Hello,
I am using PySINDy (specifically non-ensemble SINDy / WSINDy) for PDE discovery on high-dimensional spatiotemporal data. In my case, the spatial grid is very large, resulting in a huge number of samples. This leads to extremely large library matrices, slow computation, and occasional memory overflow (especially with WSINDy).
I would like to know whether it is valid in PySINDy to apply random subsampling over the spatiotemporal domain—i.e., randomly selecting a subset of rows from the candidate library matrix (Θ), similar to the approach in Rudy et al. (2017), for non-ensemble SINDy / WSINDy.
Thank you for your guidance.
Hello,
I am using PySINDy (specifically non-ensemble SINDy / WSINDy) for PDE discovery on high-dimensional spatiotemporal data. In my case, the spatial grid is very large, resulting in a huge number of samples. This leads to extremely large library matrices, slow computation, and occasional memory overflow (especially with WSINDy).
I would like to know whether it is valid in PySINDy to apply random subsampling over the spatiotemporal domain—i.e., randomly selecting a subset of rows from the candidate library matrix (Θ), similar to the approach in Rudy et al. (2017), for non-ensemble SINDy / WSINDy.
Thank you for your guidance.