control file: *.q

  2. Exodus mesh file: *.exo

The builder returned by this function accepts all SCons Builder arguments. Except for the post_action, the arguments of this function are also available as keyword arguments of the builder. When provided during task definition, the keyword arguments override the builder returned by this function.

The first target determines the working directory for the builder’s action, as shown in the action code snippet below. The action changes the working directory to the first target’s parent directory prior to executing quinoa.

The emitter will assume all emitted targets build in the current build directory. If the target(s) must be built in a build subdirectory, e.g. in a parameterized target build, then the first target must be provided with the build subdirectory, e.g. parameter_set1/target.ext. When in doubt, provide a STDOUT redirect file as a target, e.g. target.stdout.

Warning

This is an experimental builder for Quinoa support. The only emitted file is the target[0].stdout redirected STDOUT and STDERR file. All relevant application output files, e.g. out.* must be specified in the target list.

SConstruct
import waves
env = waves.scons_extensions.shell_environment("module load quinoa")
env.Append(BUILDERS={
    "QuinoaSolver": waves.scons_extensions.quinoa_solver(charmrun_options="+p1"),
})
# Serial execution with "+p1"
env.QuinoaSolver(target=["flow.stdout"], source=["flow.q", "box.exo"])
# Parallel execution with "+p4"
env.QuinoaSolver(target=["flow.stdout"], source=["flow.q", "box.exo"], charmrun_options="+p4")
Quinoa builder action
${prefix_command} ${TARGET.dir.abspath} && ${charmrun} ${charmrun_options} ${inciter} ${inciter_options} --control ${SOURCES[0].abspath} --input ${SOURCES[1].abspath} > ${TARGETS[-1].abspath} 2>&1
Parameters:

  • charmrun – The relative or absolute path to the charmrun executable

  • charmrun_options – The charmrun command line interface options

  • inciter – The relative or absolute path to the inciter (quinoa) executable

  • inciter_options – The inciter (quinoa executable) command line interface options

  • prefix_command – Optional prefix command intended for environment preparation. Primarily intended for use with waves.scons_extensions.sbatch_quinoa_solver() or when wrapping the builder with waves.scons_extensions.ssh_builder_actions(). For local, direct execution, user’s should prefer to create an SCons construction environment with waves.scons_extensions.shell_environment(). When overriding in a task definition, the prefix command must end with ' &&'.

  • post_action – List of shell command string(s) to append to the builder’s action list. Implemented to allow post target modification or introspection, e.g. inspect the Abaqus log for error keywords and throw a non-zero exit code even if Abaqus does not. Builder keyword variables are available for substitution in the post_action action using the ${} syntax. Actions are executed in the first target’s directory as cd ${TARGET.dir.abspath} && ${post_action}

Returns:

Quinoa builder

waves.scons_extensions.sbatch(program: str = 'sbatch', post_action: list[str] = [], **kwargs) → Builder[source]

SLURM sbatch SCons builder

The builder does not use a SLURM batch script. Instead, it requires the slurm_job variable to be defined with the command string to execute.

At least one target must be specified. The first target determines the working directory for the builder’s action, as shown in the action code snippet below. The action changes the working directory to the first target’s parent directory prior to executing the journal file.

The Builder emitter will append the builder managed targets automatically. Appends target[0].stdout to the target list.

SLURM sbatch builder action
cd ${TARGET.dir.abspath} && sbatch --wait --output=${TARGETS[-1].abspath} ${sbatch_options} --wrap ${slurm_job}
SConstruct
import waves
env = Environment()
env.Append(BUILDERS={"SlurmSbatch": waves.scons_extensions.sbatch()})
env.SlurmSbatch(target=["my_output.stdout"], source=["my_source.input"], slurm_job="cat $SOURCE > $TARGET")
Parameters:

  • program – An absolute path or basename string for the sbatch program.

  • post_action – List of shell command string(s) to append to the builder’s action list. Implemented to allow post target modification or introspection, e.g. inspect the Abaqus log for error keywords and throw a non-zero exit code even if Abaqus does not. Builder keyword variables are available for substitution in the post_action action using the ${} syntax. Actions are executed in the first target’s directory as cd ${TARGET.dir.abspath} && ${post_action}

Returns:

SLURM sbatch builder

waves.scons_extensions.sbatch_abaqus_journal(*args, **kwargs)[source]

Thin pass through wrapper of waves.scons_extensions.abaqus_journal()

Catenate the actions and submit with SLURM sbatch. Accepts the sbatch_options builder keyword argument to modify sbatch behavior.

Sbatch Abaqus journal builder action
sbatch --wait --output=${TARGET.base}.slurm.out ${sbatch_options} --wrap "cd ${TARGET.dir.abspath} && abaqus cae -noGui ${SOURCE.abspath} ${abaqus_options} -- ${journal_options} > ${TARGETS[-1].abspath} 2>&1"
waves.scons_extensions.sbatch_abaqus_solver(*args, **kwargs)[source]

Thin pass through wrapper of waves.scons_extensions.abaqus_solver()

Catenate the actions and submit with SLURM sbatch. Accepts the sbatch_options builder keyword argument to modify sbatch behavior.

Sbatch Abaqus solver builder action
sbatch --wait --output=${TARGET.base}.slurm.out ${sbatch_options} --wrap "cd ${TARGET.dir.abspath} && ${program} -job ${job_name} -input ${SOURCE.filebase} ${abaqus_options} -interactive -ask_delete no > ${TARGETS[-1].abspath} 2>&1"
waves.scons_extensions.sbatch_python_script(*args, **kwargs)[source]

Thin pass through wrapper of waves.scons_extensions.python_script()

Catenate the actions and submit with SLURM sbatch. Accepts the sbatch_options builder keyword argument to modify sbatch behavior.

