nomogram 0.4.0__tar.gz → 0.6.0__tar.gz

{nomogram-0.4.0 → nomogram-0.6.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nomogram
-Version: 0.4.0
+Version: 0.6.0
 Summary: Repository of scientific formulas: notations, units, validity domains
 Keywords: scientific computing,formulas,units,dimensional analysis,physics,thermodynamics,geoscience,atmospheric science,meteorology,hydrology,psychrometrics
 Author: miki4ia
@@ -22,7 +22,7 @@ Classifier: Topic :: Scientific/Engineering :: Physics
 Classifier: Typing :: Typed
 Requires-Dist: numpy>=2.0
 Requires-Dist: pint>=0.24.4
-Requires-Dist: sympy>=1.13
+Requires-Dist: sympy>=1.14
 Requires-Dist: matplotlib>=3.10.7 ; extra == 'viz'
 Requires-Python: >=3.12
 Project-URL: Homepage, https://github.com/miki4iaml/nomogram
@@ -34,18 +34,18 @@ Description-Content-Type: text/markdown
 
 <p align="center">
   <picture>
-    <source media="(prefers-color-scheme: dark)" srcset="docs/images/banner-dark.svg">
-    <img alt="nomogram" src="docs/images/banner-light.svg" width="480">
+    <source media="(prefers-color-scheme: dark)" srcset="https://raw.githubusercontent.com/miki4iaml/nomogram/v0.6.0/docs/images/banner-dark.svg">
+    <img alt="nomogram" src="https://raw.githubusercontent.com/miki4iaml/nomogram/v0.6.0/docs/images/banner-light.svg" width="480">
   </picture>
 </p>
 
-**English** | [Français](README.fr.md)
+**English** | [Français](https://github.com/miki4iaml/nomogram/blob/v0.6.0/README.fr.md)
 
 [![CI](https://github.com/miki4iaml/nomogram/actions/workflows/ci.yml/badge.svg?branch=develop)](https://github.com/miki4iaml/nomogram/actions/workflows/ci.yml)
 [![CodeQL](https://github.com/miki4iaml/nomogram/actions/workflows/github-code-scanning/codeql/badge.svg)](https://github.com/miki4iaml/nomogram/actions/workflows/github-code-scanning/codeql)
 [![codecov](https://codecov.io/gh/miki4iaml/nomogram/graph/badge.svg?branch=develop)](https://codecov.io/gh/miki4iaml/nomogram)
-[![Python](https://img.shields.io/python/required-version-toml?tomlFilePath=https%3A%2F%2Fraw.githubusercontent.com%2Fmiki4iaml%2Fnomogram%2Fdevelop%2Fpyproject.toml)](pyproject.toml)
-[![License](https://img.shields.io/github/license/miki4iaml/nomogram)](LICENSE)
+[![Python](https://img.shields.io/python/required-version-toml?tomlFilePath=https%3A%2F%2Fraw.githubusercontent.com%2Fmiki4iaml%2Fnomogram%2Fdevelop%2Fpyproject.toml)](https://github.com/miki4iaml/nomogram/blob/v0.6.0/pyproject.toml)
+[![License](https://img.shields.io/github/license/miki4iaml/nomogram)](https://github.com/miki4iaml/nomogram/blob/v0.6.0/LICENSE)
 [![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.23373134.svg)](https://doi.org/10.5281/zenodo.23373134)
 [![uv](https://img.shields.io/endpoint?url=https://raw.githubusercontent.com/astral-sh/uv/main/assets/badge/v0.json)](https://github.com/astral-sh/uv)
 [![Ruff](https://img.shields.io/endpoint?url=https://raw.githubusercontent.com/astral-sh/ruff/main/assets/badge/v2.json)](https://github.com/astral-sh/ruff)
@@ -94,7 +94,7 @@ Python 3.12 or later. Plots need the `viz` extra (matplotlib):
 Every saturation vapour pressure formulation of the catalog against the
 IAPWS reference, from 0 to 60 °C and within 1 %; curves stop where a formulation leaves its
 stated validity domain, and the figure takes the option set of nomogram's
-documentation ([script](docs/examples/saturation_vapour_pressure.py)):
+documentation ([script](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/examples/saturation_vapour_pressure.py)):
 
 ```python
 import numpy as np
@@ -113,7 +113,7 @@ with nomo.display.display_options(nomo.display.DisplayOptions.DOCUMENTATION):
     nomo.plot.plot_deviations(table, xlim=(0, 60), ylim=(-1.0, 1.0))  # extra viz
 ```
 
