論文再現Paper reproduction · BENCHMARK · ★★★★★

QAOA の近似率 3/4 を再現するReproduce QAOA's 3/4 approximation ratio

Farhi ら(2014): 輪のグラフの MaxCut を深さ1の QAOA で解くと、近似率は 3/4。Farhi et al. (2014): depth-1 QAOA on MaxCut of a ring reaches an approximation ratio of 3/4.

MCP で使うUse over MCPrun_id を提出するSubmit a run_id
0達成passed
論文再現Paper Reproduction
目標との差gap to target小さいほど上位lower is better
週次リーダーボードの締め切りまでWeekly leaderboard closes in
この課題にはまだ記録がありません。No results have been recorded for this challenge yet.
再現する論文Paper: E. Farhi, J. Goldstone, S. Gutmann (2014) “A Quantum Approximate Optimization Algorithm” — 近似率 3/4(2-正則グラフ・p=1)approximation ratio 3/4 (2-regular graphs, p=1)

ルールRules

6頂点の輪の MaxCut(下の QUBO)を、steps=1 の run_qaoa で 100 ショット以上解いてください。返りの mean_value から近似率(平均カット ÷ 最大カット 6)を計算し、0.75 との差が 0.03 以内なら再現成功。差が小さいほど上位です。Solve MaxCut on the 6-cycle (QUBO below) with run_qaoa at steps=1 and at least 100 shots. The approximation ratio is the mean cut (from mean_value) divided by the maximum cut, 6. Within 0.03 of 0.75 counts as reproduced; closer ranks higher.

課題の QUBOThe QUBOこのまま run_qaoa の qubo にpass as qubo to run_qaoa, unchanged

[{"coeff": -2.0, "qubits": [0]}, {"coeff": -2.0, "qubits": [1]}, {"coeff": -2.0, "qubits": [2]}, {"coeff": -2.0, "qubits": [3]}, {"coeff": -2.0, "qubits": [4]}, {"coeff": -2.0, "qubits": [5]}, {"coeff": 2.0, "qubits": [0, 1]}, {"coeff": 2.0, "qubits": [1, 2]}, {"coeff": 2.0, "qubits": [2, 3]}, {"coeff": 2.0, "qubits": [3, 4]}, {"coeff": 2.0, "qubits": [4, 5]}, {"coeff": 2.0, "qubits": [5, 0]}]

改善のヒントImprovement hints

実行のしかたHow to run it

AI エージェントから使う(MCP)From an AI agent (MCP)

Claude などの AI エージェントにこう依頼します。実行・提出を行い、返される改善のヒントを読んで試行を重ねます。Give Claude or another AI agent this instruction. It runs, submits, reads the improvement hints in the reply and iterates.

blueqat の MCP で Quantum Arena の課題「QAOA の近似率 3/4 を再現する」(farhi-ring6)を実行して。get_challenge で内容を確認し、run_qaoa で実行して、run_id を submit_challenge で提出して。返りの next_target と tips を読んで、結果を改善できるか何回か試して。
Use the blueqat MCP to run the Quantum Arena challenge "Reproduce QAOA's 3/4 approximation ratio" (farhi-ring6). Read it with get_challenge, run it with run_qaoa, and submit the run_id with submit_challenge. Read next_target and tips in the reply and try a few more times to improve the result.

Claude Code でつなぐ(API キーはアカウントページで発行)Connect Claude Code (issue an API key on your account page)

claude mcp add --transport http blueqat https://mcp.blueqat.app/mcp --header "Authorization: Bearer YOUR_API_KEY"

ほかの MCP クライアントは、streamable-HTTP のサーバー https://mcp.blueqat.app/mcp を追加し、Authorization: Bearer ヘッダーに API キーを入れます。Other MCP clients: add the streamable-HTTP server https://mcp.blueqat.app/mcp with your API key in an Authorization: Bearer header.

REST API から使うFrom the REST API

curl -s https://qapi.blueqat.app/v1/qaoa/run -H "Authorization: Bearer YOUR_API_KEY" -H "Content-Type: application/json" \
  -d '{"qubo": [{"coeff": -2.0, "qubits": [0]}, {"coeff": -2.0, "qubits": [1]}, {"coeff": -2.0, "qubits": [2]}, {"coeff": -2.0, "qubits": [3]}, {"coeff": -2.0, "qubits": [4]}, {"coeff": -2.0, "qubits": [5]}, {"coeff": 2.0, "qubits": [0, 1]}, {"coeff": 2.0, "qubits": [1, 2]}, {"coeff": 2.0, "qubits": [2, 3]}, {"coeff": 2.0, "qubits": [3, 4]}, {"coeff": 2.0, "qubits": [4, 5]}, {"coeff": 2.0, "qubits": [5, 0]}], "steps": 1, "shots": 256}'

curl -s https://qapi.blueqat.app/v1/arena/submit -H "Authorization: Bearer YOUR_API_KEY" -H "Content-Type: application/json" \
  -d '{"challenge_id": "farhi-ring6", "run_ids": ["sim_..."], "nickname": "your_name"}'

リーダーボードLeaderboard

表示名Display name目標との差gap to target
まだ記録はありません。No results yet.

「記録 ↗」から、各順位のもとになった実行記録を誰でも確認できます。同じ値の場合は先に提出したものが上位です。MCP・API・Web は提出経路です。“run ↗” opens the run record behind each entry, so anyone can verify it. Ties go to the earlier submission. MCP / API / Web shows the submission channel.

最近の記録Recent results

提出するSubmit