Sbatch Python script builder action
sbatch --wait --output=${TARGET.base}.slurm.out ${sbatch_options} --wrap "cd ${TARGET.dir.abspath} && python ${python_options} ${SOURCE.abspath} ${script_options} > ${TARGETS[-1].abspath} 2>&1"
waves.scons_extensions.sbatch_quinoa_solver(*args, **kwargs)[source]

Thin pass through wrapper of waves.scons_extensions.quinoa_solver()

Catenate the actions and submit with SLURM sbatch. Accepts the sbatch_options builder keyword argument to modify sbatch behavior.

Sbatch Quinoa solver builder action
sbatch --wait --output=${TARGET.base}.slurm.out ${sbatch_options} --wrap ""
waves.scons_extensions.sbatch_sierra(*args, **kwargs)[source]

Thin pass through wrapper of waves.scons_extensions.sierra()

Catenate the actions and submit with SLURM sbatch. Accepts the sbatch_options builder keyword argument to modify sbatch behavior.

sbatch Sierra builder action
sbatch --wait --output=${TARGET.base}.slurm.out ${sbatch_options} --wrap "cd ${TARGET.dir.abspath} && ${program} ${sierra_options} ${application} ${application_options} -i ${SOURCE.file} > ${TARGETS[-1].abspath} 2>&1"
waves.scons_extensions.shell_environment(command: str, cache: str | None = None, overwrite_cache: bool = False) → Base[source]

Return an SCons shell environment from a cached file or by running a shell command

If the environment is created successfully and a cache file is requested, the cache file is _always_ written. The overwrite_cache behavior forces the shell command execution, even when the cache file is present.

Warning

Currently only supports bash shells

SConstruct
import waves
env = waves.scons_extensions.shell_environment("source my_script.sh")
Parameters:

  • command – the shell command to execute

  • cache – absolute or relative path to read/write a shell environment dictionary. Will be written as YAML formatted file regardless of extension.

  • overwrite_cache – Ignore previously cached files if they exist.

Returns:

SCons shell environment

waves.scons_extensions.sierra(program: str = 'sierra', application: str = 'adagio', post_action: list[str] = []) → Builder[source]

Sierra SCons builder

This builder requires that the root input file is the first source in the list. The builder returned by this function accepts all SCons Builder arguments and adds the keyword argument(s):

  • sierra_options: The Sierra command line options provided as a string.

  • application_options: The application (e.g. adagio) command line options provided as a string.

The first target determines the working directory for the builder’s action, as shown in the action code snippet below. The action changes the working directory to the first target’s parent directory prior to executing sierra.

The emitter will assume all emitted targets build in the current build directory. If the target(s) must be built in a build subdirectory, e.g. in a parameterized target build, then the first target must be provided with the build subdirectory, e.g. parameter_set1/target.ext. When in doubt, provide a STDOUT redirect file as a target, e.g. target.stdout.

Warning

This is an experimental builder for Sierra support. The only emitted file is the application’s version report in target[0].env and the target[0].stdout redirected STDOUT and STDERR file. All relevant application output files, e.g. genesis_output.e must be specified in the target list.

SConstruct
import waves
env = waves.scons_extensions.shell_environment("module load sierra")
env.Append(BUILDERS={
    "Sierra": waves.scons_extensions.sierra(),
})
env.Sierra(target=["output.e"], source=["input.i"])
Sierra builder action
cd ${TARGET.dir.abspath} && ${program} ${sierra_options} ${application} ${application_options} -i ${SOURCE.file} > ${TARGETS[-1].abspath} 2>&1
Parameters:

  • program – An absolute path or basename string for the Sierra program

  • application – The string name for the Sierra application

  • post_action – List of shell command string(s) to append to the builder’s action list. Implemented to allow post target modification or introspection, e.g. inspect the Sierra log for error keywords and throw a non-zero exit code even if Sierra does not. Builder keyword variables are available for substitution in the post_action action using the ${} syntax. Actions are executed in the first target’s directory as cd ${TARGET.dir.abspath} && ${post_action}.

Returns:

Sierra builder

waves.scons_extensions.sphinx_build(program: str = 'sphinx-build', options: str = '', builder: str = 'html', tags: str = '') → Builder[source]

Sphinx builder using the -b specifier

This builder does not have an emitter. It requires at least one target.

action
${program} ${options} -b ${builder} ${TARGET.dir.dir.abspath} ${TARGET.dir.abspath} ${tags}
SConstruct
import waves
env = Environment()
env.Append(BUILDERS={
    "SphinxBuild": waves.scons_extensions.sphinx_build(options="-W"),
})
sources = ["conf.py", "index.rst"]
targets = ["html/index.html"]
html = env.SphinxBuild(
    target=targets,
    source=sources,
)
env.Clean(html, [Dir("html")] + sources)
env.Alias("html", html)
Parameters:

  • program – sphinx-build executable

  • options – sphinx-build options

  • builder – builder name. See the Sphinx documentation for options

  • tags – sphinx-build tags

Returns:

Sphinx builder

waves.scons_extensions.sphinx_latexpdf(program: str = 'sphinx-build', options: str = '', builder: str = 'latexpdf', tags: str = '') → Builder[source]

Sphinx builder using the -M specifier. Intended for latexpdf builds.

