Command Line¶

The cplots command makes triangle plots from MCMC chains and line plots from data files without writing any Python.

pip install "cplots[cli,matplotlib]"        # add getdist/chainconsumer/anesthetic for chains

Run cplots --help, or --help on any command, for the full option list.

Command	What it does
`cplots info CHAIN`	Print sample count, weights and parameter names of a chain
`cplots distributions triangle CHAIN`	Triangle (corner) plot from MCMC chains
`cplots lines xy FILE`	Plot columns of a data file against one another
`cplots config set\\|get\\|unset\\|show`	Manage persistent defaults

Inspect a chain¶

CHAIN is a chain file, a chain root prefix (GetDist or cobaya), or a directory holding one root. Start here to find the parameter names:

cplots info chains/planck_TTTEEE
cplots info chains/run.1.txt --skip 0.3     # discard the first 30% as burn-in

Triangle plots¶

# Pick parameters interactively, preview in a window
cplots distributions triangle chains/planck_TTTEEE

# Three parameters, in this order, as a PRD-sized PDF
cplots distributions triangle chains/ -p Omega_m,sigma8,H0 --paper-format PRD -o fig.pdf

# Truth lines and label overrides from YAML, ChainConsumer engine
cplots distributions triangle chains/ -p Omega_m,sigma8 --config truths.yaml \
    --engine chainconsumer -o corner.png

Without --params, an interactive selector lists the chain's parameters (this needs a terminal). Parameter labels come from the chain (cobaya yaml, GetDist .paramnames), then --config, then the raw names.

Option	Meaning
`-p`, `--params NAME[,NAME...]`	Parameters to plot, in order (comma-separated and/or repeated)
`-e`, `--engine`	`getdist` (default), `chainconsumer` or `anesthetic`
`--config FILE`	YAML with `truths: {name: value}` and `labels: {name: latex}`
`--latex`	Render raw parameter names as math when the chain has no labels
`-s`, `--skip`	Fraction of each chain to discard as burn-in (0 ≤ skip < 1)

Line plots from data files¶

FILE holds whitespace-delimited columns (.txt, .dat) or comma-delimited .csv; lines starting with # are ignored. Each y column becomes a trace.

# Column 1 against column 0, in a window
cplots lines xy pk.txt --xlabel '$k$' --ylabel '$P(k)$'

# Log-log, saved as PDF
cplots lines xy pk.txt --xscale log --yscale log -o pk.pdf

# Columns 1, 2, 3 as three labelled traces (same as -y 1,2,3)
cplots lines xy data.txt -y 1:4 -l "z=0" -l "z=0.5" -l "z=1"

# PRD single-column figure, interactive HTML with plotly
cplots lines xy pk.txt --paper-format PRD --backend plotly -o fig.html

Option	Meaning
`-x`, `--x-col`	0-based column for x (default 0)
`-y`, `--y-col`	0-based y column(s): `1,2,3` or `1:4` (end exclusive), and/or repeated; default 1
`-l`, `--label`	Legend label, one per y column, in order; repeat the flag
`--xlabel`, `--ylabel`	Axis labels (LaTeX in $...$ works)
`--xscale`, `--yscale`	`linear` (default) or `log`

Options shared by plot commands¶

Option	Meaning
`-b`, `--backend`	`matplotlib` (default), `plotly`, `altair` or `ultraplot`
`--style`	`publication` or `dark`, layered on the default style
`--paper-format`	Journal size and font preset: `PRD`, `PRD-2col`, `MNRAS`, `A4`
`-o`, `--output FILE`	Save the figure; the extension sets the format (`.pdf`, `.png`, `.svg`, `.html` with plotly)
`--show` / `--no-show`	Open the figure; shown by default when no `--output` is given

Persistent defaults¶

Store values you would otherwise repeat on every command. A flag given on the command line always wins over a stored value.

cplots config set backend plotly
cplots config set paper_format PRD
cplots config get backend        # prints the value, or "(not set)"
cplots config unset backend
cplots config show               # all settings and the config file location

Keys: backend, engine, theme (the stored name for --style) and paper_format. Settings live in $XDG_CONFIG_HOME/cplots/config.json (~/.config/cplots/config.json by default).