Uncertain.jl 📊

Important

Handle uncertain values with ease and high performance! 🚀

A Julia package for representing and propagating uncertainties through calculations with minimal overhead. Designed with composability and speed in mind, applicable to large datasets of uncertain values.

Key Features ✨

• ⚡ Minimal overhead: Only 2x memory usage and ~3x computational overhead compared to regular values
• 🎯 Simple API: Clean interface with the ±ᵤ operator and U module
• 🔧 Flexible uncertainty types: Support for symmetric, asymmetric, and custom uncertainty representations
• 🌐 Rich ecosystem integration: Works seamlessly with Makie, Unitful, and more

Quick Start 🚀

The most common case is numbers with measurement errors. 📏 Uncertain.jl exports the ±ᵤ operator and the U module:

using Uncertain

julia> x = 1 ±ᵤ 0.1  # Create an uncertain value
1.0 ± 0.1

julia> typeof(x)
Uncertain.ValueReal{Float64, Float64}

julia> U.value(x)     # Extract the nominal value
1.0

julia> U.uncertainty(x)  # Extract the uncertainty
0.1

julia> y = 2.5 ±ᵤ 0.2
2.5 ± 0.2

julia> x + y  # Automatic error propagation ✨
3.5 ± 0.22360679774997896

Plotting 📊

Uncertain.jl provides Makie integration for visualizing uncertain values. You can pass uncertain values directly to plotting functions:

using Uncertain, GLMakie

# Create some data with uncertainties
x = 1:10
y = [i + 0.5*randn() ±ᵤ (0.1 + 0.05*i) for i in x]

# Plot with automatic uncertainty handling
scatter(x, y)
rangebars(x, y)
lines(x, y)
band(x, y)

The plotting functions scatter(), rangebars(), lines(), and band() (and others where it makes sense) work directly with uncertain values, automatically extracting the necessary components for visualization.

Scope 🎯

The ultimate goal of Uncertain.jl is to support arbitrary uncertainty specifications – asymmetric errors, intervals, more complex distributions, and to go beyond plain numbers. All within a single uniform interface. 🌈

Currently, aside from plain numbers, Uncertain provides the U.TwoSided(lo, hi) objects to represent asymmetric two-sided uncertainties. The companion UncertainSkyCoords.jl package provides uncertainties specific to SkyCoords.jl objects – essentially, sky regions with different shapes. These types do not support all operations where error propagation would make sense yet; reports and suggestions on specific missing operations are welcome!

Furthermore, Uncertain.jl provides integrations with other parts of the Julia ecosystem: 🔗

• 📊 Plotting with Makie – you can pass uncertain values to any recipe where it makes sense
• 🔄 Conversions to and from IntervalSets.jl, Measurements.jl and MonteCarloMeasurements.jl objects for ease of interoperability
• 📐 Uncertain.Values with Unitful.jl quantities inside them support the Unitful.jl interface

Why ±ᵤ instead of ±? 🤔

We use the ±ᵤ operator (not just ±) to avoid conflicts with the popular IntervalSets.jl package, which exports the ± operator for creating intervals. This design choice ensures seamless compatibility when both packages are used together. 🤝

If you prefer to use the shorter ± symbol, you can import it explicitly: ✨

using Uncertain
import Uncertain: ±ᵤ as ±

julia> x = 1.5 ± 0.05  # Now you can use the standard ± symbol
1.5 ± 0.05

julia> y = 3.2 ± 0.1
3.2 ± 0.1

julia> x * y  # Works exactly the same ⚡
4.8 ± 0.19209372712298546

Relation to other uncertainty packages 📈

TLDR: Uncertain.jl prioritizes performance over correlation handling. Use Measurements.jl or MonteCarloMeasurements.jl if you need to track correlations; use Uncertain.jl for large datasets where you can assume uncertainties are independent.

The fundamental design difference of Uncertain.jl, compared to other Julia packages in this field like Measurements.jl and MonteCarloMeasurements.jl, is low overhead. 🚄 It handles large datasets of values with uncertainties. They occupy only 2x more memory than regular values, and have a performance overhead of only a factor of a few.

julia> n = 10^6
julia> x = rand(n)
julia> xu = x .±ᵤ 0.1

julia> Base.summarysize(x)
8000040

julia> Base.summarysize(xu)
16000040

julia> @b x .+ 1
482.584 μs (4 allocs: 7.629 MiB)

julia> @b xu .+ 1
1.201 ms (4 allocs: 15.259 MiB)

Other packages tend to have orders-of-magnitude overhead: Measurements.jl because it handles linear correlations, and MonteCarloMeasurements.jl because it uses Monte Carlo samples to represent uncertainties. 📊

Those other packages have much higher performance overhead, but they certainly have advantages over Uncertain.jl. ⚖️ In particular, here we only handle operations on independent uncertain values. All functions that involve only one uncertain number like exp(x::Uncertain.Value) or x::Uncertain.Value + y::Float64 are automatically correct as there are no dependencies possible. Computing multivariate functions like x::Uncertain.Value + y::Uncertain.Value correctly requires accounting for correlations in the general case; we do not do that here. We have a flag one can set to control whether to assume independent values, or fail on these operations: 🛠️

julia> Uncertain.assume_independent() = false

julia> x = 1 ±ᵤ 0.1
julia> y = 2 ±ᵤ 0.5

julia> x + y
ERROR: The `+` operation is only correct for independent `Uncertain.Value`s, and we have no way to prove the independence automatically.
Set `Uncertain.assume_independent() = true` to execute such functions.

julia> Uncertain.assume_independent() = true

julia> x + y
3.0 ± 0.5099019513592785

Naturally, the choice of which package to use differs in different scenarios. 🎯 Uncertain.jl values can easily be converted back and forth to corresponding objects of other packages, making interoperability more straightforward. 🔄