This builder does not have an emitter. It requires at least one target.

action
${program} -M ${builder} ${TARGET.dir.dir.abspath} ${TARGET.dir.dir.abspath} ${tags} ${options}"
SConstruct
import waves
env = Environment()
env.Append(BUILDERS={
    "SphinxPDF": waves.scons_extensions.sphinx_latexpdf(options="-W"),
})
sources = ["conf.py", "index.rst"]
targets = ["latex/project.pdf"]
latexpdf = env.SphinxBuild(
    target=targets,
    source=sources,
)
env.Clean(latexpdf, [Dir("latex")] + sources)
env.Alias("latexpdf", latexpdf)
Parameters:

  • program (str) – sphinx-build executable

  • options (str) – sphinx-build options

  • builder (str) – builder name. See the Sphinx documentation for options

  • tags (str) – sphinx-build tags

Returns:

Sphinx latexpdf builder

waves.scons_extensions.sphinx_scanner() → Scanner[source]

SCons scanner that searches for directives

  • .. include::

  • .. literalinclude::

  • .. image::

  • .. figure::

  • .. bibliography::

inside .rst and .txt files

Returns:

Abaqus input file dependency Scanner

Return type:

SCons.Scanner.Scanner

waves.scons_extensions.ssh_builder_actions(builder: Builder, remote_server: str = '${remote_server}', remote_directory: str = '${remote_directory}') → Builder[source]

Wrap a builder’s action list with remote copy operations and ssh commands

By default, the remote server and remote directory strings are written to accept (and require) task-by-task overrides via task keyword arguments. Any SCons replacement string patterns, ${variable}, will make that variable a required task keyword argument. Only if the remote server and/or remote directory are known to be constant across all possible tasks should those variables be overridden with a string literal containing no ${variable} SCons keyword replacement patterns.

Warning

The waves.scons_extensions.ssh_builder_actions() is a work-in-progress solution with some assumptions specific to the action construction used by WAVES. It _should_ work for most basic builders, but adapting this function to users’ custom builders will probably require some advanced SCons knowledge and inspection of the waves.scons_extensions_ssh_builder_actions() implementation.

Design assumptions

  • Creates the remote_directory with mkdir -p. mkdir must exist on the remote_server.

  • Copies all source files to a flat remote_directory with rsync -rlptv. rsync must exist on the local system.

  • Replaces instances of cd ${TARGET.dir.abspath} && with cd ${remote_directory} && in the original builder actions.

  • Replaces instances of SOURCE.abspath or SOURCES.abspath with SOURCE[S].file in the original builder actions.

  • Prefixes all original builder actions with cd ${remote_directory} &&.

  • All original builder actions are wrapped in single quotes as '{original action}' to preserve the && as part of the remote_server command. Shell variables, e.g. $USER, will not be expanded on the remote_server. If quotes are included in the original builder actions, they should be double quotes.

  • Returns the entire remote_directory to the original builder ${TARGET.dir.abspath} with rysnc. rsync must exist on the local system.

SConstruct
import getpass
import waves
user = getpass.getuser()
env = Environment()
env.Append(BUILDERS={
    "SSHAbaqusSolver": waves.scons_extensions.ssh_builder_actions(
        waves.scons_extensions.abaqus_solver(program="/remote/server/installation/path/of/abaqus"),
        remote_server="myserver.mydomain.com"
    )
})
env.SSHAbaqusSolver(target=["myjob.sta"], source=["input.inp"], job_name="myjob", abaqus_options="-cpus 4",
                    remote_directory="/scratch/${user}/myproject/myworkflow", user=user)
my_package.py
import SCons.Builder
import waves

def print_builder_actions(builder):
    for action in builder.action.list:
        print(action.cmd_list)

def cat(program="cat"):
    return SCons.Builder.Builder(action=
        [f"{program} ${{SOURCES.abspath}} | tee ${{TARGETS.file}}", "echo \"Hello World!\""]
    )

build_cat = cat()

ssh_build_cat = waves.scons_extensions.ssh_builder_actions(
    cat(), remote_server="myserver.mydomain.com", remote_directory="/scratch/roppenheimer/ssh_wrapper"
)

>>> import my_package
>>> my_package.print_builder_actions(my_package.build_cat)
cat ${SOURCES.abspath} | tee ${TARGETS.file}
echo "Hello World!"
>>> my_package.print_builder_actions(my_package.ssh_build_cat)
ssh myserver.mydomain.com "mkdir -p /scratch/roppenheimer/ssh_wrapper"
rsync -rlptv ${SOURCES.abspath} myserver.mydomain.com:/scratch/roppenheimer/ssh_wrapper
ssh myserver.mydomain.com 'cd /scratch/roppenheimer/ssh_wrapper && cat ${SOURCES.file} | tee ${TARGETS.file}'
ssh myserver.mydomain.com 'cd /scratch/roppenheimer/ssh_wrapper && echo "Hello World!"'
rsync -rltpv myserver.mydomain.com:/scratch/roppenheimer/ssh_wrapper/ ${TARGET.dir.abspath}
Parameters:

  • builder – The SCons builder to modify

  • remote_server – remote server where the original builder’s actions should be executed. The default string requires every task to specify a matching keyword argument string.

  • remote_directory – absolute or relative path where the original builder’s actions should be executed. The default string requires every task to specify a matching keyword argument string.

Returns:

modified builder

waves.scons_extensions.substitution_syntax(substitution_dictionary: dict, prefix: str = '@', postfix: str = '@') → dict[source]

Return a dictionary copy with the pre/postfix added to the key strings

Assumes a flat dictionary with keys of type str. Keys that aren’t strings will be converted to their string representation. Nested dictionaries can be supplied, but only the first layer keys will be modified. Dictionary values are unchanged.

