スキル一覧に戻る

formulate-quantum-problem

pjt222
更新日 Yesterday
17
2
17
GitHubで表示
その他general

について

このスキルは、ヒルベルト空間、演算子、境界条件を含む数学的枠組みを定義することで、開発者が量子力学や化学の問題を定式化するのを支援します。物理的なシナリオをシュレーディンガー方程式などの形式体系に変換する手引きを行い、摂動理論やDFT(密度汎関数理論)などの適切な解法を選択します。解析的または数値的な解法のために量子問題を設定する際にご利用ください。

クイックインストール

Claude Code

推奨
メイン
npx skills add pjt222/agent-almanac -a claude-code
プラグインコマンド代替
/plugin add https://github.com/pjt222/agent-almanac
Git クローン代替
git clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/formulate-quantum-problem

このコマンドをClaude Codeにコピー&ペーストしてスキルをインストールします

ドキュメント

Formulate Quantum Problem

Turn physical system into well-posed quantum problem. Find relevant degrees of freedom. Build Hamiltonian + state space. Set boundary conditions. Pick approximation method. Validate formulation vs known limits.

When Use

  • Set up quantum mechanics problem for analytic or numerical solution
  • Formulate quantum chemistry calculation (molecular orbitals, electronic structure)
  • Translate physical scenario into Dirac or Schrodinger formalism
  • Pick between perturbation theory, variational, DFT, exact diagonalization
  • Prep theoretical model for comparison with experimental spectroscopic or scattering data

Inputs

  • Required: Physical system description (atom, molecule, solid, field)
  • Required: Observables (energy spectrum, transition rates, ground state)
  • Optional: Experimental constraints or data (spectral lines, binding energies)
  • Optional: Desired accuracy or computational budget
  • Optional: Preferred formalism (wave mechanics, matrix mechanics, second quantization, path integral)

Steps

Step 1: Find Physical System + Relevant Degrees of Freedom

Characterize system before writing equations:

  1. Particle content: List particles (electrons, nuclei, photons, phonons) + quantum numbers (spin, charge, mass).
  2. Symmetries: Find spatial (spherical, cylindrical, translational, crystal group), internal (spin rotation, gauge), discrete (parity, time reversal).
  3. Energy scales: Find relevant energy scales. Decide which degrees of freedom active, which frozen or adiabatic.
  4. Degrees of freedom shrink: Apply Born-Oppenheimer if nuclear + electronic timescales separate. Find collective coordinates if many-body simplify.
## System Characterization
- **Particles**: [list with quantum numbers]
- **Active degrees of freedom**: [coordinates, spins, fields]
- **Frozen degrees of freedom**: [and justification for freezing]
- **Symmetry group**: [continuous and discrete]
- **Energy scale hierarchy**: [e.g., electronic >> vibrational >> rotational]

Got: Complete inventory — particles, quantum numbers, symmetries, justified active vs frozen degrees of freedom.

If fail: Energy scale hierarchy unclear? Keep all degrees of freedom, flag need for scale analysis. Premature truncation → qualitatively wrong physics.

Step 2: Build Hamiltonian + State Space

Build math framework from degrees of freedom in Step 1:

  1. Hilbert space: Define state space. Finite-dim → specify basis (spin-1/2 |up>, |down>). Infinite-dim → specify function space (L2(R^3) for single particle in 3D).
  2. Kinetic terms: Kinetic operator each particle. Position: T = -hbar^2/(2m) nabla^2.
  3. Potential terms: All interaction potentials (Coulomb, harmonic, spin-orbit, external). Explicit functional form + coupling constants.
  4. Composite Hamiltonian: Assemble H = T + V, group by interaction type. Multi-particle → include exchange + correlation or note approximation entry.
  5. Operator algebra: Verify Hamiltonian Hermitian. Find constants of motion ([H, O] = 0) for block-diagonalization.
## Hamiltonian Structure
- **Hilbert space**: [definition and basis]
- **H = T + V decomposition**:
  - T = [kinetic terms]
  - V = [potential terms, grouped by type]
- **Constants of motion**: [operators commuting with H]
- **Symmetry-adapted basis**: [if block diagonalization is possible]

Got: Complete Hermitian Hamiltonian, all terms explicit. Hilbert space defined. Constants of motion identified.

