总览 · ← 返回运行 20261002-034201-search-t1-abc-r1-C-native

节点 n8

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 2a1665e1a0: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：父节点版本 85e30575ef。改动的文件：solution/METHOD.md +5 −5、solution/run.py +89 −9

diff --git a/solution/METHOD.md b/solution/METHOD.mdindex 0f9313e..bbeff21 100644--- a/solution/METHOD.md+++ b/solution/METHOD.md@@ -1,8 +1,8 @@ ## 改了什么-1. 将基础采样限定为官方阶段（过滤 source=="external"），修复 proxy2 中从 Qiu E9.0 心脏细胞采样导致全胚预测崩溃的问题（proxy2 从 27→预期~50）。-2. 用按细胞类型分层抽样替代均匀随机抽样，保持类型比例从而改善 covariation 分数。-3. 对伪批量差值乘收缩系数 α=0.5（方法卡 k018 指出 α=1 低于 copy_last，需要收缩）。+1. 将 ALPHA 从 0.1 降至 0.05：实验表 Program 2（ALPHA=0.05）的 covariation=47.63、X3=47.41 均优于当前（ALPHA=0.1，covariation=47.04、X3=46.63）。更小的位移更接近 copy_last，更好地保留原始协方差结构。+2. 将 N_BINS 从 5 增加到 8：更细的距离分位数分箱能更精确地保持类型内部的距离分布，从而更好地保留基因间共变结构。+3. 将 quantile-bin 抽样的最小细胞数阈值从 50 降至 30：使更多中等大小的类型也能受益于保协方差抽样。 ## 用到的知识与出处 - 方法卡 k018（Damped per-type shift: shrinkage alpha）-- 任务书 CONTRACT 中关于 proxy2 外部输入阶段的说明（inputs_by_time 默认返回外部阶段、类型名不匹配）-- 方向库 T1-01（组成重加权）的思路用于分层采样保持比例+- 实验表 Program 2（ALPHA=0.05 + EB shrinkage → covariation 47.63, X3 47.41）+- 方向库 T1-12（保协方差的生成）diff --git a/solution/run.py b/solution/run.pyindex abfc55a..8906328 100644--- a/solution/run.py+++ b/solution/run.py@@ -12,7 +12,9 @@ import argparse  import numpy as np -from src.task1_temporal.baselines import shift_rows, type_deltas+from scipy import sparse as sp++from src.task1_temporal.baselines import shift_rows from src.task1_temporal.view_io import (     inputs_by_time,     labels_of,@@ -23,10 +25,57 @@ from src.task1_temporal.view_io import (     write_prediction, ) -ALPHA = 0.5+ALPHA = 0.05+N_BINS = 8+ +def _submatrix(X, mask):+    sub = X[mask]+    if sp.issparse(sub):+        return np.asarray(sub.todense())+    return np.asarray(sub) -def stratified_sample(labels: np.ndarray, n_target: int, rng: np.random.Generator) -> np.ndarray:++def eb_shrunk_deltas(prev_X, prev_labels, last_X, last_labels, alpha):+    prev_types = set(np.unique(prev_labels))+    last_types = np.unique(last_labels)+    deltas = {}+    for t in last_types:+        last_mask = last_labels == t+        n_last = int(last_mask.sum())+        if n_last == 0:+            continue+        last_sub = last_X[last_mask]+        if sp.issparse(last_sub):+            mean_last = np.asarray(last_sub.mean(axis=0)).ravel()+            var_last = np.asarray(last_sub.multiply(last_sub).mean(axis=0)).ravel() - mean_last**2+        else:+            arr = np.asarray(last_sub)+            mean_last = arr.mean(axis=0).ravel()+            var_last = arr.var(axis=0).ravel()+        var_last = np.maximum(var_last, 0.0)+        if t in prev_types:+            prev_mask = prev_labels == t+            n_prev = int(prev_mask.sum())+            prev_sub = prev_X[prev_mask]+            if sp.issparse(prev_sub):+                mean_prev = np.asarray(prev_sub.mean(axis=0)).ravel()+                var_prev = np.asarray(prev_sub.multiply(prev_sub).mean(axis=0)).ravel() - mean_prev**2+            else:+                arr_p = np.asarray(prev_sub)+                mean_prev = arr_p.mean(axis=0).ravel()+                var_prev = arr_p.var(axis=0).ravel()+            var_prev = np.maximum(var_prev, 0.0)+            delta = mean_last - mean_prev+            se2 = var_prev / max(n_prev, 1) + var_last / max(n_last, 1)+            shrink = delta**2 / (delta**2 + se2 + 1e-10)+            deltas[t] = alpha * shrink * delta+        else:+            deltas[t] = np.zeros(last_X.shape[1])+    return deltas+++def stratified_sample(labels: np.ndarray, X, n_target: int, rng: np.random.Generator) -> np.ndarray:     unique, counts = np.unique(labels, return_counts=True)     n_total = len(labels)     if n_target >= n_total:@@ -43,8 +92,40 @@ def stratified_sample(labels: np.ndarray, n_target: int, rng: np.random.Generato         if n_take <= 0:             continue         type_idx = np.where(labels == u)[0]-        chosen = rng.choice(type_idx, size=n_take, replace=False)-        indices.append(chosen)+        if n_take >= len(type_idx):+            indices.append(type_idx)+            continue+        if len(type_idx) > 30 and n_take >= 5:+            sub = _submatrix(X, type_idx)+            centroid = sub.mean(axis=0)+            dists = np.linalg.norm(sub - centroid, axis=1)+            bin_edges = np.quantile(dists, np.linspace(0, 1, N_BINS + 1)[1:-1])+            bin_ids = np.searchsorted(bin_edges, dists)+            bin_props = np.bincount(bin_ids, minlength=N_BINS) / len(type_idx)+            per_bin = np.floor(bin_props * n_take).astype(int)+            rem = n_take - per_bin.sum()+            fr = (bin_props * n_take) - per_bin+            for i in np.argsort(-fr)[:rem]:+                per_bin[i] += 1+            per_bin = np.minimum(per_bin, np.bincount(bin_ids, minlength=N_BINS))+            chosen = []+            for b in range(N_BINS):+                b_idx = type_idx[bin_ids == b]+                nb = min(per_bin[b], len(b_idx))+                if nb > 0:+                    chosen.append(rng.choice(b_idx, size=nb, replace=False))+            if chosen:+                sel = np.concatenate(chosen)+            else:+                sel = rng.choice(type_idx, size=n_take, replace=False)+            if len(sel) < n_take:+                leftover = np.setdiff1d(type_idx, sel)+                extra = rng.choice(leftover, size=n_take - len(sel), replace=False)+                sel = np.concatenate([sel, extra])+            indices.append(sel)+        else:+            chosen = rng.choice(type_idx, size=n_take, replace=False)+            indices.append(chosen)     result = np.concatenate(indices)     if len(result) < n_target:         remaining_idx = np.setdiff1d(np.arange(n_total), result)@@ -70,14 +151,13 @@ def main() -> None:     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)+    rows = stratified_sample(labels, last.X, 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)+        deltas = eb_shrunk_deltas(prev.X, labels_of(prev), last.X, labels, ALPHA)         del prev-        shrunk = {k: v * ALPHA for k, v in deltas.items()}-        X = shift_rows(X, labels[rows], shrunk)+        X = shift_rows(X, labels[rows], deltas)     write_prediction(X, genes, args.out, seed=args.seed)  

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 results, the weakest metrics are **covariation (47.04)** and **X3 (46.63)**. Program 2 showed that ALPHA=0.05 gave covariation=47.63 and X3=47.41, both better than current (ALPHA=0.1). I'll lower alpha and increase bin granularity for finer covariation preservation.

```METHOD
## 改了什么
1. 将 ALPHA 从 0.1 降至 0.05：实验表 Program 2（ALPHA=0.05）的 covariation=47.63、X3=47.41 均优于当前（ALPHA=0.1，covariation=47.04、X3=46.63）。更小的位移更接近 copy_last，更好地保留原始协方差结构。
2. 将 N_BINS 从 5 增加到 8：更细的距离分位数分箱能更精确地保持类型内部的距离分布，从而更好地保留基因间共变结构。
3. 将 quantile-bin 抽样的最小细胞数阈值从 50 降至 30：使更多中等大小的类型也能受益于保协方差抽样。
## 用到的知识与出处
- 方法卡 k018（Damped per-type shift: shrinkage alpha）
- 实验表 Program 2（ALPHA=0.05 + EB shrinkage → covariation 47.63, X3 47.41）
- 方向库 T1-12（保协方差的生成）
```

<<<<<<< SEARCH
ALPHA = 0.1
N_BINS = 5
=======
ALPHA = 0.05
N_BINS = 8
>>>>>>> REPLACE

<<<<<<< SEARCH
        if len(type_idx) > 50 and n_take >= 10:
=======
        if len(type_idx) > 30 and n_take >= 5:
>>>>>>> REPLACE