Parameters:

  • substitution_dictionary (dict) – Original dictionary to copy

  • prefix (string) – String to prepend to all dictionary keys

  • postfix (string) – String to append to all dictionary keys

Returns:

Copy of the dictionary with key strings modified by the pre/posfix

Parameter Generators

class waves.parameter_generators.CartesianProduct(parameter_schema: dict, output_file_template: str = None, output_file: str = None, output_file_type: str = 'yaml', set_name_template: str = 'parameter_set@number', previous_parameter_study: str = None, overwrite: bool = False, dryrun: bool = False, write_meta: bool = False, **kwargs)[source]

Bases: _ParameterGenerator

Builds a cartesian product parameter study

Parameters:

  • parameter_schema (dict) – The YAML loaded parameter study schema dictionary - {parameter_name: schema value} CartesianProduct expects “schema value” to be an iterable. For example, when read from a YAML file “schema value” will be a Python list.

  • output_file_template (str) – Output file name template. Required if parameter sets will be written to files instead of printed to STDOUT. May contain pathseps for an absolute or relative path template. May contain the @number set number placeholder in the file basename but not in the path. If the placeholder is not found it will be appended to the template string.

  • output_file (str) – Output file name for a single file output of the parameter study. May contain pathseps for an absolute or relative path. output_file and output_file_template are mutually exclusive. Output file is always overwritten.

  • output_file_type (str) – Output file syntax or type. Options are: ‘yaml’, ‘h5’.

  • set_name_template (str) – Parameter set name template. Overridden by output_file_template, if provided.

  • previous_parameter_study (str) – A relative or absolute file path to a previously created parameter study Xarray Dataset

  • overwrite (bool) – Overwrite existing output files

  • dryrun (bool) – Print contents of new parameter study output files to STDOUT and exit

  • write_meta (bool) – Write a meta file named “parameter_study_meta.txt” containing the parameter set file names. Useful for command line execution with build systems that require an explicit file list for target creation.

Example

>>> import waves
>>> parameter_schema = {
...     'parameter_1': [1, 2],
...     'parameter_2': ['a', 'b']
... }
>>> parameter_generator = waves.parameter_generators.CartesianProduct(parameter_schema)
>>> print(parameter_generator.parameter_study)
<xarray.Dataset>
Dimensions:             (data_type: 1, parameter_set_hash: 4)
Coordinates:
  * data_type           (data_type) object 'samples'
    parameter_set_hash  (parameter_set_hash) <U32 'de3cb3eaecb767ff63973820b2...
  * parameter_sets      (parameter_set_hash) <U14 'parameter_set0' ... 'param...
Data variables:
    parameter_1         (data_type, parameter_set_hash) object 1 1 2 2
    parameter_2         (data_type, parameter_set_hash) object 'a' 'b' 'a' 'b'
Variables:

self.parameter_study – The final parameter study XArray Dataset object

parameter_study_to_dict(*args, **kwargs) → dict[source]

Return parameter study as a dictionary

Used for iterating on parameter sets in an SCons workflow with parameter substitution dictionaries, e.g.

>>> for set_name, parameters in parameter_generator.parameter_study_to_dict().items():
...     print(f"{set_name}: {parameters}")
...
parameter_set0: {'parameter_1': 1, 'parameter_2': 'a'}
parameter_set1: {'parameter_1': 1, 'parameter_2': 'b'}
parameter_set2: {'parameter_1': 2, 'parameter_2': 'a'}
parameter_set3: {'parameter_1': 2, 'parameter_2': 'b'}
Parameters:

data_type (str) – The data_type selection to return - samples or quantiles

Returns:

parameter study sets and samples as a dictionary: {set_name: {parameter: value}, …}

Return type:

dict - {str: {str: value}}

write() → None[source]

Write the parameter study to STDOUT or an output file.

Writes to STDOUT by default. Requires non-default output_file_template or output_file specification to write to files.

If printing to STDOUT, print all parameter sets together. If printing to files, overwrite when contents of existing files have changed. If overwrite is specified, overwrite all parameter set files. If a dry run is requested print file-content associations for files that would have been written.

Writes parameter set files in YAML syntax by default. Output formatting is controlled by output_file_type.

parameter_1: 1
parameter_2: a
class waves.parameter_generators.CustomStudy(parameter_schema: dict, output_file_template: str = None, output_file: str = None, output_file_type: str = 'yaml', set_name_template: str = 'parameter_set@number', previous_parameter_study: str = None, overwrite: bool = False, dryrun: bool = False, write_meta: bool = False, **kwargs)[source]

Bases: _ParameterGenerator

Builds a custom parameter study from user-specified values

Parameters:

  • parameter_schema (array) – Dictionary with two keys: parameter_samples and parameter_names. Parameter samples in the form of a 2D array with shape M x N, where M is the number of parameter sets and N is the number of parameters. Parameter names in the form of a 1D array with length N. When creating a parameter_samples array with mixed type (e.g. string and floats) use dtype=object to preserve the mixed types and avoid casting all values to a common type (e.g. all your floats will become strings).

  • output_file_template (str) – Output file name template. Required if parameter sets will be written to files instead of printed to STDOUT. May contain pathseps for an absolute or relative path template. May contain the @number set number placeholder in the file basename but not in the path. If the placeholder is not found it will be appended to the template string.

  • output_file (str) – Output file name for a single file output of the parameter study. May contain pathseps for an absolute or relative path. output_file and output_file_template are mutually exclusive. Output file is always overwritten.

  • output_file_type (str) – Output file syntax or type. Options are: ‘yaml’, ‘h5’.

  • set_name_template (str) – Parameter set name template. Overridden by output_file_template, if provided.