If fail: Not manifestly Hermitian? Check missing conjugate terms or gauge phases. Hilbert space ambiguous (relativistic)? Specify formalism explicit, note issue.

Step 3: Set Boundary + Initial Conditions

Constrain problem for unique solution:

  1. Boundary conditions: Bound state → normalizability (psi -> 0 at infinity). Scattering → incoming wave boundary. Periodic → Bloch or Born-von Karman.
  2. Domain restrictions: Spatial domain. Particle in box → walls. Hydrogen atom → radial + angular. Lattice models → lattice + topology.
  3. Initial state (time-dependent): State at t=0 as expansion in energy eigenbasis or wave packet with center + width.
  4. Constraint equations: Indistinguishable particles → symmetrize (bosons) or antisymmetrize (fermions). Gauge theories → gauge-fixing.
## Boundary and Initial Conditions
- **Spatial domain**: [definition]
- **Boundary type**: [Dirichlet / Neumann / periodic / scattering]
- **Normalization**: [condition]
- **Particle statistics**: [bosonic / fermionic / distinguishable]
- **Initial state** (if time-dependent): [specification]

Got: Boundary conditions physically motivated, mathematically consistent with Hamiltonian domain, sufficient for unique solution (or well-defined scattering matrix).

If fail: Over- or under-determined? Check self-adjointness of Hamiltonian on chosen domain. Non-self-adjoint → careful treatment of deficiency indices.

Step 4: Pick Approximation Method

Pick solution strategy for problem structure:

  1. Check exact solvability: Problem reduces to known exactly solvable model (harmonic oscillator, hydrogen atom, Ising)? Yes → use exact + perturbation for corrections.

  2. Perturbation theory (weak coupling):

    • Split H = H0 + lambda V, H0 exactly solvable
    • Verify lambda V small vs level spacing of H0
    • Check degeneracy; degenerate perturbation theory if needed
    • Good for: weak interaction, few-body, analytic results
  3. Variational methods (ground state):

    • Trial wave function with adjustable parameters
    • Trial function satisfies boundary + symmetry
    • Good for: ground state energy target, many-body
  4. Density Functional Theory (many-electron):

    • Exchange-correlation functional (LDA, GGA, hybrid)
    • Basis set (plane waves, Gaussian, numerical atomic orbitals)
    • Good for: many-electron, ground state density + energy
  5. Numerical exact methods (small, benchmarking):

    • Exact diagonalization for small Hilbert spaces
    • Quantum Monte Carlo for ground state sampling
    • DMRG for 1D or quasi-1D
    • Good for: high accuracy, small system
## Approximation Method Selection
- **Method chosen**: [name]
- **Justification**: [why this method fits the problem structure]
- **Expected accuracy**: [order of perturbation, variational bound quality, DFT functional accuracy]
- **Computational cost**: [scaling with system size]
- **Alternatives considered**: [and why they were rejected]

Got: Justified choice with clear accuracy + cost. Alternatives documented.

If fail: No single method clearly right? Formulate for two methods + compare. Disagreement reveals difficulty + guides refinement.

Step 5: Validate Formulation vs Known Limits

Before solving, verify formulation reproduces known physics:

  1. Classical limit: Take hbar -> 0 (or large quantum numbers), verify Hamiltonian reduces to correct classical mechanics.
  2. Non-interacting limit: Set couplings to zero, verify solution = product of single-particle states.
  3. Symmetry limits: Verify formulation respects all identified symmetries. Check Hamiltonian transforms correctly under symmetry group.
  4. Dimensional analysis: Verify every term has units of energy. Check characteristic length, energy, time scales physically reasonable.
  5. Known exact results: Special cases (hydrogen atom Z=1, harmonic oscillator quadratic potential)? Verify formulation reproduces them.
## Validation Checks
| Check | Expected Result | Status |
|-------|----------------|--------|
| Classical limit (hbar -> 0) | [classical Hamiltonian] | [Pass/Fail] |
| Non-interacting limit | [product states] | [Pass/Fail] |
| Symmetry transformation | [correct representation] | [Pass/Fail] |
| Dimensional analysis | [all terms in energy units] | [Pass/Fail] |
| Known exact case | [reproduced result] | [Pass/Fail] |

Got: All validation checks pass. Formulation self-consistent, ready to solve.

