Name: thomson-problem-experiment
Author: KonkovDV

Prompt file imported from KonkovDV/OMEGA-MATH (.github/prompts/thomson-problem-experiment.prompt.md). Copyright stays with the author.

description: Run or analyze the executable Thomson-problem optimization lane from the OMEGA math workspace. agent: omega-math argument-hint: "phase1 N=2..20 restarts=20" or "phase2 N=7..20 niter=30 restarts=4" model: GPT-5.4 (copilot)

Work inside the standalone math repository on thomson-problem.

Required sequence:

Read problem brief, literature packet, and attack plan.
Treat the currently shipped experiment lanes as:
- phase1_multistart.py
- phase2_basin_hopping.py
Use explicit seeds and bounded restart / basin-hopping counts first.
Compare every energy value against the Cambridge known energies embedded in the scripts and brief.
Do not claim a new minimum unless the energy is strictly lower than the known value by a credible margin, the best configuration is persisted, and the run is reproducible.

Return:

parameters used
energies found and deltas vs known values
whether the result is a reproduction, a gap, or a plausible improvement
the next best bounded run

--- description: Run or analyze the executable Thomson-problem optimization lane from the OMEGA math workspace. agent: omega-math argument-hint: "phase1 N=2..20 restarts=20" or "phase2 N=7..20 niter=30 restarts=4" model: GPT-5.4 (copilot) --- Work inside the standalone `math` repository on `thomson-problem`. Required sequence: 1. Read [problem brief](../../research/active/thomson-problem/input_files/data_description.md), [literature packet](../../research/active/thomson-problem/input_files/literature.md), and [attack plan](../../research/active/thomson-problem/planning/attack_plan.md). 2. Treat the currently shipped experiment lanes as: - `phase1_multistart.py` - `phase2_basin_hopping.py` 3. Use explicit seeds and bounded restart / basin-hopping counts first. 4. Compare every energy value against the Cambridge known energies embedded in the scripts and brief. 5. Do not claim a new minimum unless the energy is strictly lower than the known value by a credible margin, the best configuration is persisted, and the run is reproducible. Return: - parameters used - energies found and deltas vs known values - whether the result is a reproduction, a gap, or a plausible improvement - the next best bounded run

{ "prompt": "---\ndescription: Run or analyze the executable Thomson-problem optimization lane from the OMEGA math workspace.\nagent: omega-math\nargument-hint: \"phase1 N=2..20 restarts=20\" or \"phase2 N=7..20 niter=30 restarts=4\"\nmodel: GPT-5.4 (copilot)\n---\n\nWork inside the standalone `math` repository on `thomson-problem`.\n\nRequired sequence:\n\n1. Read [problem brief](../../research/active/thomson-problem/input_files/data_description.md), [literature packet](../../research/active/thomson-problem/input_files/literature.md), and [attack plan](../../research/active/thomson-problem/planning/attack_plan.md).\n2. Treat the currently shipped experiment lanes as:\n\t- `phase1_multistart.py`\n\t- `phase2_basin_hopping.py`\n3. Use explicit seeds and bounded restart / basin-hopping counts first.\n4. Compare every energy value against the Cambridge known energies embedded in the scripts and brief.\n5. Do not claim a new minimum unless the energy is strictly lower than the known value by a credible margin, the best configuration is persisted, and the run is reproducible.\n\nReturn:\n\n- parameters used\n- energies found and deltas vs known values\n- whether the result is a reproduction, a gap, or a plausible improvement\n- the next best bounded run", "variables": [], "notes": "Work inside the standalone `math` repository on `thomson-problem`.", "metadata": { "source": { "provider": "github-prompts", "repository": "https://github.com/KonkovDV/OMEGA-MATH", "path": ".github/prompts/thomson-problem-experiment.prompt.md", "ref": "1110663349b1582722a3cbbc6bf8668dfef8caa7", "url": "https://github.com/KonkovDV/OMEGA-MATH/blob/1110663349b1582722a3cbbc6bf8668dfef8caa7/.github/prompts/thomson-problem-experiment.prompt.md", "key": "KonkovDV/OMEGA-MATH/.github/prompts/thomson-problem-experiment.prompt.md" } } }

thomson-problem-experiment

About

description: Run or analyze the executable Thomson-problem optimization lane from the OMEGA math workspace. agent: omega-math argument-hint: "phase1 N=2..20 restarts=20" or "phase2 N=7..20 niter=30 restarts=4" model: GPT-5.4 (copilot)

Use it

Manifest

Reviews