"Use when PySR needs TemplateExpressionSpec, parametric templates,
Scanned 9/9/2026
Install to Claude Code
npx -y skills add VectorSpaceLab/AREX-Skill --skill structured-expressions --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Structured Expressions?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/vectorspacelab-structured-expressions)More formats (shields.io, HTML) on the badges page.
---
name: structured-expressions
description: "Use when PySR needs TemplateExpressionSpec, parametric templates,
shared or vector-valued expressions, differential operators, or seeded
template guesses."
disable-model-invocation: true
metadata:
disco-role: operating
license: Apache 2.0
---
# Structured expressions
Use this sub-skill when the equation shape is partly known and PySR should search only the unknown pieces.
## Use this sub-skill for
- `TemplateExpressionSpec` planning or repair
- parametric expressions with category-specific learned constants
- shared subexpressions or vector-valued residual templates
- differential templates that use `D`
- seeded guesses for known template placeholders
## Route to another sub-skill when
- the task is a free-form fit without known structure → `fit-and-diagnose`
- the task needs custom losses, operator constraints, or tree walks → `customization-and-constraints`
- the task needs SymPy / LaTeX / JAX / PyTorch exports or reload handling → `export-and-artifacts`
## Operating rules
1. Keep the template shape explicit: `combine`, `expressions`, `variable_names`, and optional `parameters`.
2. Treat category indices as Julia indices inside the template. If the source labels are 0-based, shift them before `fit`.
3. For vector-valued templates, move the extra targets into `X`, use a dummy `y`, and make the elementwise loss return the residual from the template.
4. For differential templates, use `D` and verify the argument index.
5. Do not promise template export methods that are unsupported. Use `model.equations_["julia_expression"]` and component trees instead.
6. If the template also needs a custom objective, hand off to `customization-and-constraints` and use `loss_function_expression`.
## Bundled assets
- `references/template-workflows.md`
- `references/api-reference.md`
- `references/troubleshooting.md`
- `scripts/template_spec_builder.py`
## Verification target
A later agent should be able to build a valid template plan, explain indexing and parameters, seed guesses correctly, and identify unsupported exports without running a search.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!