You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi, Ni Trieu! I found mistake in your unbalanced setting and the simple ideas might be useful!
In unbalanced setting test, assume sender vs receiver : 2^16 vs 5000. expectedIntersection=300
I change your code above , so the intersection not in sender's first 5000.
The result seems wrong.
Idea 1:
I set the sender degree = mMyInputSize - 1, so the sender build sub product tree and evaluate according to sender's all inputs .And the poly he received from receiver don't change(degree = receiver_size -1). It looks like use a low degree poly to modulus the high-degree sub product tree, and the reminder poly in leaf node is the final result.
Finally, I get the right result, but the cost seems expensive.
Idea 2:
Splite the sender set in multple batch ,each batch match with the size of the receiver.
Maybe I was wrong about that.
The text was updated successfully, but these errors were encountered:
Hi, Ni Trieu! I found mistake in your unbalanced setting and the simple ideas might be useful!
In unbalanced setting test, assume sender vs receiver : 2^16 vs 5000. expectedIntersection=300
I change your code above , so the intersection not in sender's first 5000.
The result seems wrong.
I set the sender degree = mMyInputSize - 1, so the sender build sub product tree and evaluate according to sender's all inputs .And the poly he received from receiver don't change(degree = receiver_size -1). It looks like use a low degree poly to modulus the high-degree sub product tree, and the reminder poly in leaf node is the final result.
![image](https://user-images.githubusercontent.com/29690159/133218663-268c5f13-e41d-4f71-ab27-0b92f5c33c4c.png)
![image](https://user-images.githubusercontent.com/29690159/133218763-6a3427e7-8646-4b9e-8c12-afe21b641baf.png)
![image](https://user-images.githubusercontent.com/29690159/133219194-122cfda2-2bf5-4381-b0a9-21cfe8b15948.png)
Finally, I get the right result, but the cost seems expensive.
Splite the sender set in multple batch ,each batch match with the size of the receiver.
Maybe I was wrong about that.
The text was updated successfully, but these errors were encountered: