Inconsistent Verification Results with Specific Zero-Knowledge Proofs

[nihat99]

When submitting certain zero-knowledge proofs for verification, the system intermittently returns incorrect verification statuses. This issue appears to be sporadic, with the same proof sometimes passing and other times failing without any changes to the input data.

Steps to Reproduce:

Deploy the aligned_layer verification system in a test environment.
Submit a specific zero-knowledge proof (details of the proof can be provided upon request).
Observe the verification result.
Repeat the submission multiple times.
Expected Behavior:

The verification result should be consistent across multiple submissions of the same proof, either always passing or always failing, depending on the validity of the proof.

Actual Behavior:

The verification result varies between submissions of the same proof, with some attempts passing and others failing.

Environment:

aligned_layer version: [insert version]
Operating System: [insert OS]
Deployment method: [e.g., Docker, local installation]
Additional Information:

This issue does not occur with all proofs; it seems to be specific to certain proofs with particular characteristics.
No error messages or logs indicate a failure; the verification result simply toggles between pass and fail.
Network conditions and system load do not appear to influence the occurrence of this issue.

[MarcosNicolau]

Hi @nihat99. Could you be more specific with your issue: What is your proving system and version? Can you share your program code? How are you sending the proofs (via cli, sdk)? Please share those details so that we can reproduce it.