config - bossanova

Global configuration for bossanova.

This module provides package-wide configuration settings that can be modified at runtime. Currently manages singular tolerance for mixed models.

Functions:

Name	Description
`get_display_digits`	Get the number of significant figures for DataFrame display output.
`get_singular_tolerance`	Get the current singular tolerance for mixed models.
`is_singular`	Check whether a mixed model fit is singular.
`set_display_digits`	Set the number of significant figures for DataFrame display output.
`set_singular_tolerance`	Set the global singular tolerance for mixed models.

Functions¶

get_display_digits¶

get_display_digits() -> int

Get the number of significant figures for DataFrame display output.

Controls how many significant figures are shown in user-facing DataFrames returned by model properties (.params, .effects, .diagnostics, etc.) and compare(). Matches R’s signif() convention.

Returns:

Type	Description
`int`	Current display digits setting (default 4).

Examples:

from bossanova import get_display_digits
get_display_digits()
# 4

get_singular_tolerance¶

get_singular_tolerance() -> float

Get the current singular tolerance for mixed models.

The singular tolerance is used by isSingular() to determine if a mixed model fit is singular (has variance components at or near zero).

Returns:

Type	Description
`float`	The current singular tolerance threshold.

Examples:

from bossanova import get_singular_tolerance
get_singular_tolerance()
# 0.0001

is_singular¶

is_singular(theta: np.ndarray | None, tol: float | None = None) -> bool

Check whether a mixed model fit is singular.

A fit is singular when any theta parameter (variance component on the relative scale) is at or near zero, indicating a variance component on the boundary of the parameter space.

Parameters:

Name	Type	Description	Default
`theta`	`ndarray \| None`	Relative variance parameters from REML/ML optimization. None indicates no random effects (returns False).	required
`tol`	`float \| None`	Tolerance threshold. If None, uses the global singular tolerance from `get_singular_tolerance()`.	`None`

Returns:

Type	Description
`bool`	True if any element of theta has absolute value below the tolerance.

Examples:

from config import is_singular
import numpy as np
is_singular(np.array([0.5, 1e-6]))  # True (1e-6 < 1e-4)
is_singular(np.array([0.5, 0.3]))   # False
is_singular(None)                    # False

set_display_digits¶

set_display_digits(digits: int) -> None

Set the number of significant figures for DataFrame display output.

The summary() method has its own digits parameter and is unaffected by this setting.

Parameters:

Name	Type	Description	Default
`digits`	`int`	Number of significant figures. Must be >= 1.	required

Examples:

from bossanova import set_display_digits
set_display_digits(6)  # Show 6 significant figures

set_singular_tolerance¶

set_singular_tolerance(tol: float) -> None

Set the global singular tolerance for mixed models.

The singular tolerance is used by isSingular() to determine if a mixed model fit is singular. Values below this threshold are considered effectively zero.

Parameters:

Name	Type	Description	Default
`tol`	`float`	New tolerance threshold. Must be positive.	required

Examples:

from bossanova import set_singular_tolerance
set_singular_tolerance(1e-6)  # More strict threshold