CurveMole

FREE, OPEN-SOURCE SCIENTIFIC SOFTWARE

CurveMoleInteractive
curve fitting.

Extensible scientific analysis.

Build models visually, analyse individual curves or full series, and inspect residuals and parameter uncertainty in one modular desktop toolbox.

Spectroscopy · Diffraction · Kinetics · Experimental x–y data

Windows, Linux & macOS · GPL-3.0-or-later

Explore the workflow ↓
THE ACTUAL WORKSPACEView full size ↗
Actual CurveMole workspace showing measured fluorescence data, five model components, a total fit, residuals and fitted parameters
Data, model components, fit and residuals — together.
About this experimental example

Measured fluorescence data: five pseudo-Voigt profiles multiplied by filter transmittance. The saved project contains 14 series and 98 curves, with block-bootstrap intervals, residual structure and attention flags retained. Sample and measurement labels have been renamed. Project provenance ↗

FROM DATA TO RESULTS

One workspace for the whole analysis.

Import → Build a model → Fit → Inspect → Save and export

01 / MODEL

Build visually.

Place and adjust peaks on the plot. Add backgrounds and custom functions, and inspect each component.

02 / CONSTRAIN

Define the parameters.

Set bounds, fix values and define relationships within a curve or across several datasets.

03 / PREPARE

Preserve the originals.

Apply reversible masks and transformations while retaining your imported data.

04 / KEEP

Save the analysis.

Keep models, results and laboratory notes in a portable project. Export data, parameters, figures and reports together.

SINGLE CURVES · EXPERIMENTAL SERIES

From one curve
to a full series.

Follow changes with temperature, pressure or time.
Compare the data, then choose how to fit the series.

INDEPENDENT

Fit each curve separately.

Analyse individual datasets with their own models and parameters.

SEQUENTIAL

Propagate a model.

Carry a prepared model through the series, with inspection and pause/resume.

GLOBAL SIMULTANEOUS

Estimate shared parameters.

Fit several curves in one optimisation, linking shared or related quantities.

WATERFALL VIEW

Follow changes across a series.

Use vertical offsets to compare peak positions, widths and amplitudes across related curves.

View documentation ↗Illustrative Raman-like data. Actual application screenshots.
Illustrative Raman-like spectra and their fitted models in CurveMole Waterfall view
DIAGNOSTICS AND UNCERTAINTY

Inspect the fit.
Assess uncertainty.

Examine residual structure, parameter correlations and confidence intervals. Choose additional uncertainty analyses to suit your model and noise assumptions.

Uncertainty methods and assumptions ↗
Covariance and correlations
Inspect local parameter uncertainty and relationships between estimates.
Monte Carlo and bootstrap
Explore uncertainty with simulations, residual bootstrap or block bootstrap.
Profile likelihood
Examine confidence regions beyond the local covariance approximation.
Fitting methods and scientific conventions

Weighted least squares

For weighted least squares with supplied standard deviations in y, the objective is

χ2(θ)=∑i[yi−f(xi;θ)σy,i]2

Here θ denotes the free model parameters. X uncertainties and correlations between data points are not included in this objective.

Weighting conventions ↗
Models and constraints
User-defined formulas, polynomial and cubic-spline functions, and Gaussian, Lorentzian, Voigt and pseudo-Voigt profiles with signed integrated area. Fixed values, bounds and expression links within and across curves.
Optimization
SciPy nonlinear least squares, with Levenberg–Marquardt or trust-region reflective selection. Alternative solvers and differential evolution with local refinement are available.
Diagnostics and uncertainty
Residuals, parameter correlations and local Jacobian covariance; profile likelihood, Monte Carlo, residual bootstrap and block bootstrap. Interpretation depends on the model, constraints and noise assumptions.
CUSTOM TOOLS · REPEATABLE PROCEDURES

Extend and automate
your analysis.

Start with the graphical interface, without writing code.
Add custom models and procedures when you need them.

YOUR MODELS

Custom formulas
and function libraries.

