diskmelts 2.2.1__tar.gz → 2.3.0__tar.gz

{diskmelts-2.2.1/src/diskmelts.egg-info → diskmelts-2.3.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: diskmelts
-Version: 2.2.1
+Version: 2.3.0
 Summary: Machine-learning Enabled Line fitting Tool for Spectra of protoplanetary disks
 Author: Chengyan Xie, Dingshan Deng
 License-Expression: MIT
@@ -12,7 +12,7 @@ Keywords: astronomy,protoplanetary disks,JWST,spectral fitting,machine learning
 Classifier: Programming Language :: Python :: 3
 Classifier: Operating System :: OS Independent
 Classifier: Topic :: Scientific/Engineering :: Astronomy
-Requires-Python: >=3.11
+Requires-Python: >=3.9
 Description-Content-Type: text/markdown
 License-File: LICENSE
 Requires-Dist: torch>=1.13
@@ -20,8 +20,9 @@ Requires-Dist: numpy<2,>=1.26; sys_platform == "darwin" and platform_machine ==
 Requires-Dist: numpy; sys_platform != "darwin" or platform_machine != "x86_64"
 Requires-Dist: pandas
 Requires-Dist: scipy
-Requires-Dist: scikit-learn<1.9,>=1.8
+Requires-Dist: scikit-learn>=1.0
 Requires-Dist: matplotlib
+Requires-Dist: emcee>=3
 Provides-Extra: io
 Requires-Dist: astropy; extra == "io"
 Provides-Extra: dev
@@ -50,7 +51,7 @@ density (`logN`), and emitting area scaling (`A`) from observed spectra.
 ## Installation
 
 DiskMELTS is available on [PyPI](https://pypi.org/project/diskmelts/) and
-requires Python 3.11 or newer. It bundles 46 production checkpoints, 23 each
+requires Python 3.9 or newer. It bundles 46 production checkpoints, 23 each
 for the sigma=0 and sigma=2 grids, covering the same 18 species (see
 [Available molecules](#available-molecules)).
 Install it from any directory. These commands use
@@ -107,9 +108,11 @@ fit = fit_stage(
     wave, flux, molecule="C2H2", pretrained=models,
     wavelength_range=(13.0, 15.0), fit_ranges=(13.5, 14.1),
     distance_pc=140.0, n_samples=1024, n_refine=4, n_top=4,
+    n_mcmc=20000,  # quick look; omit for the converged default (~1 min)
+    # errorbars=False,  # skip the MCMC error bars and return the best fit only
 )
 