-![Deviation of saturation vapour pressure formulations from IAPWS](https://raw.githubusercontent.com/miki4iaml/nomogram/v0.4.0/docs/images/saturation-vapour-pressure.png)
+![Deviation of saturation vapour pressure formulations from IAPWS](https://raw.githubusercontent.com/miki4iaml/nomogram/v0.6.0/docs/images/saturation-vapour-pressure.png)
 
 ## Usage
 
@@ -129,8 +129,29 @@ ha = nomo.comparison.compose(
 ha.evaluate(RH=Q_(100, "percent"), T=Q_(20, "degC")).to("g/m^3")  # 17.29
 ```
 
+`compare` gives each value to the input of its quantity kind. A
+formulation with several inputs of one kind, such as a scheme
+interpolating between two levels (the target pressure p and the
+pressures p_1 and p_2 of the levels), compares once all but one of them
+are fixed, which `compare` asks for: `partial` fixes inputs as `functools.partial` fixes arguments,
+here the two levels, and leaves a function of p alone, evaluated,
+compared and plotted like any formulation
+([Humidity between two levels](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/vertical.md#humidity-between-two-levels)).
+It also fixes what is known in advance, such as the levels of a sounding,
+and names the curve in legends:
+
+```python
+levels = {"p_1": Q_(850, "hPa"), "p_2": Q_(500, "hPa"),
+          "T_1": Q_(278, "K"), "T_2": Q_(252, "K")}
+in_ln_p = nomo.comparison.partial(
+    catalog.formulations["log_pressure_interpolation_temperature"],
+    label="T linear in ln p", **levels,
+)
+in_ln_p.evaluate(p=Q_(700, "hPa")).to("K")  # 268.49
+```
+
 Formulations are found by words, wildcards and fields, whatever their
-notation ([Finding formulations](docs/search.md)):
+notation ([Finding formulations](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/search.md)):
 
