Types of Computation Tasks

In the Arcium Network, two high-level categories of Computation exist: system Computations and normal Computations. System Computations are generated by processes related to the operation of the Arcium Network itself, while all other Computations are tasks requested by Computation Customers.

System Computations

System Computations manage essential processes within the Arcium Network itself and come in three types:

• DKGSystemComputation: These are distributed key generation (or DKG) operations that are run when a new MXE joins a new Cluster (see the Keyshare Distribution and Security section for more). Priority fees for these Computations are set by the Computation Customer when they request to add an MXE to a Cluster (see the Pricing section for more on priority fees).

• MigrationSystemComputation: These are Cluster migration Computations (either planned or forced). Priority fees in this case can be contributed towards by all of the Arx nodes in the Cluster that is being migrated — meaning that the portion of the priority fee paid by each node is not equal between them, each may contribute to the priority as much, or as little, as it would like (based on the importance of the execution speed to each Arx node).

• NonParticipationDetectionSystemComputation: These Computations occur when non-participation has been detected inside a Cluster and cannot be resolved by intra-Cluster consensus (a broader consensus of Arx nodes is needed, see the Slashing section for more). Priority fees in this case are directly inherited from the original Computations that led to the non-participation dispute — if the dispute results in a slash, then the cost of this duplicate priority fee is reimbursed to the nodes that executed the system computation, otherwise the duplicate priority fee represents a minor loss in revenue for the nodes involved.

Computations

Standard Computations in the Arcium Network are submitted by Computation Customers to execute confidential computing. A Computation is always defined within the context of an associated MXE, providing the state that the Computation operates on.

A Computation may optionally have an associated dependency, meaning that the Computation will not be executed until the specified dependent Computation is executed first.