-# 3. Best-fit T (K), logN (cm^-2), A (au^2), and the spread of top solutions.
+# 3. Best fit, then posterior medians and 16-84% error bars (from MCMC).
 print(fit["params"]["C2H2"])
 print(fit["uncertainty"]["C2H2"])
 ```
@@ -126,8 +129,9 @@ covers multi-stage fits, saved outputs, and plots.
 | `numpy` | Array operations |
 | `pandas` | CSV I/O |
 | `scipy` | L-BFGS-B optimizer, NNLS, Sobol sampling |
-| `scikit-learn` | Model scaling and PCA metadata |
+| `scikit-learn` | PCA and scaling when training new models |
 | `matplotlib` | Plotting |
+| `emcee` | MCMC sampling of parameter error bars |
 
 ## Continuum subtraction
 
@@ -227,6 +231,14 @@ makes the returned emitting area physical. `fit_observation` accepts either a
 CSV path or in-memory `obs_wav`, `obs_flux`, and optional `obs_err` arrays
 (new in 2.2.1).
 
+Every fit returns its best fit and, by default, error bars from MCMC sampling
+of the posterior with [emcee](https://emcee.readthedocs.io/). Reported values
+are posterior medians with 16–84% error bars; the noise level accounts for
+correlated pixels and model mismatch, and a `*` marks a range that reaches a
+fit bound. Sampling takes about a minute per stage. Pass `errorbars=False`
+for best fits only. See
+[What the uncertainties mean](https://diskmelts.readthedocs.io/en/latest/quickstart.html#uncertainties).
+
 For a synthetic spectrum, `wave, flux = diskmelts.synthesize_spectrum('HCN',
 T=300, logN=18, A=1)` returns the native wavelength axis and model flux (new
 in 2.2.1). See the
@@ -270,8 +282,11 @@ assumes 2 km/s turbulent broadening. Select sigma=2 with `load_fitting_models(..
 | OH | 15–26 | 100–1400 | 13–19.5 |
 
 `diskmelts.MODEL_SPECS` records the exact domains, PCA sizes, and network
-layers. Version 2.2.0 lacks the sigma=0 OH and 6–9 µm C2H2 checkpoints;
-upgrade with `python -m pip install -U diskmelts`.
+layers. Version 2.3.0 retrains 26 of these checkpoints (13C12CH2, 13CO2,
+C2H2 11.5–17 µm, C2H4, C2H6, C2HD, C4H2, CH4, CO17O, CO18O, CO2, HC3N, and HCN at
+both sigmas) with lower held-out error; see the
+[changelog](https://diskmelts.readthedocs.io/en/latest/changelog.html).
+Upgrade with `python -m pip install -U diskmelts`.
 
 ## Training and validation data
 

{diskmelts-2.2.1 → diskmelts-2.3.0}/README.md

@@ -11,7 +11,7 @@ density (`logN`), and emitting area scaling (`A`) from observed spectra.
 ## Installation
 
 DiskMELTS is available on [PyPI](https://pypi.org/project/diskmelts/) and
-requires Python 3.11 or newer. It bundles 46 production checkpoints, 23 each
+requires Python 3.9 or newer. It bundles 46 production checkpoints, 23 each
 for the sigma=0 and sigma=2 grids, covering the same 18 species (see
 [Available molecules](#available-molecules)).
 Install it from any directory. These commands use
@@ -68,9 +68,11 @@ fit = fit_stage(
     wave, flux, molecule="C2H2", pretrained=models,
     wavelength_range=(13.0, 15.0), fit_ranges=(13.5, 14.1),
     distance_pc=140.0, n_samples=1024, n_refine=4, n_top=4,
+    n_mcmc=20000,  # quick look; omit for the converged default (~1 min)
+    # errorbars=False,  # skip the MCMC error bars and return the best fit only
 )
 
-# 3. Best-fit T (K), logN (cm^-2), A (au^2), and the spread of top solutions.
+# 3. Best fit, then posterior medians and 16-84% error bars (from MCMC).
 print(fit["params"]["C2H2"])
 print(fit["uncertainty"]["C2H2"])
 ```
@@ -87,8 +89,9 @@ covers multi-stage fits, saved outputs, and plots.
 | `numpy` | Array operations |
 | `pandas` | CSV I/O |
 | `scipy` | L-BFGS-B optimizer, NNLS, Sobol sampling |
-| `scikit-learn` | Model scaling and PCA metadata |
+| `scikit-learn` | PCA and scaling when training new models |
 | `matplotlib` | Plotting |
+| `emcee` | MCMC sampling of parameter error bars |
 
 ## Continuum subtraction
 
@@ -188,6 +191,14 @@ makes the returned emitting area physical. `fit_observation` accepts either a
 CSV path or in-memory `obs_wav`, `obs_flux`, and optional `obs_err` arrays
 (new in 2.2.1).
 
+Every fit returns its best fit and, by default, error bars from MCMC sampling
+of the posterior with [emcee](https://emcee.readthedocs.io/). Reported values
+are posterior medians with 16–84% error bars; the noise level accounts for
+correlated pixels and model mismatch, and a `*` marks a range that reaches a
+fit bound. Sampling takes about a minute per stage. Pass `errorbars=False`
+for best fits only. See
+[What the uncertainties mean](https://diskmelts.readthedocs.io/en/latest/quickstart.html#uncertainties).
+
 For a synthetic spectrum, `wave, flux = diskmelts.synthesize_spectrum('HCN',
 T=300, logN=18, A=1)` returns the native wavelength axis and model flux (new
 in 2.2.1). See the
@@ -231,8 +242,11 @@ assumes 2 km/s turbulent broadening. Select sigma=2 with `load_fitting_models(..
 | OH | 15–26 | 100–1400 | 13–19.5 |
 