  • previous_parameter_study (str) – A relative or absolute file path to a previously created parameter study Xarray Dataset

  • overwrite (bool) – Overwrite existing output files

  • dryrun (bool) – Print contents of new parameter study output files to STDOUT and exit

  • write_meta (bool) – Write a meta file named “parameter_study_meta.txt” containing the parameter set file names. Useful for command line execution with build systems that require an explicit file list for target creation.

Example

>>> import waves
>>> import numpy
>>> parameter_schema = dict(
...     parameter_samples = numpy.array([[1.0, 'a', 5], [2.0, 'b', 6]], dtype=object),
...     parameter_names = numpy.array(['height', 'prefix', 'index']))
>>> parameter_generator = waves.parameter_generators.CustomStudy(parameter_schema)
>>> print(parameter_generator.parameter_study)
<xarray.Dataset>
Dimensions:             (data_type: 1, parameter_set_hash: 2)
Coordinates:
  * data_type           (data_type) object 'samples'
    parameter_set_hash  (parameter_set_hash) <U32 '50ba1a2716e42f8c4fcc34a90a...
 *  parameter_sets      (parameter_set_hash) <U14 'parameter_set0' 'parameter...
Data variables:
    height              (data_type, parameter_set_hash) object 1.0 2.0
    prefix              (data_type, parameter_set_hash) object 'a' 'b'
    index               (data_type, parameter_set_hash) object 5 6
Variables:

self.parameter_study – The final parameter study XArray Dataset object

parameter_study_to_dict(*args, **kwargs) → dict[source]

Return parameter study as a dictionary

Used for iterating on parameter sets in an SCons workflow with parameter substitution dictionaries, e.g.

>>> for set_name, parameters in parameter_generator.parameter_study_to_dict().items():
...     print(f"{set_name}: {parameters}")
...
parameter_set0: {'parameter_1': 1, 'parameter_2': 'a'}
parameter_set1: {'parameter_1': 1, 'parameter_2': 'b'}
parameter_set2: {'parameter_1': 2, 'parameter_2': 'a'}
parameter_set3: {'parameter_1': 2, 'parameter_2': 'b'}
Parameters:

data_type (str) – The data_type selection to return - samples or quantiles

Returns:

parameter study sets and samples as a dictionary: {set_name: {parameter: value}, …}

Return type:

dict - {str: {str: value}}

write() → None[source]

Write the parameter study to STDOUT or an output file.

Writes to STDOUT by default. Requires non-default output_file_template or output_file specification to write to files.

If printing to STDOUT, print all parameter sets together. If printing to files, overwrite when contents of existing files have changed. If overwrite is specified, overwrite all parameter set files. If a dry run is requested print file-content associations for files that would have been written.

Writes parameter set files in YAML syntax by default. Output formatting is controlled by output_file_type.

parameter_1: 1
parameter_2: a
class waves.parameter_generators.LatinHypercube(*args, **kwargs)[source]

Bases: _ScipyGenerator

Builds a Latin-Hypercube parameter study from the scipy Latin Hypercube class

The h5 output_file_type is the only output type that contains both the parameter samples and quantiles.

Warning

The merged parameter study feature does not check for consistent parameter distributions. Changing the parameter definitions and merging with a previous parameter study will result in incorrect relationships between parameters and the parameter study samples and quantiles.

Parameters:

  • parameter_schema (dict) – The YAML loaded parameter study schema dictionary - {parameter_name: schema value} LatinHypercube expects “schema value” to be a dictionary with a strict structure and several required keys. Validated on class instantiation.

  • output_file_template (str) – Output file name template. Required if parameter sets will be written to files instead of printed to STDOUT. May contain pathseps for an absolute or relative path template. May contain the @number set number placeholder in the file basename but not in the path. If the placeholder is not found it will be appended to the template string.

  • output_file (str) – Output file name for a single file output of the parameter study. May contain pathseps for an absolute or relative path. output_file and output_file_template are mutually exclusive. Output file is always overwritten.

  • output_file_type (str) – Output file syntax or type. Options are: ‘yaml’, ‘h5’.

  • set_name_template (str) – Parameter set name template. Overridden by output_file_template, if provided.

  • previous_parameter_study (str) – A relative or absolute file path to a previously created parameter study Xarray Dataset

  • overwrite (bool) – Overwrite existing output files

  • dryrun (bool) – Print contents of new parameter study output files to STDOUT and exit

  • write_meta (bool) – Write a meta file named “parameter_study_meta.txt” containing the parameter set file names. Useful for command line execution with build systems that require an explicit file list for target creation.

  • kwargs – Any additional keyword arguments are passed through to the sampler method

To produce consistent Latin Hypercubes on repeat instantiations, the **kwargs must include {'seed': <int>}. See the scipy Latin Hypercube scipy.stats.qmc.LatinHypercube class documentation for details The d keyword argument is internally managed and will be overwritten to match the number of parameters defined in the parameter schema.

Example

>>> import waves
>>> parameter_schema = {
...     'num_simulations': 4,  # Required key. Value must be an integer.
...     'parameter_1': {
...         'distribution': 'norm',  # Required key. Value must be a valid scipy.stats
...         'loc': 50,               # distribution name.
...         'scale': 1
...     },
...     'parameter_2': {
...         'distribution': 'skewnorm',
...         'a': 4,
...         'loc': 30,
...         'scale': 2
...     }
... }
>>> parameter_generator = waves.parameter_generators.LatinHypercube(parameter_schema)
>>> print(parameter_generator.parameter_study)
<xarray.Dataset>
Dimensions:             (data_type: 2, parameter_set_hash: 4)
Coordinates:
    parameter_set_hash  (parameter_set_hash) <U32 '1e8219dae27faa5388328e225a...
  * data_type           (data_type) <U9 'quantiles' 'samples'
  * parameter_sets      (parameter_set_hash) <U14 'parameter_set0' ... 'param...
Data variables:
    parameter_1         (data_type, parameter_set_hash) float64 0.125 ... 51.15
    parameter_2         (data_type, parameter_set_hash) float64 0.625 ... 30.97
Variables:

  • self.parameter_distributions – A dictionary mapping parameter names to the scipy.stats distribution

  • self.parameter_study – The final parameter study XArray Dataset object

parameter_study_to_dict(*args, **kwargs) → dict[source]

Return parameter study as a dictionary

Used for iterating on parameter sets in an SCons workflow with parameter substitution dictionaries, e.g.

