hrrsmjd/kuramoto_sivashinsky
收藏Hugging Face2024-07-18 更新2024-07-22 收录
下载链接:
https://hf-mirror.com/datasets/hrrsmjd/kuramoto_sivashinsky
下载链接
链接失效反馈官方服务:
资源简介:
该数据集包含一维Kuramoto Sivashinsky方程的数据,通过方法线生成,空间导数使用伪谱方法计算。训练数据的T值为100,验证和测试数据的T值为200,时间步长∆t为0.2。空间范围X为[0, 64],空间步长∆x为0.25。初始条件是从具有随机系数的截断傅里叶级数的分布中采样的。数据集包含2048条训练轨迹,每条验证和测试数据集包含128条轨迹。每条轨迹的前360步被视为预热阶段并被丢弃。数据以双精度浮点格式(float64)生成,建议在大多数使用情况下将数据转换为单精度浮点格式(float32)。
This dataset contains data for the one-dimensional Kuramoto Sivashinsky equation, generated using the method of lines with spatial derivatives computed using the pseudo-spectral method. The training data has T = 100, while validation and testing data have T = 200, with a time step ∆t = 0.2. The spatial domain X is [0, 64] with a spatial step ∆x = 0.25. Initial conditions were sampled from a distribution over the truncated Fourier series with random coefficients. The dataset includes 2048 trajectories for training and 128 trajectories each for validation and testing. The first 360 steps of each trajectory were considered part of the warmup phase and subsequently discarded. The data was generated in double-precision floating-point format (float64), and it is recommended to convert the data to single-precision floating-point format (float32) for most use cases.
提供机构:
hrrsmjd
原始信息汇总
Kuramoto Sivashinsky 方程数据集
数据生成
- 方法:使用线法生成,空间导数通过伪谱法计算。
- 时间设置:
- 训练数据:T = 100,∆t = 0.2
- 验证和测试数据:T = 200,∆t = 0.2
- 空间设置:X = [0, 64],∆x = 0.25
- 初始条件:从截断傅里叶级数分布中采样随机系数。
- 边界条件:周期性边界条件。
数据集结构
- 训练数据:包含2048条轨迹。
- 验证数据:包含128条轨迹。
- 测试数据:包含128条轨迹。
- 数据格式:生成时使用双精度浮点格式(float64),推荐转换为单精度浮点格式(float32)。
数据处理
- 预热阶段:每条轨迹的前360步被视为预热阶段并被丢弃。



