Utilities

class baseclasses.utils.CaseInsensitiveDict(*args, **kwargs)

Python dictionary where the keys are case-insensitive. Note that this assumes the keys are strings, and indeed will fail if you try to create an instance where keys are not strings. All common Python dictionary operations are supported, and additional operations can be added easily. In order to preserve capitalization on key initialization, the implementation relies on storing a dictionary of mappings, which are used to check any new keys against existing keys and compare them in a case-insensitive fashion.

Warning

This container preserves the initial capitalization, such that any operation which operates on an existing entry will not modify it. This means that for example __setitem__() will NOT update the original capitalization.

Attributes:

datadict

The equivalent case-sensitive dictionary. This stores the actual values.

mapdict

Dictionary of mappings between the lowercase representation and the initial capitalization.

class baseclasses.utils.CaseInsensitiveSet(*args, **kwargs)

Python set where the elements are case-insensitive. Note that this assumes the elements are strings, and indeed will fail if you try to create an instance where elements are not strings. All common Python set operations are supported, and additional operations can be added easily. In order to preserve capitalization on key initialization, the implementation relies on storing a dictionary of mappings which are used to check any new keys against existing keys and compare them in a case-insensitive fashion.

Warning

This container preserves the initial capitalization, such that any operation which operates on an existing entry will not modify it. This means that add() and update() will NOT update the original capitalization.

Attributes:

dataset

The equivalent case-sensitive set.

mapdict

Dictionary of mappings between the lowercase representation and the initial capitalization.

add(item)

Add an element.

discard(item)

Remove an element. Do not raise an exception if absent.

issubset(other)

We convert both to regular set, and compare their lower case values

update(d)

Just call add() iteratively

exception baseclasses.utils.Error(message)

Format the error message in a box to make it clear this was a explicitly raised exception.

class baseclasses.utils.SolverHistory(includeIter=True, includeTime=True)

The SolverHistory class can be used to store and print various useful values during the execution of a solver.

NOTE: The implementation of this class contains no consideration of parallelism. If you are using a solverHistory object in your parallel solver, you will need to take care over which procs you make calls to the SolverHistory object on.

Create a solver history instance

Parameters:

includeIterbool, optional

Whether to include the history’s internal iteration variable in the history, by default True

includeTimebool, optional

Whether to include the history’s internal timing variable in the history, by default True

addMetadata(name, data)

Add a piece of metadata to the history

The metadata attribute is simply a dictionary that can be used to store arbitrary information related to the solution being recorded, e.g solver options

Parameters:

namestr

Item name/key

dataAny

Item to store

addVariable(name, varType, printVar=False, valueFormat=None, overwrite=False)

Define a new field to be stored in the history.

Parameters:

namestr

Variable name

varTypeType

Variable type, i.e int, float, str etc

printVarbool, optional

Whether to include the variable in the iteration printout, by default False

valueFormatstr, optional

Format string valid for use with the str.format() method (e.g “{:17.11e}” for a float or “{:03d}” for an int), only important for variables that are to be printed. By default a predefined format for the given varType is used

overwritebool, optional

Whether to overwrite any existing variables with the same name, by default False

getData()

Get the recorded data

Returns:

dict

Dictionary of recorded data

getIter()

Get the current number of iterations recorded

Returns:

int

Number of iterations recorded

getMetadata()

Get the recorded metadata

Returns:

dict

Dictionary of recorded metadata

getVariables()

Get the recorded variables

Returns:

dict

Dictionary of recorded variables

printData(iters=None)

Print a selection of lines from the history

Each line will look something like this:

|      0  |  1.000e-01  |      -84     |   2.87098307489e-01  |      ...      |

Parameters:

itersint or Iterable of ints, optional

Iteration numbers to print, by default only the last iteration will be printed

printHeader()

Print the header of the iteration printout

The header will look something like this:

+--------------------------------------------------------------------...------+
|  Iter   |    Time     |  Random Int  |     Random Float     |      ...      |
+--------------------------------------------------------------------...------+

reset(clearMetadata=False)

Reset the history to its initial state.

