We have identified a performance bottleneck in the initialization phase of our solver algorithm, particularly when dealing with large matrices. Our current process involves identifying and fixing problematic voltage nodes and current loops one at a time when a matrix cannot be solved due to ill-conditioning. This approach, while effective, results in significant delays in matrices with a large number of nodes (e.g., more than 1000 nodes).
We have modified the algorithm to allow fixing multiple problematic nodes simultaneously during the initialization phase. By doing so, we significantly reduce the time taken for initialization in large matrices.
We have identified a performance bottleneck in the initialization phase of our solver algorithm, particularly when dealing with large matrices. Our current process involves identifying and fixing problematic voltage nodes and current loops one at a time when a matrix cannot be solved due to ill-conditioning. This approach, while effective, results in significant delays in matrices with a large number of nodes (e.g., more than 1000 nodes).
We have modified the algorithm to allow fixing multiple problematic nodes simultaneously during the initialization phase. By doing so, we significantly reduce the time taken for initialization in large matrices.