 ```python
 catalog.search(variable="rho*")                    # ρ in every notation
@@ -138,20 +159,28 @@ catalog.search(kind="VapourPressure", source="buck*")
 ```
 
 Each formulation keeps its units, validity domain, temperature scale and
-references. The vocabulary is mapped in [docs/rosetta.md](docs/rosetta.md);
-[How nomogram earns trust](docs/approach.md) explains what is checked and
-how, and [Displaying formulas](docs/display.md) shows formulas in notebooks,
-Markdown and plots. [Sea water](docs/sea-water.md) shows the salinity
+references. The vocabulary is mapped in [docs/rosetta.md](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/rosetta.md),
+and [Shared vocabularies](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/shared-vocabularies.md) explains how nomogram
+builds on QUDT, CF and Wikidata and contributes to them;
+[How nomogram earns trust](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/approach.md) explains what is checked and
+how, and [Displaying formulas](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/display.md) shows formulas in notebooks,
+Markdown and plots. [Sea water](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/sea-water.md) shows the salinity
 scales and the vapour pressure of sea water, and what its 2 % lowering
-changes at the sea surface. [The logo](docs/logo.md) is a working
-nomogram.
+changes at the sea surface. [Humidity aloft](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/humidity-aloft.md)
+sets the formula of a radiosonde's producer against its users' along a
+vertical profile, through the dew point, over water or over ice, and
+against WMO's requirements.
+[Security and compliance](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/security-and-compliance.md) gathers, for
+those who approve its use on a laboratory network, how nomogram is
+developed, checked and published, what remains to do, and a checklist.
+[The logo](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/logo.md) is a working nomogram.
 
 ## Contributing
 
 References to integrate, information or corrections about catalog
 entries, new formulas and code are all welcome: see the
-[ways to contribute](.github/CONTRIBUTING.md#ways-to-contribute) and
-the [catalog guide](docs/catalog.md).
+[ways to contribute](https://github.com/miki4iaml/nomogram/blob/v0.6.0/.github/CONTRIBUTING.md#ways-to-contribute) and
+the [catalog guide](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/catalog.md).
 
 ## Data sources
 
@@ -162,7 +191,7 @@ their canonical units, from the CF standard name table. The package
 bundles a subset of each source:
 
 - QUDT — Quantities, Units, Dimensions and Types, by QUDT.org
-  (<https://qudt.org>), release 3.5.2, licensed under
+  (<https://qudt.org>), release 3.5.3, licensed under
   [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/);
 - the [CF standard name table](https://cfconventions.org/Data/cf-standard-names/95/build/cf-standard-name-table.html),
   version 95, [CC0 1.0](https://creativecommons.org/publicdomain/zero/1.0/);
@@ -170,9 +199,9 @@ bundles a subset of each source:
   Institute, Inc., [UCUM License](https://ucum.org/license).
 
 What nomogram selects, changes and adds is listed in
-[NOTICE.md](src/nomogram/data/NOTICE.md).
+[NOTICE.md](https://github.com/miki4iaml/nomogram/blob/v0.6.0/src/nomogram/data/NOTICE.md).
 
 ## License
 
-Code: [BSD-3-Clause](LICENSE). Bundled data: see
+Code: [BSD-3-Clause](https://github.com/miki4iaml/nomogram/blob/v0.6.0/LICENSE). Bundled data: see
 [Data sources](#data-sources).

{nomogram-0.4.0 → nomogram-0.6.0}/README.md

@@ -1,17 +1,17 @@
 <p align="center">
   <picture>
-    <source media="(prefers-color-scheme: dark)" srcset="docs/images/banner-dark.svg">
-    <img alt="nomogram" src="docs/images/banner-light.svg" width="480">
+    <source media="(prefers-color-scheme: dark)" srcset="https://raw.githubusercontent.com/miki4iaml/nomogram/v0.6.0/docs/images/banner-dark.svg">
+    <img alt="nomogram" src="https://raw.githubusercontent.com/miki4iaml/nomogram/v0.6.0/docs/images/banner-light.svg" width="480">
   </picture>
 </p>
 
-**English** | [Français](README.fr.md)
+**English** | [Français](https://github.com/miki4iaml/nomogram/blob/v0.6.0/README.fr.md)
 
 [![CI](https://github.com/miki4iaml/nomogram/actions/workflows/ci.yml/badge.svg?branch=develop)](https://github.com/miki4iaml/nomogram/actions/workflows/ci.yml)
 [![CodeQL](https://github.com/miki4iaml/nomogram/actions/workflows/github-code-scanning/codeql/badge.svg)](https://github.com/miki4iaml/nomogram/actions/workflows/github-code-scanning/codeql)
 [![codecov](https://codecov.io/gh/miki4iaml/nomogram/graph/badge.svg?branch=develop)](https://codecov.io/gh/miki4iaml/nomogram)
-[![Python](https://img.shields.io/python/required-version-toml?tomlFilePath=https%3A%2F%2Fraw.githubusercontent.com%2Fmiki4iaml%2Fnomogram%2Fdevelop%2Fpyproject.toml)](pyproject.toml)
-[![License](https://img.shields.io/github/license/miki4iaml/nomogram)](LICENSE)
+[![Python](https://img.shields.io/python/required-version-toml?tomlFilePath=https%3A%2F%2Fraw.githubusercontent.com%2Fmiki4iaml%2Fnomogram%2Fdevelop%2Fpyproject.toml)](https://github.com/miki4iaml/nomogram/blob/v0.6.0/pyproject.toml)
+[![License](https://img.shields.io/github/license/miki4iaml/nomogram)](https://github.com/miki4iaml/nomogram/blob/v0.6.0/LICENSE)
 [![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.23373134.svg)](https://doi.org/10.5281/zenodo.23373134)
 [![uv](https://img.shields.io/endpoint?url=https://raw.githubusercontent.com/astral-sh/uv/main/assets/badge/v0.json)](https://github.com/astral-sh/uv)
 [![Ruff](https://img.shields.io/endpoint?url=https://raw.githubusercontent.com/astral-sh/ruff/main/assets/badge/v2.json)](https://github.com/astral-sh/ruff)
@@ -60,7 +60,7 @@ Python 3.12 or later. Plots need the `viz` extra (matplotlib):
 Every saturation vapour pressure formulation of the catalog against the
 IAPWS reference, from 0 to 60 °C and within 1 %; curves stop where a formulation leaves its
 stated validity domain, and the figure takes the option set of nomogram's
-documentation ([script](docs/examples/saturation_vapour_pressure.py)):
+documentation ([script](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/examples/saturation_vapour_pressure.py)):
 
