# 运行时错误：cusolver错误：计算奇异矩阵的SVD时出现CUSOLVER_STATUS_INTERNAL_ERROR

- **ID:** `cuda/cusolver-internal-error-on-svd`
- **领域:** cuda
- **类别:** runtime_error
- **错误码:** `CUSOLVER_STATUS_INTERNAL_ERROR`
- **验证级别:** ai_generated
- **修复率:** 76%

## 根因

当输入矩阵恰好是奇异矩阵或包含NaN/inf值时，cuSolver的SVD例程（gesvdj或gesvd）内部失败，导致迭代求解器中的缓冲区溢出或除零错误。

## 版本兼容性

| 版本 | 状态 | 引入 | 弃用 |
|------|------|------|------|
| CUDA 12.4 | active | — | — |
| cuSolver 11.5.1 | active | — | — |
| PyTorch 2.3.0 | active | — | — |

## 解决方案

1. ```
   Preprocess the matrix to remove exact singularities: add a small regularization term (e.g., A += 1e-8 * torch.eye(n, device=A.device)) before calling torch.linalg.svd. Example: A_reg = A + 1e-8 * torch.eye(A.size(0), device=A.device); U, S, V = torch.linalg.svd(A_reg).
   ```
2. ```
   Use torch.linalg.lstsq instead of SVD for solving least-squares problems, as it handles singular matrices more robustly.
   ```

## 无效尝试

- **** — This works but defeats the purpose of GPU acceleration; also, the error may still occur on CPU if the matrix is singular. (60% 失败率)
- **** — Singular matrices remain singular regardless of precision; the error is algorithmic, not numerical. (85% 失败率)