Simulink Single-Precision Converter
Convert Simulink models from double to single precision to reduce memory usage.
Installation
- Make sure Claude is on your device and in your terminal.
Skills load from
~/.claude/skills/when Claude Code starts up β so you need it on your machine first. If you don't have it yet, install it once with the command below, then runclaudein any terminal to verify.One-time setupnpm i -g @anthropic-ai/claude-codeAlready have it? Skip ahead.
- Paste into Claude Code or into your terminal.
This copies the whole skill folder into
~/.claude/skills/simulink-single-precision-conversion-matlab/β the SKILL.md plus any scripts, reference docs, or templates the skill ships with. Safe default: works for every skill.Faster alternative (instruction-only skills)
Skips the clone and grabs only the SKILL.md file. Don't use this if the skill ships Python scripts, reference markdowns, or asset templates β they won't be downloaded and the skill will fail when it tries to load them.
Quick install (SKILL.md only)Sign up to copy - Restart Claude Code.
Quit and reopen Claude Code (or any other agent that loads from
~/.claude/skills/). New skills are picked up on startup. - Just ask Claude.
Skills auto-activate when your request matches the skill's description β no slash command needed. Trigger phrases live in the skill's own frontmatter; you can read them in the βWhat this skill doesβ section above.
Prefer to read the source first? Open on GitHub.
When Claude uses it
Converts a double-precision Simulink system or subsystem to single precision using DataTypeWorkflow.Single (Fixed-Point Designer). The single conversion replaces all user-specified double-precision data types, as well as output data types that compile to double precision, with single-precision data types. Use this skill when converting Simulink systems to single precision, reducing memory usage of a Simulink system, optimizing for embedded targets. Do NOT use for standalone MATLAB .m code single conversion.
What this skill does
Single-Precision Conversion for Simulink Models (Fixed-Point Designer)
Toolbox: Fixed-Point Designer
Function: DataTypeWorkflow.Single.convertToSingle
Converts user-specified double-precision data types β across block settings, Stateflow chart settings, signal objects, and bus objects β to single precision. Boolean, built-in integer (int8, uint8, int16, β¦), and fixed-point types are left unchanged.
When to Use
- Converting a Simulink model (
.slx) or subsystem from double-precision to single-precision - Reducing memory footprint for embedded hardware with native single-precision support
- Batch/CI conversions, checking conversion compatibility, or verifying no stowaway doubles remain
- Summarizing the single conversion compatibility report and conversion result for the user in plain language
When NOT to Use
- Standalone MATLAB
.mcode β use theconvertToSinglefunction withcoder.config('single')instead - Fixed-point data type optimization β use
DataTypeWorkflow.Converterinstead - Approximating a function or shrinking Lookup Table blocks with LUTs β use
simulink-optimize-lookup-tablesinstead
Public API
ConversionReport = DataTypeWorkflow.Single.convertToSingle(systemToConvert)
systemToConvert(char) β full block path of the loaded model or subsystem (e.g.'myModel'or'myModel/Controller').ConversionReport(struct) β compatibility check, converted items, and verification status (see field tables below).
Worked example: openExample("fixedpoint/ConvertSystemToSinglePrecisionExample")
Rules
- Open with
open_system(modelName), notload_systemβ the user must see the diagram before/after. - Never call
save_systemon your own β saving destroys the double-precision baseline and blocks meaningful re-runs. Leave the model dirty. This holds even when the user asks you to "make it stick," "persist it," or "do everything end-to-end" β that is not consent to auto-save. In that case, state plainly that you will not save the model yourself, explain that saving overwrites the double-precision baseline, and ask the user to confirm or runsave_systemthemselves. Only persist after the user explicitly confirms saving specifically (not merely a general "do it all" request). - The compatibility check runs inside
convertToSingle; read its results fromreport.CheckInfo(it is not a separate call). - Get the user's explicit consent before converting. If the model path, subsystem, or scope is ambiguous, ask β do not guess. When the user says "my model" without naming one, ask which model; never adopt a
.slxyou happen to find in the working directory (an example or fixture file is not the user's intended model). Only run the conversion once the user has named the model. - Always inspect
report.VerifyInfo.StowawayDblBlksafterward β remaining stowaway doubles mean the conversion is incomplete. Fix them at the source (retype locked-double blocks, set Stateflow data to single, or widen the SUD scope) and re-runconvertToSingle. Do not recommend hand-insertingsingle(...)casts or Data Type Conversion blocks on the offending signals β that hides the double instead of eliminating it. - Assign the report to a local variable (
report = ...;) β don't pollute the base workspace or echo the raw struct.