 ```python
 import numpy as np
@@ -79,7 +79,7 @@ with nomo.display.display_options(nomo.display.DisplayOptions.DOCUMENTATION):
     nomo.plot.plot_deviations(table, xlim=(0, 60), ylim=(-1.0, 1.0))  # extra viz
 ```
 
-![Deviation of saturation vapour pressure formulations from IAPWS](https://raw.githubusercontent.com/miki4iaml/nomogram/v0.4.0/docs/images/saturation-vapour-pressure.png)
+![Deviation of saturation vapour pressure formulations from IAPWS](https://raw.githubusercontent.com/miki4iaml/nomogram/v0.6.0/docs/images/saturation-vapour-pressure.png)
 
 ## Usage
 
@@ -95,8 +95,29 @@ ha = nomo.comparison.compose(
 ha.evaluate(RH=Q_(100, "percent"), T=Q_(20, "degC")).to("g/m^3")  # 17.29
 ```
 
+`compare` gives each value to the input of its quantity kind. A
+formulation with several inputs of one kind, such as a scheme
+interpolating between two levels (the target pressure p and the
+pressures p_1 and p_2 of the levels), compares once all but one of them
+are fixed, which `compare` asks for: `partial` fixes inputs as `functools.partial` fixes arguments,
+here the two levels, and leaves a function of p alone, evaluated,
+compared and plotted like any formulation
+([Humidity between two levels](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/vertical.md#humidity-between-two-levels)).
+It also fixes what is known in advance, such as the levels of a sounding,
+and names the curve in legends:
+
+```python
+levels = {"p_1": Q_(850, "hPa"), "p_2": Q_(500, "hPa"),
+          "T_1": Q_(278, "K"), "T_2": Q_(252, "K")}
+in_ln_p = nomo.comparison.partial(
+    catalog.formulations["log_pressure_interpolation_temperature"],
+    label="T linear in ln p", **levels,
+)
+in_ln_p.evaluate(p=Q_(700, "hPa")).to("K")  # 268.49
+```
+
 Formulations are found by words, wildcards and fields, whatever their
-notation ([Finding formulations](docs/search.md)):
+notation ([Finding formulations](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/search.md)):
 
 ```python
 catalog.search(variable="rho*")                    # ρ in every notation
@@ -104,20 +125,28 @@ catalog.search(kind="VapourPressure", source="buck*")
 ```
 
 Each formulation keeps its units, validity domain, temperature scale and
-references. The vocabulary is mapped in [docs/rosetta.md](docs/rosetta.md);
-[How nomogram earns trust](docs/approach.md) explains what is checked and
-how, and [Displaying formulas](docs/display.md) shows formulas in notebooks,
-Markdown and plots. [Sea water](docs/sea-water.md) shows the salinity
+references. The vocabulary is mapped in [docs/rosetta.md](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/rosetta.md),
+and [Shared vocabularies](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/shared-vocabularies.md) explains how nomogram
+builds on QUDT, CF and Wikidata and contributes to them;
+[How nomogram earns trust](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/approach.md) explains what is checked and
+how, and [Displaying formulas](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/display.md) shows formulas in notebooks,
+Markdown and plots. [Sea water](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/sea-water.md) shows the salinity
 scales and the vapour pressure of sea water, and what its 2 % lowering
-changes at the sea surface. [The logo](docs/logo.md) is a working
-nomogram.
+changes at the sea surface. [Humidity aloft](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/humidity-aloft.md)
+sets the formula of a radiosonde's producer against its users' along a
+vertical profile, through the dew point, over water or over ice, and
+against WMO's requirements.
+[Security and compliance](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/security-and-compliance.md) gathers, for
+those who approve its use on a laboratory network, how nomogram is
+developed, checked and published, what remains to do, and a checklist.
+[The logo](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/logo.md) is a working nomogram.
 