If fail: Failed check = error in Hamiltonian construction or boundary. Trace to term or condition, fix before solving.

Checks

  • All particles + quantum numbers explicit
  • Hilbert space defined with clear basis
  • Hamiltonian Hermitian + all terms correct units
  • Constants of motion identified + used for simplification
  • Boundary conditions physically motivated + mathematically sufficient
  • Particle statistics (bosonic/fermionic) correctly enforced
  • Approximation method choice justified + accuracy stated
  • Classical, non-interacting, symmetry limits checked
  • Known exact results reproduced special cases
  • Formulation complete for implementation

Pitfalls

  • Dropping degrees of freedom early: Freezing without energy scale check misses physics. Always justify with scale argument.
  • Non-Hermitian Hamiltonian: Forgetting conjugate terms in spin-orbit or complex potentials. Verify H = H-dagger explicit.
  • Wrong boundary for scattering: Bound-state boundary (normalizability) for scattering discards continuous spectrum. Match boundary to physical question.
  • Ignoring degeneracy in perturbation theory: Non-degenerate on degenerate level → divergent corrections. Check degeneracy before expanding.
  • Over-rely on single approximation: Different methods = complementary failure modes. Variational → upper bounds but miss excited states. Perturbation diverges at strong coupling. Cross-validate when possible.
  • Dimensional inconsistency: Mixing natural units (hbar = 1) with SI in same expression. Pick unit system at start, state it explicit.

See Also

  • derive-theoretical-result -- derive analytic results from formulated problem
  • survey-theoretical-literature -- prior work on similar quantum systems

GitHub リポジトリ

pjt222/agent-almanac
パス: i18n/caveman/skills/formulate-quantum-problem
0
agentsagentskillsai-assisted-developmentclaude-codeskillsteams

関連スキル

llamaguard

その他

LlamaGuardは、暴力やヘイトスピーチなど6つの安全性カテゴリーにおいて、LLMの入力と出力をモデレートするMetaの70-80億パラメータモデルです。94〜95%の精度を提供し、vLLM、Hugging Face、Amazon SageMakerを使用してデプロイ可能です。このスキルを使用して、AIアプリケーションにコンテンツフィルタリングと安全策を簡単に統合できます。

スキルを見る

cost-optimization

その他

このClaudeスキルは、リソースの適正サイジング、タグ付け戦略、支出分析を通じて、開発者がクラウドコストを最適化することを支援します。AWS、Azure、GCPにわたるクラウド支出の削減とコストガバナンスの実施のためのフレームワークを提供します。インフラコストの分析、リソースの適正サイジング、または予算制約への対応が必要な際にご利用ください。

スキルを見る

quantizing-models-bitsandbytes

その他

このスキルは、bitsandbytesを使用してLLMを8ビットまたは4ビット精度に量子化し、精度の低下を最小限に抑えつつ50〜75%のメモリ削減を実現します。限られたGPUメモリでより大規模なモデルを実行したり、推論を高速化するのに理想的で、INT8、NF4、FP4などのフォーマットをサポートしています。HuggingFace Transformersと統合され、QLoRAトレーニングや8ビットオプティマイザーを可能にします。

スキルを見る

dispatching-parallel-agents

その他

このClaudeスキルは、複数のエージェントを配備し、3つ以上の独立した問題を並行して調査・修正します。共有状態や依存関係がなく解決可能な、無関係な障害が発生するシナリオ向けに設計されています。中核となる機能は並列問題解決であり、効率を最大化するために独立した問題領域ごとに1つのエージェントを割り当てます。

スキルを見る