Getting started¶

QUENDS (Quantifying Uncertainty in ENsemble Data Streams) turns raw time-series simulation output into trustworthy statistics: it trims transients, detects steady state, and quantifies uncertainty for single signals and for ensembles of runs.

Installation¶

pip install quends

To work from a checkout (editable install with the development extras):

git clone https://github.com/sandialabs/quends.git
cd quends
pip install -e ".[dev]"

Basic usage¶

The core object is the DataStream. A typical single-signal analysis is three steps — load → trim → quantify:

import quends as qnds

# 1. Load one signal (plus its time column) from a CSV file.
ds = qnds.from_csv("examples/data/cgyro/output_nu0_50.csv", "Q_D/Q_GBD")

# 2. Trim the warm-up transient, keeping only the steady-state region.
trimmed = ds.trim(method="threshold", window_size=100, threshold=0.1)

# 3. Compute statistics with an honest uncertainty estimate.
stats = trimmed.compute_statistics()
print(stats)

{'Q_D/Q_GBD': {'mean': 25.07, 'mean_uncertainty': 1.14, 'window_size': 85,
               'effective_sample_size': 42, ...}}

Every loader takes the file and one variable name and returns a DataStream that holds just time and that signal. Use ds.variables() to list available columns and ds.data to reach the underlying pandas DataFrame.

Ensembles in one step¶

For a collection of runs, build an Ensemble and compute an ensemble estimate (ensemble_average, pooled_block_means, or ivw_member_means):

paths = ["run01.csv", "run02.csv", "run03.csv"]
ens = qnds.Ensemble.from_files(paths, "Q_D/Q_GBD")
result = ens.compute_uncertainty(method="ensemble_average")
print(result)

Where to next¶

• User Guide — the concepts and the recommended workflow for each part of the library (loading, trimming, statistics, ensembles, plotting, workflows).

• Gallery of examples — runnable, end-to-end tutorials.

• API Reference — the full, auto-generated API.