 `diskmelts.MODEL_SPECS` records the exact domains, PCA sizes, and network
-layers. Version 2.2.0 lacks the sigma=0 OH and 6–9 µm C2H2 checkpoints;
-upgrade with `python -m pip install -U diskmelts`.
+layers. Version 2.3.0 retrains 26 of these checkpoints (13C12CH2, 13CO2,
+C2H2 11.5–17 µm, C2H4, C2H6, C2HD, C4H2, CH4, CO17O, CO18O, CO2, HC3N, and HCN at
+both sigmas) with lower held-out error; see the
+[changelog](https://diskmelts.readthedocs.io/en/latest/changelog.html).
+Upgrade with `python -m pip install -U diskmelts`.
 
 ## Training and validation data
 

{diskmelts-2.2.1 → diskmelts-2.3.0}/pyproject.toml

@@ -4,10 +4,10 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "diskmelts"
-version = "2.2.1"
+version = "2.3.0"
 description = "Machine-learning Enabled Line fitting Tool for Spectra of protoplanetary disks"
 readme = "README.md"
-requires-python = ">=3.11"
+requires-python = ">=3.9"
 license = "MIT"
 license-files = ["LICENSE"]
 authors = [
@@ -21,17 +21,18 @@ classifiers = [
     "Topic :: Scientific/Engineering :: Astronomy",
 ]
 dependencies = [
-    # Checkpoints load with torch.load(weights_only=False), added in 1.13.
+    # Checkpoints load with torch.load(weights_only=True), added in 1.13.
     "torch>=1.13",
     # Intel macOS wheels stop at PyTorch 2.2, whose NumPy bridge needs NumPy 1.x.
     "numpy>=1.26,<2; sys_platform == 'darwin' and platform_machine == 'x86_64'",
     "numpy; sys_platform != 'darwin' or platform_machine != 'x86_64'",
     "pandas",
     "scipy",
-    # Bundled checkpoints pickle scikit-learn 1.8 PCA/scaler objects; keep the
-    # same minor version so they unpickle without a cross-version warning.
-    "scikit-learn>=1.8,<1.9",
+    # Used only to fit PCA/scalers when training; checkpoints store plain arrays.
+    "scikit-learn>=1.0",
     "matplotlib",
+    # Ensemble MCMC for posterior error bars (pure Python, NumPy only).
+    "emcee>=3",
 ]
 
 [project.urls]

{diskmelts-2.2.1 → diskmelts-2.3.0}/src/diskmelts/__init__.py

@@ -12,6 +12,8 @@ from diskmelts.trainmodel import (
     train_model_bank,
     select_grid_holdout,
     tune_pca_components,
+    convert_legacy_checkpoint,
+    grid_sha256,
 )
 from diskmelts.fitting import (
     load_models,
@@ -57,6 +59,9 @@ from diskmelts.v2_read_plot import (
 from diskmelts.v2_paper_comparison import plot_literature_comparisons
 from diskmelts import consub
 
+# fit_nested, validate_nt, and plot_validation_split stay importable for
+# existing scripts but are deprecated since 2.3.0 and left out of __all__.
+
 __all__ = [
     "consub",
     "MODEL_SPECS",
@@ -78,11 +83,12 @@ __all__ = [
     "train_model_bank",
     "select_grid_holdout",
     "tune_pca_components",
+    "convert_legacy_checkpoint",
+    "grid_sha256",
     # fitting
     "load_models",
     "load_model_bank",
     "generate_spectrum",
-    "fit_nested",
     "fit_molecules",
     "load_observed_spectrum",
     "save_fit_outputs",
@@ -91,12 +97,10 @@ __all__ = [
     "print_fit_params",
     "save_fitted_comparison",
     # validation
-    "validate_nt",
     "validate_nt_holdout",
     "validate_full",
     # plotting
     "plot_fit",
     "plot_validation",
-    "plot_validation_split",
     "plot_bank_reconstruction",
 ]