 ## Contributing
 
 References to integrate, information or corrections about catalog
 entries, new formulas and code are all welcome: see the
-[ways to contribute](.github/CONTRIBUTING.md#ways-to-contribute) and
-the [catalog guide](docs/catalog.md).
+[ways to contribute](https://github.com/miki4iaml/nomogram/blob/v0.6.0/.github/CONTRIBUTING.md#ways-to-contribute) and
+the [catalog guide](https://github.com/miki4iaml/nomogram/blob/v0.6.0/docs/catalog.md).
 
 ## Data sources
 
@@ -128,7 +157,7 @@ their canonical units, from the CF standard name table. The package
 bundles a subset of each source:
 
 - QUDT — Quantities, Units, Dimensions and Types, by QUDT.org
-  (<https://qudt.org>), release 3.5.2, licensed under
+  (<https://qudt.org>), release 3.5.3, licensed under
   [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/);
 - the [CF standard name table](https://cfconventions.org/Data/cf-standard-names/95/build/cf-standard-name-table.html),
   version 95, [CC0 1.0](https://creativecommons.org/publicdomain/zero/1.0/);
@@ -136,9 +165,9 @@ bundles a subset of each source:
   Institute, Inc., [UCUM License](https://ucum.org/license).
 
 What nomogram selects, changes and adds is listed in
-[NOTICE.md](src/nomogram/data/NOTICE.md).
+[NOTICE.md](https://github.com/miki4iaml/nomogram/blob/v0.6.0/src/nomogram/data/NOTICE.md).
 
 ## License
 
-Code: [BSD-3-Clause](LICENSE). Bundled data: see
+Code: [BSD-3-Clause](https://github.com/miki4iaml/nomogram/blob/v0.6.0/LICENSE). Bundled data: see
 [Data sources](#data-sources).

{nomogram-0.4.0 → nomogram-0.6.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "nomogram"
-version = "0.4.0"
+version = "0.6.0"
 description = "Repository of scientific formulas: notations, units, validity domains"
 readme = "README.md"
 requires-python = ">=3.12"
@@ -39,7 +39,7 @@ classifiers = [
 dependencies = [
     "numpy>=2.0",
     "pint>=0.24.4",
-    "sympy>=1.13",
+    "sympy>=1.14",
 ]
 
 [[project.authors]]
@@ -56,7 +56,7 @@ Changelog = "https://github.com/miki4iaml/nomogram/blob/main/CHANGELOG.md"
 viz = ["matplotlib>=3.10.7"]
 
 [build-system]
-requires = ["uv_build>=0.12.18,<0.13.0"]
+requires = ["uv_build>=0.12.18,<0.14.0"]
 build-backend = "uv_build"
 
 [dependency-groups]
@@ -133,8 +133,24 @@ extend-select = [
     "PTH",
     "RET",
     "T20",
+    "TID251",
 ]
 
+[tool.ruff.lint.flake8-tidy-imports.banned-api."sympy.sympify"]
+msg = "read formulas with nomogram.expression.parse_expression"
+
+[tool.ruff.lint.flake8-tidy-imports.banned-api."sympy.core.sympify"]
+msg = "read formulas with nomogram.expression.parse_expression"
+
+[tool.ruff.lint.flake8-tidy-imports.banned-api."sympy.parse_expr"]
+msg = "read formulas with nomogram.expression.parse_expression"
+
+[tool.ruff.lint.flake8-tidy-imports.banned-api."sympy.parsing"]
+msg = "read formulas with nomogram.expression.parse_expression"
+
+[tool.ruff.lint.flake8-tidy-imports.banned-api."sympy.parsing.sympy_parser"]
+msg = "read formulas with nomogram.expression.parse_expression"
+
 [tool.ruff.lint.per-file-ignores]
 "tests/**" = ["ARG001"]
 "scripts/**" = ["T20"]

