Python API

import phytkit as pk

Commands

These run pk commands and return what pk would print: a dict for JSON output, or a str for format="csv". Options are pk's options as keyword arguments, with _ in place of - (see Scripting with Python). They raise PhytKitError on failure.

Function	pk command
pk.describe(model=None)	pk describe [model]
pk.fit(tool, *files, **options)	pk fit <tool> <file...>
pk.li600_correct(file, **options)	pk li600 correct <file>
pk.li600_estimate(file, **options)	pk li600 estimate <file>
pk.concat(*files, **options)	pk concat <file...>
pk.run(argv)	Any pk command line, as a list of strings

How keyword values become options:

Python value	Becomes
True	the option alone: predicted=True → --predicted
False or None	nothing
a dict	one --name key=value per entry: set={"gm": 1e6}
a list	for fit and fix, one comma-separated option; for others, the option repeated: survey=["a", "b"]
anything else	--name value, with paths converted to strings

Fitting data in memory

fit_arrays

pk.fit_arrays(model_name, x, y, *, fit_params=None, init=None, max_iter=None,
              curves=None, constraints=None, constraint_weights=None) -> dict

Argument	Meaning
model_name	A model from pk.models()
x	Dict of input name → sequence of values; see Models and parameters for each model's inputs and units (temperature in Kelvin)
y	Sequence of measured responses, the same length as each input
fit_params	Parameters to optimize (default: those not fixed by default)
init	Dict of starting values, or the values fixed parameters are held at
max_iter	Iteration limit (default 200)
curves	List of row-index lists, one per curve, for per-curve constraints
constraints	Constraint keys to apply (default: all)
constraint_weights	Dict of constraint key → weight

Returns a dict:

Key	Meaning
params	Every parameter's value
fitted	Names of the optimized parameters
predicted	Model prediction for each point
rmse, r2	Fit quality on y
iterations, converged	How the fit finished; see How fitting works

fit_many

pk.fit_many(model_name, datasets, *, fit_params=None, init=None, max_iter=None,
            constraints=None, constraint_weights=None, threads=None) -> list[dict]

Fits each (x, y) pair in datasets in parallel and returns one result per dataset, in order, each as from fit_arrays. threads limits the worker threads (default: every core).

predict

pk.predict(model_name, params, x) -> list[float]

The model's prediction at x. Parameters missing from params take their defaults.

models

pk.models() -> list[str]

The names of every built-in model.

Errors

PhytKitError : Raised by every function. It has the fields of pk's error object:

| Attribute | |
|---|---|
| `str(e)` | The message |
| `e.code` | The kind of error, such as `unknown_option` (see [Commands](commands.md#errors)) |
| `e.hint` | What to do about it, or `None` |
| `e.usage` | `True` when the call was wrong, `False` when a file or its data was |
| `e.help` | The command's usage line, on usage errors from command functions, or `None` |

Version

pk.__version__ is the package version.