On this page
Numerical and statistical methods
Autoplot's built-in analysis cards wrap established numerical libraries rather than opaque heuristics. When reporting fitted parameters or distribution curves in a methods section, consult the underlying reference definitions:
- Nonlinear least squares: SciPy curve_fit documentation covers trust-region reflective and Levenberg–Marquardt optimisation, parameter covariance matrices, and bounds; see Autoplot's nonlinear curve fitting guide.
- Histogram density conventions: NumPy histogram reference defines bin-edge inclusion, unequal-width density normalisation (
density=True), and weight handling; paired with histogram normalization and log binning. - Empirical cumulative curves: SciPy empirical CDF and survival function (ecdf) documents step-function empirical cumulative and complementary cumulative estimates; paired with how to make a CCDF plot.
- Power-law tail inference: Clauset, Shalizi and Newman (2009), Power-law distributions in empirical data sets out continuous maximum-likelihood tail estimation, Kolmogorov–Smirnov xmin selection, bootstrap goodness-of-fit testing, and likelihood-ratio model comparison; paired with how to fit a power law to data.
- Heavy-tail scaling: Newman (2005), Power laws, Pareto distributions and Zipf's law explains why logarithmic binning and cumulative tail views behave differently on heavy-tailed measurements, and how exponent conventions relate across PDF and CCDF forms.
- Tail-fitting library: Python powerlaw package is the library behind Autoplot's Log-space CCDF tail fit and xmin diagnostic scan.
- Exploratory diagnostics: NIST/SEMATECH e-Handbook of Statistical Methods provides reference procedures on residual diagnostics and histogram construction; paired with how to interpret residual plots.
Journal figure and PDF verification
Exporting a plot is only half of figure preparation; verifying that vector geometry, font subsets, and physical column widths survive submission prevents late production rewrites:
- Column widths and typography: Nature Research Figure Guide specifies single, 1.5-column, and double-column widths, minimum type sizes at final print scale, and vector-versus-raster rules; paired with how to make a multi-panel figure.
- Vector graph delivery: Nature Communications submission instructions detail vector file expectations for quantitative graphs and schematics; paired with how to export a vector PDF figure.
- Font and raster inspection: Poppler command-line utilities (
pdffontsandpdfimages) verify locally that every font in a figure PDF is embedded and no unexpected low-DPI raster layer sits behind vector axes; paired with how to embed fonts in a figure PDF. - Pre-submission preflight: Adobe Acrobat Preflight checks colour spaces, transparency flattening, font subsets, and effective image resolution.
Mac scientific software and platform notes
Researchers choosing or migrating plotting workflows on Apple silicon Macs often pair a native figure workspace with code libraries or keep legacy licences for specialist instruments:
- Binary translation on M-series Macs: Apple Silicon Rosetta 2 documentation explains how macOS runs Intel-only binaries when maintaining older scientific software.
- Igor Pro on macOS: WaveMetrics Igor Pro macOS update documents the status of Igor Pro 9.05 on macOS and Windows-only Igor Pro 10; paired with Igor Pro alternative for Mac.
- Origin on macOS: OriginLab Origin on Mac outlines virtual-machine routes for Windows-only Origin and OriginPro projects; paired with OriginPro alternative for Mac.
- Scripted figure handoff: Matplotlib documentation is the reference for editing and extending Matplotlib scripts exported from Autoplot; paired with Jupyter alternative for scientific plotting.
- Interactive web charts: Plotly Python graphing library covers browser-interactive charting when colleagues need live filters rather than fixed print dimensions; paired with Plotly Studio alternative for scientists.
Coursework, theses and academic AI boundaries
Graduate and undergraduate students using software for lab reports, problem sets, and dissertations have to separate numerical calculation from prose generation, and check what their department's academic-integrity policy requires them to disclose.
- Local computation vs cloud assistants: Autoplot runs imports, variables, fits, and plots locally on your Mac without sending raw measurement tables to a cloud model; read what leaves your machine and Autoplot for students before using the optional assistant on unpublished thesis data.
- Reproducible coursework records: When an instructor or supervisor asks how a curve or exponent was obtained, export the figure's Matplotlib script or save the assistant's generated Python alongside the figure rather than pasting unverified numbers from a chat window.
- Broader study, literature, and dissertation tools: Plotting and curve fitting cover the quantitative figure side of a thesis; for comparing literature-review tools, note-taking assistants, and university AI-disclosure rules across the rest of an academic degree, see Jobbit's guide to AI for Students in 2026: Coursework, Research and Dissertations.