{nomogram-0.4.0 → nomogram-0.6.0}/pyproject.toml.orig

@@ -1,6 +1,6 @@
 [project]
 name = "nomogram"
-version = "0.4.0"
+version = "0.6.0"
 description = "Repository of scientific formulas: notations, units, validity domains"
 readme = "README.md"
 authors = [
@@ -39,7 +39,7 @@ classifiers = [
 dependencies = [
     "numpy>=2.0",
     "pint>=0.24.4",
-    "sympy>=1.13",
+    "sympy>=1.14",
 ]
 
 [project.urls]
@@ -54,7 +54,7 @@ viz = [
 ]
 
 [build-system]
-requires = ["uv_build>=0.12.18,<0.13.0"]
+requires = ["uv_build>=0.12.18,<0.14.0"]
 build-backend = "uv_build"
 
 [dependency-groups]
@@ -110,7 +110,16 @@ exclude = ["*.md"]
 
 [tool.ruff.lint]
 # On top of ruff's default selection, which Dependabot bumps surface in CI.
-extend-select = ["ARG", "B", "C4", "PIE", "PT", "PTH", "RET", "T20"]
+extend-select = ["ARG", "B", "C4", "PIE", "PT", "PTH", "RET", "T20", "TID251"]
+
+# A formula is read by nomogram.expression.parse_expression, against a
+# whitelist: never by SymPy's parsers, which rely on eval (AGENTS.md).
+[tool.ruff.lint.flake8-tidy-imports.banned-api]
+"sympy.sympify".msg = "read formulas with nomogram.expression.parse_expression"
+"sympy.core.sympify".msg = "read formulas with nomogram.expression.parse_expression"
+"sympy.parse_expr".msg = "read formulas with nomogram.expression.parse_expression"
+"sympy.parsing".msg = "read formulas with nomogram.expression.parse_expression"
+"sympy.parsing.sympy_parser".msg = "read formulas with nomogram.expression.parse_expression"
 
 [tool.ruff.lint.per-file-ignores]
 # Fixtures requested for their side effects look like unused arguments.

{nomogram-0.4.0 → nomogram-0.6.0}/src/nomogram/catalog.py

@@ -87,6 +87,9 @@ from nomogram.model import (
     _origin,
 )
 from nomogram.relations import (
+    _alike,
+    _head,
+    _shape,
     _unit_factor,
     match_inputs,
     record,
@@ -292,8 +295,8 @@ class Catalog:
         >>> c = catalog()
         >>> "formulations.buck1981_magnus" in c.search("buck*", section="formulations")
         True
-        >>> c.search(variable="rho*").refs[:2]
-        ('formulations.mass_concentration_ideal_gas', 'formulations.itu_r_p453_vapour_pressure_from_density')
+        >>> "formulations.mass_concentration_ideal_gas" in c.search(variable="rho*")
+        True
         """
         from nomogram.search import search
 
@@ -1014,7 +1017,7 @@ class _Builder:
         names = {
             self.constants[n].identifier: n for n in self.constants[key].components
         }
-        return expr.subs({s: value(names[s.name]) for s in expr.free_symbols})
+        return expr.xreplace({s: value(names[s.name]) for s in expr.free_symbols})
 
     def unrounded(self, key: str) -> Any:
         """The exact value of a constant: its definition, recursively, else as printed."""
@@ -1713,6 +1716,70 @@ class _Builder:
             raise CatalogError(msg)
         return formulations[name]
 
