How far is the thermostat-driven infrared-gating textile from the best possible switching schedule when the room temperature steps?

What it found

A stylised, preliminary follow-up to garment-twonodestep. Exhaustive search over open/closed schedules (at most 3 toggles on a 10 min grid, instant actuation) in the switchable two-node model, scored by mean |skin temperature − 33.7 C| (a proxy, not a comfort index). One toggle is enough (more improve the score by at most 0.001 C). The best toggle is not always at the step: it is early or late after a step down. The gate is within 0.02 C of the optimum after a step up but 0.06–0.11 C away after a step down (0.18 with 10 min lag in one case). One person, two assumed transmittances, a coarse grid; not a claim about any real garment.

A figure from its own output

Paper, code and results

• README and method
• Experiments and results
• Code
• Tests

More in Engineered clothing

• What does a degree of comfort cost in fabric?Joins garment-thermal comfort shifts to garment-teardown costs: only one fabric treatment helps in each scenario, levers combine sub-additively, and at assumed premiums a degree costs $0.5 to $19 on a replica trench but $3 to $114 on a designer one. Stylised; premiums assumed.
• Does a fabric lever still pay once you weight it by where the garment is worn?Weights garment-comfortcost's fabric levers by wearing time and adds a gated infrared lever: a fixed infrared-transparent shell pays only with a warm-room majority (above 75.5% of time against the cool room), white never harms, and the gate is chosen only when it costs about the fixed shell's premium or less. Stylised; premiums and wearing shares assumed.
• How many home-sewn coats before the labour cost per coat settles?How many home-sewn replica coats until labour cost settles? On a Wright learning curve (assumed first coat 4× an experienced operator, rate 0.85) it takes 20 coats to get within 2× of factory time; the first coat stays under $50 only if your time is valued at ≤$1.45/h. Learning parameters assumed. Stylised.
• How much longer must a $2,850 coat last to beat a $56 replica on cost per wear?How much longer must a $2,850 coat last than a $56 replica to win on cost per wear? About 46× with no resale and equal care; needs a median life near 3,100 wears to win half the time. Lifetimes, care and resale assumed. Stylised.
• Is a second garment better than a dearer single one?Is a second garment better than a dearer single one? A shell-and-layer pair beats the best single garment at the same budget in 54 of 66 wearing mixes at $50 and $100 (mean +0.49 C and +1.02 C) once thin cloth is removed; the apparent $20 gain is an artefact of thin cloth's negative assumed cost. Stylised; premiums and wearing shares assumed.
• How often can a wearer pick the wrong garment before a second garment stops paying?How often can a wearer pick the wrong garment before a second garment stops paying? With random wrong picks the pair's edge over one garment holds (+0.57 C at 30% error, $100); if every mistake is the worst option it falls from +0.55 to +0.09 C (stylised).

All 45 Engineered clothing projects →

Other research topics

Proper scoring & elicitation28Effort & peer prediction9Verification games & slashing19Detection & tolerances14Markets as routers13Prediction-market design9Paying contributors8Decentralised training systems23Digital twins & sim-to-real164LLM chip design loops40Robot manufacturing loops53Ambient AI assistant46Self-replicating manufacturing34Simulation-first design64Engineered clothing45Task-specific AI accelerators40Algorithmic chip-design loops42Open-source purpose-built robots44The software-engineering loop43Adversarial human imitation41Open-source residential construction41Autonomous multi-angle capture34Sensing, imaging and 3D31Energy supply and dense energy storage39Robot societies among people31Agentic co-design of low-cost robots27Laundry-hanging robots25Machine learning on qualitative data23Robots on the building site27Civilizations & group psychology1History & human sciences1

Produced with AI assistance (Claude). Results are from the code in this repository and are simulated or stylised unless the abstract says otherwise; nothing here implies affiliation with or endorsement by any lab, researcher or company cited. MIT licensed.