Define analytical expressions and reuse your model functions across analyses.

Define a custom function ↗
REPEATABLE WORKFLOWS

Python, CLI
and YAML.

Automate import, preprocessing, fitting and export, or integrate CurveMole with other software.

Python API ↗YAML workflows ↗

AI-assisted automation. Give an AI assistant the Python API audit or YAML workflow audit to help write scripts and workflows. These self-contained interface references avoid loading the entire codebase into the AI’s context, saving tokens. Use the audit matching your installed version, and review the scientific assumptions.

One scientific engine, shared by the desktop interface, Python API, command line and YAML workflows.

TAKE A CLOSER LOOK

Explore the examples.

Inspect fitted models and residuals, download the projects, or compare the interface themes.

Worked examples: spectra, oscillations and kinetics
WORKED EXAMPLES · SYNTHETIC DATA

Weighted nonlinear
least-squares examples

Three reproducible examples of weighted nonlinear least squares.
Data, fitted models and normalized residuals are shown together.

Fits computed with CurveMole. Figures rendered with Matplotlib. These examples use synthetic data and CurveMole’s Python API. The plotting style was created for this website and does not represent the appearance of the CurveMole desktop interface. Error bars represent supplied ±1σ uncertainties in y. Fits minimize Σ[(y − f(x))/σ]² using those uncertainties. The lower panels show normalized residuals; χ²ᵣ is reported with the data count. The application screenshots in the sections above show the actual CurveMole interface.
Actual CurveMole fit to synthetic overlapping Gaussian bands with one-sigma error bars, individual components and normalized residuals
01 / SUM OF GAUSSIANS

Two Gaussians and a linear term

Two Gaussian components and a linear term fitted simultaneously to data with nonuniform uncertainties. Individual components are displayed with the linear term added.

y(x) = b₀ + b₁x + G₁(x) + G₂(x)

The fit reports strong parameter correlation (|r| ≥ 0.95), illustrating why residuals and parameter covariance should be examined together.

Fit: CurveMole · Plot: Matplotlib
Website styling; not an application screenshot.

CurveMole weighted fit of a damped oscillation, showing one-sigma measurement uncertainties and normalized residuals
02 / DAMPED COSINE

Exponentially damped cosine

A user-defined function combines amplitude, frequency, phase, exponential damping and an offset. All five parameters are estimated in one weighted fit.

y(x) = c + A exp(−x/τ) cos(ωx + φ)

A general x–y analysis example: the model is entered as a custom formula using the same engine as the desktop interface.

Fit: CurveMole · Plot: Matplotlib
Website styling; not an application screenshot.

CurveMole weighted fit of a difference of exponentials with one-sigma error bars and normalized residuals
03 / DIFFERENCE OF EXPONENTIALS

Difference of exponentials

A custom formula combines two exponential terms. The amplitude, offset and two exponential coefficients are refined with explicit parameter bounds.

y(x) = c + A [exp(−k₁x) − exp(−k₂x)]

Nonuniform uncertainties enter the fitting objective explicitly. Bounds separate the two exponential coefficients.

Fit: CurveMole · Plot: Matplotlib
Website styling; not an application screenshot.

Fitting engine: CurveMole 0.46.0 · Figure rendering: Matplotlib with a website-specific style · fixed random seed · supplied σ treated as absolute · parameter uncertainties from local covariance. These examples demonstrate the workflow; they are not experimental validation of the software. Reproduce the examples ↓

The same workspace in light and dark
DISPLAY OPTIONS

Light and dark themes

The same illustrative Raman dataset in both display modes.
Drag the divider to compare the interface.

● ● ●CurveMole / Display themesSAME DATA · TWO THEMES
The original CurveMole Raman fitting example in light modeThe same original CurveMole Raman fitting example in dark mode

Illustrative Raman-like data · Matched viewport and model · Actual application screenshots

START WITH YOUR DATA

Your next analysis starts here.

Download CurveMole, import a curve and build your first model.
The quick start walks you through fitting, inspection and export.