+    @staticmethod
+    def bindings(
+        place: str, table: Any, inner: Mapping[str, Formulation]
+    ) -> dict[str, dict[str, str]]:
+        """The renamings of the inputs of each inner formulation, by its identifier.
+
+        ``bind = { theta = "t" }`` renames an input of every inner
+        formulation; ``bind = { theta_1 = { T = "T_1" } }`` those of the
+        formulation computing ``theta_1`` only, so that one formulation may
+        compute two inputs at different arguments.
+        """
+        if table is None:
+            table = {}
+        if not isinstance(table, dict):
+            msg = f"{place}.bind: expected a table"
+            raise CatalogError(msg)
+        every: dict[str, str] = {}
+        own: dict[str, dict[str, str]] = {}
+
+        def target(where: str, name: Any) -> str:
+            if not isinstance(name, str) or not name:
+                msg = f"{where}: expected the identifier of an input"
+                raise CatalogError(msg)
+            if name in inner:
+                msg = f"{where}: {name!r} is computed by {inner[name].key}"
+                raise CatalogError(msg)
+            return name
+
+        for source, name in table.items():
+            if isinstance(name, dict):
+                part = inner.get(source)
+                if part is None:
+                    msg = f"{place}.bind.{source}: no inner formulation computes {source!r}"
+                    raise CatalogError(msg)
+                names = {v.identifier for v in part.inputs}
+                for input_name, renamed in name.items():
+                    if input_name not in names:
+                        msg = (
+                            f"{place}.bind.{source}: {input_name!r} is not an "
+                            f"input of {part.key}"
+                        )
+                        raise CatalogError(msg)
+                    own.setdefault(source, {})[input_name] = target(
+                        f"{place}.bind.{source}", renamed
+                    )
+                continue
+            if not any(
+                source in {v.identifier for v in p.inputs} for p in inner.values()
+            ):
+                msg = (
+                    f"{place}.bind: {source!r} is not an input of an inner formulation"
+                )
+                raise CatalogError(msg)
+            every[source] = target(f"{place}.bind", name)
+        renames = {}
+        for identifier, part in inner.items():
+            rename = {**every, **own.get(identifier, {})}
+            renames[identifier] = {
+                v.identifier: rename[v.identifier]
+                for v in part.inputs
+                if v.identifier in rename
+            }
+        return renames
+
     def composed(
         self, key: str, table: Any, formulations: Mapping[str, Formulation]
     ) -> Formulation:
@@ -1741,6 +1808,7 @@ class _Builder:
             if isinstance(node, sympy.Piecewise)
         }
         inner: dict[str, Formulation] = {}
+        scaled: dict[str, Variable] = {}
         for identifier, name in c["inner"].items():
             w = f"{place}.inner.{identifier}"
             part = self.part(w, name, formulations)
@@ -1769,33 +1837,26 @@ class _Builder:
                 )
                 raise CatalogError(msg)
             if part.output.scale != target.scale:
-                msg = (
-                    f"{w}: {part.key} gives a value on scale "
-                    f"{getattr(part.output.scale, 'key', None)}, but {identifier} "
-                    f"is on {getattr(target.scale, 'key', None)}"
-                )
-                raise CatalogError(msg)
+                # A part on no scale (an interpolation, a definition) feeds an
+                # input on the reference scale, where nothing is converted: its
+                # temperatures take that scale. On another scale, converting
+                # its inputs would not convert what it computes.
+                if (
+                    part.output.scale is None
+                    and target.scale is not None
+                    and target.scale.is_reference
+                    and not any(v.scale is not None for v in part.inputs)
+                ):
+                    scaled[identifier] = target
+                else:
+                    msg = (
+                        f"{w}: {part.key} gives a value on scale "
+                        f"{getattr(part.output.scale, 'key', None)}, but {identifier} "
+                        f"is on {getattr(target.scale, 'key', None)}"
+                    )
+                    raise CatalogError(msg)
             inner[identifier] = part
-        bind = dict(c["bind"] or {})
-        for source, name in bind.items():
-            if not any(
-                source in {v.identifier for v in p.inputs} for p in inner.values()
-            ):
-                msg = (
-                    f"{place}.bind: {source!r} is not an input of an inner formulation"
-                )
-                raise CatalogError(msg)
-            if name in inner:
-                msg = f"{place}.bind: {name!r} is computed by {inner[name].key}"
-                raise CatalogError(msg)
-        renames = {
-            identifier: {
-                v.identifier: bind[v.identifier]
-                for v in part.inputs
-                if v.identifier in bind
-            }
-            for identifier, part in inner.items()
-        }
+        renames = self.bindings(place, c["bind"], inner)
 
