How do Compute Nodes Work

Compute Node is one of the core components of the Lumoz Protocol, responsible for computing ZKPs and AI results within the protocol. By utilizing task splitting, parallel computation, and proof aggregation, we effectively enhance the efficiency of ZKP/AI computation and reduce its overall cost. Below is an overview of the workflow of Compute nodes:

1. Listening for Tasks: Compute Nodes will continuously monitor the chain for the publication of corresponding tasks based on the type of proof and initiate calculations accordingly.

2. Proof Calculation: A single task will be recursively broken down into multiple sub-tasks, allowing for parallel computation. At the same time, nodes will check the local cache for any precomputed or cached results that can be reused to optimize the process.

3. Sub-task Aggregation: Once all sub-tasks have been computed, the results will be aggregated into a single, complete proof.

4. Aggregation and Submission of Multiple Proofs: Instead of submitting a single proof immediately after generation, nodes will wait to aggregate it with subsequent proofs. The combined proofs and node information will be consolidated into a compact submission, which will then be posted on-chain to minimize cost overhead.

5. Waiting for Subsequent Verification Steps: Following submission, nodes will proceed to the next stage of the verification process.