In this document, we list common questions that users may ask about simulating their applications.
This is likely due to the fact that your input data is large. Simulating an application on large data usually takes a while. You can either leave the process running and wait for the result, or you can try simulating your application on smaller data.
If you are building a graph-based application, you can try to simulate the application with a smaller graph. For example, NetworkX provides functions to generate random social graph at arbitrary size:
import networkx as nx
# Create a caveman graph with 1 clique and 500 nodes in the clique
g = nx.caveman_graph(1, 500)If you are building an neural network application, you can try to simulate the application with a smaller feature size or hidden size.
Then, you can linearly scale the time required for a larger dataset based on the numbers obtained for the smaller ones. The power number should be roughly the same across different data sizes.
This is likely due to the fact that your kernel issued an OOM killer. You can either run your simulation with a more powerful machine, or you should try to clean up your system memory before doing the simulation.