>>> for set_name, parameters in parameter_generator.parameter_study_to_dict().items():
...     print(f"{set_name}: {parameters}")
...
parameter_set0: {'parameter_1': 1, 'parameter_2': 'a'}
parameter_set1: {'parameter_1': 1, 'parameter_2': 'b'}
parameter_set2: {'parameter_1': 2, 'parameter_2': 'a'}
parameter_set3: {'parameter_1': 2, 'parameter_2': 'b'}
Parameters:

data_type (str) – The data_type selection to return - samples or quantiles

Returns:

parameter study sets and samples as a dictionary: {set_name: {parameter: value}, …}

Return type:

dict - {str: {str: value}}

write() → None[source]

Write the parameter study to STDOUT or an output file.

Writes to STDOUT by default. Requires non-default output_file_template or output_file specification to write to files.

If printing to STDOUT, print all parameter sets together. If printing to files, overwrite when contents of existing files have changed. If overwrite is specified, overwrite all parameter set files. If a dry run is requested print file-content associations for files that would have been written.

Writes parameter set files in YAML syntax by default. Output formatting is controlled by output_file_type.

parameter_1: 1
parameter_2: a
class waves.parameter_generators.SALibSampler(sampler_class, *args, **kwargs)[source]

Bases: _ParameterGenerator, ABC

Builds a SALib sampler parameter study from a SALib.sample sampler_class

Samplers must use the N sample count argument. Note that in SALib.sample N is not always equivalent to the number of simulations. The following samplers are tested for parameter study shape and merge behavior:

  • fast_sampler

  • finite_diff

  • latin

  • sobol

  • morris

Warning

For small numbers of parameters, some SALib generators produce duplicate parameter sets. These duplicate sets are removed during parameter study generation. This may cause the SALib analyze method(s) to raise errors related to the expected parameter set count.

Warning

The merged parameter study feature does not check for consistent parameter distributions. Changing the parameter definitions and merging with a previous parameter study will result in incorrect relationships between parameters and the parameter study samples.

Parameters:

  • sampler_class (str) – The SALib.sample sampler class name. Case sensitive.

  • parameter_schema (dict) – The YAML loaded parameter study schema dictionary - {parameter_name: schema value} SALibSampler expects “schema value” to be a dictionary with a strict structure and several required keys. Validated on class instantiation.

  • output_file_template (str) – Output file name template. Required if parameter sets will be written to files instead of printed to STDOUT. May contain pathseps for an absolute or relative path template. May contain the @number set number placeholder in the file basename but not in the path. If the placeholder is not found it will be appended to the template string.

  • output_file (str) – Output file name for a single file output of the parameter study. May contain pathseps for an absolute or relative path. output_file and output_file_template are mutually exclusive. Output file is always overwritten.

  • output_file_type (str) – Output file syntax or type. Options are: ‘yaml’, ‘h5’.

  • set_name_template (str) – Parameter set name template. Overridden by output_file_template, if provided.

  • previous_parameter_study (str) – A relative or absolute file path to a previously created parameter study Xarray Dataset

  • overwrite (bool) – Overwrite existing output files

  • dryrun (bool) – Print contents of new parameter study output files to STDOUT and exit

  • write_meta (bool) – Write a meta file named “parameter_study_meta.txt” containing the parameter set file names. Useful for command line execution with build systems that require an explicit file list for target creation.

  • kwargs – Any additional keyword arguments are passed through to the sampler method

Keyword arguments for the SALib.sample sampler_class sample method.

Example

>>> import waves
>>> parameter_schema = {
...     "N": 4,  # Required key. Value must be an integer.
...     "problem": {  # Required key. See the SALib sampler interface documentation
...         "num_vars": 3,
...         "names": ["parameter_1", "parameter_2", "parameter_3"],
...         "bounds": [[-1, 1], [-2, 2], [-3, 3]]
...     }
... }
>>> parameter_generator = waves.parameter_generators.SALibSampler("sobol", parameter_schema)
>>> print(parameter_generator.parameter_study)
<xarray.Dataset>
Dimensions:             (data_type: 1, parameter_sets: 32)
Coordinates:
  * data_type           (data_type) object 'samples'
    parameter_set_hash  (parameter_sets) <U32 'e0cb1990f9d70070eaf5638101dcaf...
  * parameter_sets      (parameter_sets) <U15 'parameter_set0' ... 'parameter...
Data variables:
    parameter_1         (data_type, parameter_sets) float64 -0.2029 ... 0.187
    parameter_2         (data_type, parameter_sets) float64 -0.801 ... 0.6682
    parameter_3         (data_type, parameter_sets) float64 0.4287 ... -2.871
Variables:

self.parameter_study – The final parameter study XArray Dataset object

Raises:

  • ValueError – If the SALib sobol or SALib morris sampler is specified and there are fewer than 2 parameters.