Parameters:

clearMetadatabool, optional

Whether to clear the metadata too, by default False

save(fileName)

Write the solution history to a pickle file

Only the data dictionary is saved

Parameters:

fileNamestr

File path to save the solution history to, file extension not required, will be ignored if supplied

startTiming()

Record the start time of the solver

This function only needs to be called explicitly if the start time of your solver is separate from the first time the write method is called.

write(data)

Record data for a single iteration

Note that each call to this method is treated as a new iteration. All data to be recorded for a single solver iteration must therefore be recorded in a single call to this method.

Parameters:

datadict

Dictionary of values to record, with variable names as keys

writeFullVariableHistory(name, values)

Write the entire history of a variable in one go

This function should be used in the case where your solver already handles the recording of variables during a solution (e.g ADflow) but you want to use a SolverHistory object to facilitate writing it to a file

Parameters:

namestr

Variable name

valuesIterable

Values to record, will be converted to a list

baseclasses.utils.getPy3SafeString(string)

Accepts a string and makes sure it’s converted to unicode for python 3.6 and above

baseclasses.utils.pp(obj, comm=None, flush=True)

Parallel safe printing routine. This method prints obj (via pprint) on the root proc of self.comm if it exists. Otherwise it will just print obj.

Parameters:

objobject

Any Python object to be printed

commMPI comm

The MPI comm object on this processor

flushbool

If True, the stream will be flushed.

baseclasses.utils.readJSON(fname, comm=None)

Reads a JSON file and return the contents as a dictionary. This includes a custom NumPy reader to retrieve NumPy arrays, matching the writeJSON() function.

Parameters:

file_namestr

The file name

commmpi4py.MPI.Comm, optional

The communicator over which this function is called. If supplied, only the root proc will be used for file IO.

References

This is based on this stack overflow answer

baseclasses.utils.readPickle(fname, comm=None)

This is a parallel pickle.load function, which is performed on the root proc only. Error checking is necessary to provide py2 compatibility.

Parameters:

fnamestr

The pickle file name

commmpi4py.MPI.Comm, optional

The communicator over which this function is called. If supplied, only the root proc will be used for file IO.

Returns:

objThe object stored in the pickle file

baseclasses.utils.redirectIO(f_out, f_err=None)

This function redirects stdout/stderr to the given file handle.

Parameters:

f_outfile

A file stream to redirect stdout to

f_errfile

A file stream to redirect stderr to. If none is specified it is set to f_out

baseclasses.utils.redirectingIO(f_out, f_err=None)

A function that redirects stdout in a with block and returns to the stdout after the with block completes. The filestream passed to this function will be closed after exiting the with block.

Here is an example of usage where all adflow output is redirected to the file adflow_out.txt:

>>> from baseclasses.utils import redirectIO
>>> print("Printing some information to terminal")
>>> with redirectIO.redirectingIO(open("adflow_out.txt", "w")):
...     CFDSolver = ADFLOW(options=options)
...     CFDSolver(AeroProblem(**apOptions)
>>> print("Printing some more information to terminal")

Parameters:

f_outfile

A file stream that stdout should be redirected to

f_errfile

A file stream to redirect stderr to. If none is specified it is set to f_out

baseclasses.utils.writeJSON(fname, obj, comm=None)

Write an object to a JSON file. This includes a custom NumPy encoder to reliably write NumPy arrays to JSON, which can then be read back via readJSON().

Parameters:

fnamestr

The file name

objdict or ndarray

The object to be written to JSON

commmpi4py.MPI.Comm, optional

The communicator over which this function is called. If supplied, only the root proc will be used for file IO.

baseclasses.utils.writePickle(fname, obj, comm=None)

Parallel pickle writing function, only performs operations on the root proc

Parameters:

fnamestr

The pickle file name

objany object which can be pickled by Python

The object to be pickled

commmpi4py.MPI.Comm, optional

The communicator over which this function is called. If supplied, only the root proc will be used for file IO.