         # Inputs: those of the outer formulation it still needs, then those of
         # the inner ones; one identifier is one input, alike in every part.
@@ -1804,22 +1865,40 @@ class _Builder:
         pieces = [(outer, v) for v in outer.inputs if v.identifier not in inner]
         for identifier, part in inner.items():
             rename = renames[identifier]
-            pieces += [
-                (
-                    part,
-                    dataclasses.replace(
-                        v, identifier=rename.get(v.identifier, v.identifier)
-                    ),
+            fed = scaled.get(identifier)
+            for v in part.inputs:
+                v = dataclasses.replace(
+                    v, identifier=rename.get(v.identifier, v.identifier)
                 )
-                for v in part.inputs
-            ]
+                if v.identifier in inner:
+                    # One identifier would name two quantities: an input of
+                    # this part, and what another part computes.
+                    msg = (
+                        f"{where}: input {v.identifier!r} of {part.key} is "
+                        f"computed by {inner[v.identifier].key}; rename it with bind"
+                    )
+                    raise CatalogError(msg)
+                if (
+                    fed is not None
+                    and fed.scale is not None
+                    and fed.scale.applies_to(v.kind, self.kinds)
+                ):
+                    v = dataclasses.replace(
+                        v, scale=fed.scale, scale_basis=fed.scale_basis
+                    )
+                pieces.append((part, v))
         for piece, variable in pieces:
             came_from.setdefault(variable.identifier, piece.key)
             known = inputs.setdefault(variable.identifier, variable)
-            if (known.kind, known.unit, known.scale) != (
-                variable.kind,
-                variable.unit,
-                variable.scale,
+            # An input on no scale merges with one on the reference scale
+            # only: on another, its part would receive the converted value.
+            other = variable.scale if known.scale is None else known.scale
+            if (known.kind, known.unit) != (variable.kind, variable.unit) or (
+                known.scale != variable.scale
+                and (
+                    None not in (known.scale, variable.scale)
+                    or not getattr(other, "is_reference", False)
+                )
             ):
                 msg = (
                     f"{where}: input {variable.identifier} of {piece.key} differs "
@@ -1827,6 +1906,12 @@ class _Builder:
                     "another part; rename it with bind"
                 )
                 raise CatalogError(msg)
+            if known.scale is None and variable.scale is not None:
+                # An input on no scale takes any: that of the other part.
+                inputs[variable.identifier] = dataclasses.replace(
+                    known, scale=variable.scale, scale_basis=variable.scale_basis
+                )
+                came_from[variable.identifier] = piece.key
 
         # Part, and formulation that defines it, of each coefficient.
         owner: dict[str, tuple[str, str]] = {}
@@ -2737,6 +2822,11 @@ class _Builder:
                 for name in part.rename.values()
                 if name not in inputs
             }
+        elif formulation.parts:
+            # A composition: checked on the parts substituted as written,
+            # since SymPy may simplify a factor of two calls of a part away.
+            tree = ast.parse(formulation.expression, mode="eval")
+            used = {node.id for node in ast.walk(tree) if isinstance(node, ast.Name)}
         unused = sorted(set(names) - used)
         if unused:
             msg = f"{where}: declared but not used: {', '.join(unused)}"
@@ -2790,8 +2880,27 @@ class _Builder:
                 built[key] = self.composed(key, tables[key], built)
         formulations = {key: built[key] for key in tables}
         # Two sources of one formulation make one entry with two citations.
+        # Only formulations of the same head (kinds, units, coefficients),
+        # then of alike shapes, may be one: the costly signature, which
+        # tries every order of the inputs, is computed for those alone.
+        heads: dict[str, list[str]] = {}
+        for key, f in formulations.items():
+            heads.setdefault(_head(f), []).append(key)
+        candidates = []
+        for keys in heads.values():
+            shapes = {key: _shape(formulations[key]) for key in keys if len(keys) > 1}
+            candidates += [
+                key
+                for key in shapes
+                if any(
+                    _alike(shapes[key], shapes[other])
+                    for other in shapes
+                    if other != key
+                )
+            ]
         seen: dict[str, str] = {}
-        for key, formulation in formulations.items():
+        for key in candidates:
+            formulation = formulations[key]
             fingerprint = signature(formulation)
             if fingerprint in seen:
                 first = seen[fingerprint]