  • AttributeError –

    • N is not a key of parameter_schema

    • problem is not a key of parameter_schema

    • names is not a key of parameter_schema['problem']

  • TypeError –

    • parameter_schema is not a dictionary

    • parameter_schema['N'] is not an integer

    • parameter_schema['problem'] is not a dictionary

    • parameter_schema['problem']['names'] is not a YAML compliant iterable (list, set, tuple)

parameter_study_to_dict(*args, **kwargs) → dict[source]

Return parameter study as a dictionary

Used for iterating on parameter sets in an SCons workflow with parameter substitution dictionaries, e.g.

>>> for set_name, parameters in parameter_generator.parameter_study_to_dict().items():
...     print(f"{set_name}: {parameters}")
...
parameter_set0: {'parameter_1': 1, 'parameter_2': 'a'}
parameter_set1: {'parameter_1': 1, 'parameter_2': 'b'}
parameter_set2: {'parameter_1': 2, 'parameter_2': 'a'}
parameter_set3: {'parameter_1': 2, 'parameter_2': 'b'}
Parameters:

data_type (str) – The data_type selection to return - samples or quantiles

Returns:

parameter study sets and samples as a dictionary: {set_name: {parameter: value}, …}

Return type:

dict - {str: {str: value}}

write() → None[source]

Write the parameter study to STDOUT or an output file.

Writes to STDOUT by default. Requires non-default output_file_template or output_file specification to write to files.

If printing to STDOUT, print all parameter sets together. If printing to files, overwrite when contents of existing files have changed. If overwrite is specified, overwrite all parameter set files. If a dry run is requested print file-content associations for files that would have been written.

Writes parameter set files in YAML syntax by default. Output formatting is controlled by output_file_type.

parameter_1: 1
parameter_2: a
class waves.parameter_generators.ScipySampler(sampler_class, *args, **kwargs)[source]

Bases: _ScipyGenerator

Builds a scipy sampler parameter study from a scipy.stats.qmc sampler_class

Samplers must use the d parameter space dimension keyword argument. The following samplers are tested for parameter study shape and merge behavior:

  • Sobol

  • Halton

  • LatinHypercube

  • PoissonDisk

The h5 output_file_type is the only output type that contains both the parameter samples and quantiles.

Warning

The merged parameter study feature does not check for consistent parameter distributions. Changing the parameter definitions and merging with a previous parameter study will result in incorrect relationships between parameters and the parameter study samples and quantiles.

Parameters:

  • sampler_class (str) – The scipy.stats.qmc sampler class name. Case sensitive.

  • parameter_schema (dict) – The YAML loaded parameter study schema dictionary - {parameter_name: schema value} ScipySampler expects “schema value” to be a dictionary with a strict structure and several required keys. Validated on class instantiation.

  • output_file_template (str) – Output file name template. Required if parameter sets will be written to files instead of printed to STDOUT. May contain pathseps for an absolute or relative path template. May contain the @number set number placeholder in the file basename but not in the path. If the placeholder is not found it will be appended to the template string.

  • output_file (str) – Output file name for a single file output of the parameter study. May contain pathseps for an absolute or relative path. output_file and output_file_template are mutually exclusive. Output file is always overwritten.

  • output_file_type (str) – Output file syntax or type. Options are: ‘yaml’, ‘h5’.

  • set_name_template (str) – Parameter set name template. Overridden by output_file_template, if provided.

  • previous_parameter_study (str) – A relative or absolute file path to a previously created parameter study Xarray Dataset

  • overwrite (bool) – Overwrite existing output files

  • dryrun (bool) – Print contents of new parameter study output files to STDOUT and exit

  • write_meta (bool) – Write a meta file named “parameter_study_meta.txt” containing the parameter set file names. Useful for command line execution with build systems that require an explicit file list for target creation.

  • kwargs – Any additional keyword arguments are passed through to the sampler method

Keyword arguments for the scipy.stats.qmc sampler_class. The d keyword argument is internally managed and will be overwritten to match the number of parameters defined in the parameter schema.

Example

>>> import waves
>>> parameter_schema = {
...     'num_simulations': 4,  # Required key. Value must be an integer.
...     'parameter_1': {
...         'distribution': 'norm',  # Required key. Value must be a valid scipy.stats
...         'loc': 50,               # distribution name.
...         'scale': 1
...     },
...     'parameter_2': {
...         'distribution': 'skewnorm',
...         'a': 4,
...         'loc': 30,
...         'scale': 2
...     }
... }
>>> parameter_generator = waves.parameter_generators.ScipySampler("LatinHypercube", parameter_schema)
>>> print(parameter_generator.parameter_study)
<xarray.Dataset>
Dimensions:             (data_type: 2, parameter_set_hash: 4)
Coordinates:
    parameter_set_hash  (parameter_set_hash) <U32 '1e8219dae27faa5388328e225a...
  * data_type           (data_type) <U9 'quantiles' 'samples'
  * parameter_sets      (parameter_set_hash) <U14 'parameter_set0' ... 'param...
Data variables:
    parameter_1         (data_type, parameter_set_hash) float64 0.125 ... 51.15
    parameter_2         (data_type, parameter_set_hash) float64 0.625 ... 30.97
Variables:

  • parameter_distributions – A dictionary mapping parameter names to the scipy.stats distribution

  • self.parameter_study – The final parameter study XArray Dataset object

parameter_study_to_dict(*args, **kwargs) → dict[source]

Return parameter study as a dictionary

Used for iterating on parameter sets in an SCons workflow with parameter substitution dictionaries, e.g.

