Using the command line

pk has four commands:

Command	What it does
pk describe [model]	Lists tools, models, parameters, bounds and constraints
pk fit <tool> <file...>	Fits a model: photosynthesis, stomatal, hydraulics or soil
pk li600 correct <file>	Corrects LI-600 stomatal conductance for evaporative cooling
pk li600 estimate <file>	Solves A, Ci and Cc from LI-600 conductance
pk concat <file...>	Merges LI-6800 logs or CSVs, with a CurveID per file

pk --help prints this list, and pk <command> --help prints a command's options. Commands documents every option.

Input files

pk reads:

• CSV files with one header row of column names.
• .xlsx workbooks, using the first sheet. For photosynthesis, the LI-6800's Measurements sheet.
• LI-6800 native logs, the files with no extension that the instrument writes. PhytKit finds the [Data] block and the column header itself.
• Standard input, if you give - as the file name.

PhytKit finds the columns it needs by name and accepts common variants, such as Psi (MPa) or water_potential for water potential. It also converts common units, such as °C to Kelvin, bar to MPa, and a percent to a fraction. It reports each choice in warnings. See Input files.

Options

Options follow the command: --name value, --name=value, or --flag alone for a yes/no option. A few options can be repeated, such as --set and --survey. Lists can be comma-separated: --fix gm,alphaG.

pk checks every option before it runs anything. A misspelled or unknown option, a value of the wrong type, or an extra argument is an error, never silently ignored. Run pk <command> --help to see a command's options.

The options you'll use most with pk fit:

Option	Example	Effect
--model	--model fvcbC4	Model to fit (each tool has a default)
--fit	--fit Vcmax25,Jmax25	Fit only these parameters
--fix	--fix gm,alphaG	Hold these parameters fixed
--set	--set Vcmax25=80	Starting value, or the fixed value of a fixed parameter
--max-iter	--max-iter 400	Iteration limit (default 200)
--group-by	--group-by Leaf	Split one file into one fit per value of a column
--format	--format csv	Print CSV instead of JSON

Output and errors

Every command prints one JSON object to standard output. It has "ok": true and the results, unless you asked for --format csv.

When something goes wrong, pk prints an error object and exits with code 1:

{
  "ok": false,
  "command": "li600 correct",
  "error": {
    "code": "unknown_option",
    "message": "Unknown option --sidednes for pk li600 correct.",
    "usage": true,
    "hint": "Did you mean --sidedness? Options: --sidedness, --thermal-conductance, --rows, --out, --format.",
    "help": "pk li600 correct <file> [options]"
  }
}

• code names the kind of error, such as unknown_option or file_not_found. Codes don't change between versions, so scripts can branch on them. They're all listed in Commands.
• message says what was wrong.
• hint, when present, says what to do: the option, model, parameter or tool you probably meant, and the valid choices.
• usage is true when the command line was wrong, so fixing the arguments and retrying will help. It's false when a file or its data was the problem.
• help, on usage errors, is the command's usage line.

For scripts and coding agents

pk is designed to be driven by other programs:

• It never asks questions or waits for input.
• pk describe and pk describe <model> say what each model takes, so a script doesn't need to read these docs.
• Output is always JSON, or CSV when you ask for it.
• Errors have a stable code, and a hint naming the likely fix. Nothing is silently ignored: an unknown option, a stray argument or an out-of-range value stops the command before it runs.
• pk describe also lists every command's options, with types, defaults and choices, and every error code.
• --format csv and --out write tables you can open directly.