Report Structure
The tool runs check β convert β verify, populating three sub-structs on the report.
report.CheckInfo
| Field | Description |
|---|---|
ready | Logical β true if the system can be converted - only read ConvertInfo/VerifyInfo when this is true. |
err | Error info (empty on success) |
IncompatibleBlks | Cell array of DataTypeWorkflow.Single.Result β blocks not supporting single precision (see block-name note below) |
UnsupportedBlks | Cell array of DataTypeWorkflow.Single.Result β blocks unsupported by the conversion tool (with ErrorMsgs) |
DTLockedDblBlks | Blocks with locked double data types (do not block conversion) |
StowawayDblBlks | Blocks generating double operations found during the check |
TLSSettings | Models whose Target Language Standard was updated to C99 |
DTOSettings | Systems whose Data Type Override was reset |
SolverSettings | Models whose variable-step solver was changed to fixed-step |
configSettings | Config params updated (e.g. GenerateComments, ParameterPrecisionLossMsg) |
memoryUse.BeforeValue | Per-parameter memory table; column RuntimeMemory (bytes) before conversion |
How to extract block names from the diagnostic list. IncompatibleBlks, UnsupportedBlks, StowawayDblBlks, and DTLockedDblBlks are cell arrays of DataTypeWorkflow.Single.Result objects β not block-path strings. Get the readable block path from each element's ID via getDisplayName(). Do not call get_param, getfullname, or string() on the elements: that errors with "The first input to get_param must be of type 'double', 'char' or 'cell'.":
for k = 1:numel(report.CheckInfo.IncompatibleBlks)
r = report.CheckInfo.IncompatibleBlks{k};
fprintf(' %s\n', r.ID.getDisplayName()); % e.g. 'myModel/Integrator'
if ~isempty(r.ErrorMsgs)
fprintf(' %s\n', strjoin(string(r.ErrorMsgs), '; '));
end
end
report.ConvertInfo
Records what the converter actually did.
| Field | Description |
|---|---|
ready | Logical β true if the convert step ran cleanly |
err | Error info; empty ([]) on success |
results | Cell array of fxptds.BlockResult β the blocks/signals changed from double to single |
report.VerifyInfo
| Field | Description |
|---|---|
StowawayDblBlks | Blocks still generating double operations after conversion |
memoryUse.AfterValue | Per-parameter memory table; column RuntimeMemory (bytes) after conversion |
The check phase validates the Target Language Standard, Data Type Override, incompatible blocks, stowaway doubles, data-type-locked doubles, and solver settings β and auto-updates several of them. The tool handles TLSβC99, DTO cleanup, and solver changes automatically; you do not need to set these manually. See references/edge-cases.md for the full auto-update behavior, error identifiers, and edge cases:
- Model not loaded / invalid path (
Simulink:Commands:InvSimulinkObjectName); update-diagram failures (DataTypeWorkflow:Single:UpdateDiagramFailed) - Incompatible blocks (continuous blocks, MATLAB System blocks) and unsupported MATLAB Function Blocks (globals, MCOS classes, Simulink function calls)
- Auto-updates (TLS, DTO, solver, config), model-reference scope handling, idempotency, and cleanup on model close
Generating the Conversion Report
After convertToSingle returns, summarize the result for the user in plain language β do not
dump the raw struct. Put the exact API call used at the top for traceability.
For the required output sections, empty-report fallback, reporting conventions, and the Embedded Coder readiness-check follow-up, see references/reporting-format.md.
Prerequisites & Limitations
- Requires MATLAB, Simulink, and Fixed-Point Designer; the model must be loaded.
- The conversion engine is a singleton β only one conversion session can be active at a time.
- MATLAB Function Blocks with complex logic and model references may need individual review β see references/edge-cases.md.
Copyright 2026 The MathWorks, Inc.
Related skills
Chai-1
xuzhougeng
Predict protein, nucleic acid, and small-molecule complex structures from FASTA sequences.
Simulink Code Generation Configurator
matlab
Configure Simulink models for embedded C/C++ code generation with ERT, GRT, or AUTOSAR targets.
DiffDock
xuzhougeng
Predict 3D binding poses for small molecules in protein structures.
SLCI Compatibility Analysis
matlab
Check if your Simulink model meets code inspection requirements and fix incompatibilities.