总览 · ← 返回运行 20261001-233757-search-t1-abc-r0-C-native

节点 n42

改了什么

方法说明?节点程序自带的 METHOD.md：这个程序做了什么、为什么。

来自 programs.git de8f3bc57a:solution/METHOD.md

调研员的计划

代码改动?这个节点的程序和父节点程序的逐行差别：绿色是新增，红色是删除。

对比：这个提交的上一版（种子程序：相对空仓库）。改动的文件：solution/METHOD.md +6 −0、solution/README.md +4 −0、solution/run.py +75 −0

diff --git a/solution/METHOD.md b/solution/METHOD.mdnew file mode 100644index 0000000..56edbc8--- /dev/null+++ b/solution/METHOD.md@@ -0,0 +1,6 @@+## 改了什么+1. 修复 proxy2 致命问题：原代码从 `stages[-1]`（proxy2 中是外部 Qiu E9.0，仅 2174 个心脏细胞）抽样作为输出，导致细胞类型覆盖极差（cell_state=32.86）。改为始终从最新**官方**阶段（`official[-1]`）抽样。+2. 用分层抽样（stratified_sample）替代 view_io 的 sample_rows：按细胞类型比例分配每个类型的抽样数，保持组成结构。这应改善 covariation 和 cell_state 分数。+3. delta 计算仅在 ≥2 个官方阶段时执行（final 视图），且直接从 `last`（=official[-1]）取 labels，不再重复读取。+## 用到的知识与出处+CONTRACT "外部输入阶段"节：proxy2 中 Qiu E9.0 与官方类型标签不同，不能直接当预测输出；方法卡 §平移幅度：pseudobulk_shift 在 T1 上 48.6，低于 copy_last，α 应收缩。diff --git a/solution/README.md b/solution/README.mdnew file mode 100644index 0000000..ba29577--- /dev/null+++ b/solution/README.md@@ -0,0 +1,4 @@+# pseudobulk_shift++最新阶段抽样后，每个细胞加上所属类型在最后一步的伪批量差值 mean(last|type) − mean(prev|type)，夹到 ≥0；前一阶段没有的类型原样复制。+T1 proxy 只有一个输入阶段，没有差值可取，退化成 copy_last（同样的抽样），所以 proxy 分 = copy_last（seed 0 实测 49.77）。final 才真正平移；官方在真实 T1 上报的常数位移是 48.6，低于地板。diff --git a/solution/run.py b/solution/run.pynew file mode 100644index 0000000..af3df61--- /dev/null+++ b/solution/run.py@@ -0,0 +1,75 @@+#!/usr/bin/env python3+"""pseudobulk_shift: latest stage + per-cell-type pseudobulk delta of the last step.++The delta is mean(last|type) - mean(prev|type) over the two latest inputs,+computed on the full stages and added once to a subsample of the latest stage+(clipped at 0). Types missing from the earlier stage are copied unchanged.++With a single input stage (T1 proxy: E8.5 only) there is no step to take a+delta from, so this falls back to copy_last with the same sampling. The proxy+therefore cannot tell this seed from copy_last; that gap is expected.+"""++from __future__ import annotations++import argparse++import numpy as np++from src.task1_temporal.baselines import shift_rows, type_deltas+from src.task1_temporal.view_io import (+    inputs_by_time,+    labels_of,+    load_manifest,+    panel_genes,+    read_stage,+    sample_rows,+    target_n_cells,+    write_prediction,+)+++def stratified_sample(labels: np.ndarray, n_target: int, rng: np.random.Generator) -> np.ndarray:+    unique, counts = np.unique(labels, return_counts=True)+    total = len(labels)+    indices = []+    for ct, count in zip(unique, counts):+        n_sample = max(1, int(round(count / total * n_target)))+        ct_idx = np.where(labels == ct)[0]+        chosen = rng.choice(ct_idx, size=min(n_sample, len(ct_idx)), replace=False)+        indices.append(chosen)+    indices = np.concatenate(indices)+    if len(indices) > n_target:+        indices = rng.choice(indices, size=n_target, replace=False)+    elif len(indices) < n_target:+        extra = rng.choice(np.arange(total), size=n_target - len(indices), replace=True)+        indices = np.concatenate([indices, extra])+    return indices+++def main() -> None:+    parser = argparse.ArgumentParser()+    parser.add_argument("--data", required=True)+    parser.add_argument("--out", required=True)+    parser.add_argument("--seed", type=int, default=0)+    args = parser.parse_args()++    manifest = load_manifest(args.data)+    genes = panel_genes(args.data, manifest)+    official = inputs_by_time(manifest, include_external=False)+    last = read_stage(args.data, official[-1], genes)+    labels = labels_of(last)+    rng = np.random.default_rng(args.seed)+    n_target = target_n_cells(manifest, last.n_obs)+    rows = stratified_sample(labels, n_target, rng)+    X = last.X[rows]+    if len(official) >= 2:+        prev = read_stage(args.data, official[-2], genes)+        deltas = type_deltas(prev.X, labels_of(prev), last.X, labels)+        del prev+        X = shift_rows(X, labels[rows], deltas)+    write_prediction(X, genes, args.out, seed=args.seed)+++if __name__ == "__main__":+    main()

调研来源?调研员查到并用到的知识条目和文献检索结果（只列标题和编号）。

没有记录调研来源。

分析结果?分析员写的 ANALYSIS.json：改了什么、各组分数怎么变、假设是否成立、经验和下一步建议。

对话摘要?每个角色和大模型对话的统计：轮数、工具调用、用时、token 数和最后的回答摘录；原始记录只给路径。

只给统计和最后回答的摘录；完整对话请到原始记录位置里列出的文件看。

分析员

ANALYSIS.json written

native

stream

Looking at the scores, the main weaknesses are **covariation (39.06)** and **X3 (40.19)**. The stratified sampling may be hurting covariation by distorting natural gene-gene correlations. I'll revert to `sample_rows`, add composition trend extrapolation (T1-03 direction) for multi-stage views, and use shrinkage on deltas.

<<<<<<< SEARCH
def stratified_sample(labels: np.ndarray, n_target: int, rng: np.random.Generator) -> np.ndarray:
    unique, counts = np.unique(labels, return_counts=True)
    total = len(labels)
    indices = []
    for ct, count in zip(unique, counts):
        n_sample = max(1, int(round(count / total * n_target)))
        ct_idx = np.where(labels == ct)[0]
        chosen = rng.choice(ct_idx, size=min(n_sample, len(ct_idx)), replace=False)
        indices.append(chosen)
    indices = np.concatenate(indices)
    if len(indices) > n_target:
        indices = rng.choice(indices, size=n_target, replace=False)
    elif len(indices) < n_target:
        extra = rng.choice(np.arange(total), size=n_target - len(indices), replace=True)
        indices = np.concatenate([indices, extra])
    return indices


def main() -> None:
    parser = argparse.ArgumentParser()
    parser.add_argument("--data", required=True)
    parser.add_argument("--out", required=True)
    parser.add_argument("--seed", type=int, default=0)
    args = parser.parse_args()

    manifest = load_manifest(args.data)
    genes = panel_genes(args.data, manifest)
    official = inputs_by_time(manifest, include_external=False)
    last = read_stage(args.data, official[-1], genes)
    labels = labels_of(last)
    rng = np.random.default_rng(args.seed)
    n_target = target_n_cells(manifest, last.n_obs)
    rows = stratified_sample(labels, n_target, rng)
    X = last.X[rows]
    if len(official) >= 2:
        prev = read_stage(args.data, official[-2], genes)
        deltas = type_deltas(prev.X, labels_of(prev), last.X, labels)