>>> for set_name, parameters in parameter_generator.parameter_study_to_dict().items():
...     print(f"{set_name}: {parameters}")
...
parameter_set0: {'parameter_1': 1, 'parameter_2': 'a'}
parameter_set1: {'parameter_1': 1, 'parameter_2': 'b'}
parameter_set2: {'parameter_1': 2, 'parameter_2': 'a'}
parameter_set3: {'parameter_1': 2, 'parameter_2': 'b'}
Parameters:

data_type (str) – The data_type selection to return - samples or quantiles

Returns:

parameter study sets and samples as a dictionary: {set_name: {parameter: value}, …}

Return type:

dict - {str: {str: value}}

write() → None[source]

Write the parameter study to STDOUT or an output file.

Writes to STDOUT by default. Requires non-default output_file_template or output_file specification to write to files.

If printing to STDOUT, print all parameter sets together. If printing to files, overwrite when contents of existing files have changed. If overwrite is specified, overwrite all parameter set files. If a dry run is requested print file-content associations for files that would have been written.

Writes parameter set files in YAML syntax by default. Output formatting is controlled by output_file_type.

parameter_1: 1
parameter_2: a
class waves.parameter_generators.SobolSequence(*args, **kwargs)[source]

Bases: _ScipyGenerator

Builds a Sobol sequence parameter study from the scipy Sobol class random method.

The h5 output_file_type is the only output type that contains both the parameter samples and quantiles.

Warning

The merged parameter study feature does not check for consistent parameter distributions. Changing the parameter definitions and merging with a previous parameter study will result in incorrect relationships between parameters and the parameter study samples and quantiles.

Parameters:

  • parameter_schema (dict) – The YAML loaded parameter study schema dictionary - {parameter_name: schema value} SobolSequence expects “schema value” to be a dictionary with a strict structure and several required keys. Validated on class instantiation.

  • output_file_template (str) – Output file name template. Required if parameter sets will be written to files instead of printed to STDOUT. May contain pathseps for an absolute or relative path template. May contain the @number set number placeholder in the file basename but not in the path. If the placeholder is not found it will be appended to the template string.

  • output_file (str) – Output file name for a single file output of the parameter study. May contain pathseps for an absolute or relative path. output_file and output_file_template are mutually exclusive. Output file is always overwritten.

  • output_file_type (str) – Output file syntax or type. Options are: ‘yaml’, ‘h5’.

  • set_name_template (str) – Parameter set name template. Overridden by output_file_template, if provided.

  • previous_parameter_study (str) – A relative or absolute file path to a previously created parameter study Xarray Dataset

  • overwrite (bool) – Overwrite existing output files

  • dryrun (bool) – Print contents of new parameter study output files to STDOUT and exit

  • write_meta (bool) – Write a meta file named “parameter_study_meta.txt” containing the parameter set file names. Useful for command line execution with build systems that require an explicit file list for target creation.

  • kwargs – Any additional keyword arguments are passed through to the sampler method

To produce consistent Sobol sequences on repeat instantiations, the **kwargs must include either scramble=False or seed=<int>. See the scipy Sobol scipy.stats.qmc.Sobol class documentation for details. The d keyword argument is internally managed and will be overwritten to match the number of parameters defined in the parameter schema.

Example

>>> import waves
>>> parameter_schema = {
...     'num_simulations': 4,  # Required key. Value must be an integer.
...     'parameter_1': {
...         'distribution': 'uniform',  # Required key. Value must be a valid scipy.stats
...         'loc': 0,                   # distribution name.
...         'scale': 10
...     },
...     'parameter_2': {
...         'distribution': 'uniform',
...         'loc': 2,
...         'scale': 3
...     }
... }
>>> parameter_generator = waves.parameter_generators.SobolSequence(parameter_schema)
>>> print(parameter_generator.parameter_study)
<xarray.Dataset>
Dimensions:             (data_type: 2, parameter_sets: 4)
Coordinates:
    parameter_set_hash  (parameter_sets) <U32 'c1fa74da12c0991379d1df6541c421...
  * data_type           (data_type) <U9 'quantiles' 'samples'
  * parameter_sets      (parameter_sets) <U14 'parameter_set0' ... 'parameter...
Data variables:
    parameter_1         (data_type, parameter_sets) float64 0.0 0.5 ... 7.5 2.5
    parameter_2         (data_type, parameter_sets) float64 0.0 0.5 ... 4.25
parameter_study_to_dict(*args, **kwargs) → dict[source]

Return parameter study as a dictionary

Used for iterating on parameter sets in an SCons workflow with parameter substitution dictionaries, e.g.

>>> for set_name, parameters in parameter_generator.parameter_study_to_dict().items():
...     print(f"{set_name}: {parameters}")
...
parameter_set0: {'parameter_1': 1, 'parameter_2': 'a'}
parameter_set1: {'parameter_1': 1, 'parameter_2': 'b'}
parameter_set2: {'parameter_1': 2, 'parameter_2': 'a'}
parameter_set3: {'parameter_1': 2, 'parameter_2': 'b'}
Parameters:

data_type (str) – The data_type selection to return - samples or quantiles

Returns:

parameter study sets and samples as a dictionary: {set_name: {parameter: value}, …}

Return type:

dict - {str: {str: value}}

write() → None[source]

Write the parameter study to STDOUT or an output file.

Writes to STDOUT by default. Requires non-default output_file_template or output_file specification to write to files.

If printing to STDOUT, print all parameter sets together. If printing to files, overwrite when contents of existing files have changed. If overwrite is specified, overwrite all parameter set files. If a dry run is requested print file-content associations for files that would have been written.

Writes parameter set files in YAML syntax by default. Output formatting is controlled by output_file_type.

parameter_1: 1
parameter_2: a