The DynamicProof
class enables circuits to verify proofs using in-circuit verification keys.
This is opposed to the baked-in verification keys of the Proof
class.
In order to use this, a subclass of DynamicProof that specifies the public input and output types along with the maxProofsVerified number has to be created.
export class SideloadedProgramProof extends DynamicProof<MyStruct, Field> {
static publicInputType = MyStruct;
static publicOutputType = Field;
static maxProofsVerified = 0 as const;
}
The maxProofsVerified
constant is a product of the child circuit and indicates the maximum number that that circuit verifies itself.
If you are unsure about what that is for you, you should use 2
.
Any DynamicProof
subclass can be used as private input to ZkPrograms or SmartContracts along with a VerificationKey
input.
proof.verify(verificationKey)
NOTE: In the case of DynamicProof
s, the circuit makes no assertions about the verificationKey used on its own.
This is the responsibility of the application developer and should always implement appropriate checks.
This pattern differs a lot from the usage of normal Proof
, where the verification key is baked into the compiled circuit.
src/examples/zkprogram/dynamic-keys-merkletree.ts for an example of how this can be done using merkle trees
Assertions generally only happen using the vk hash that is part of the VerificationKey
struct along with the raw vk data as auxiliary data.
When using verify() on a DynamicProof
, Pickles makes sure that the verification key data matches the hash.
Therefore all manual assertions have to be made on the vk's hash and it can be assumed that the vk's data is checked to match the hash if it is used with verify().
ProofBase
<Input
,Output
>
• Input
• Output
new DynamicProof<Input, Output>(__namedParameters: {
"maxProofsVerified": 0 | 1 | 2;
"proof": unknown;
"publicInput": Input;
"publicOutput": Output;
}): DynamicProof<Input, Output>
• __namedParameters
• __namedParameters.maxProofsVerified: 0
| 1
| 2
• __namedParameters.proof: unknown
• __namedParameters.publicInput: Input
• __namedParameters.publicOutput: Output
DynamicProof
<Input
, Output
>
maxProofsVerified: 0 | 1 | 2;
proof: unknown;
publicInput: Input;
publicOutput: Output;
shouldVerify: Bool;
optional usedVerificationKey: VerificationKey;
static featureFlags: FeatureFlags = FeatureFlags.allNone;
As the name indicates, feature flags are features of the proof system.
If we want to side load proofs and verification keys, we first have to tell Pickles what shape of proofs it should expect.
For example, if we want to side load proofs that use foreign field arithmetic custom gates, we have to make Pickles aware of that by defining these custom gates.
Note: Only proofs that use the exact same composition of custom gates which were expected by Pickles can be verified using side loading. If you want to verify any proof, no matter what custom gates it uses, you can use FeatureFlags.allMaybe. Please note that this might incur a significant overhead.
You can also toggle specific feature flags manually by specifying them here. Alternatively, you can use FeatureFlags.fromZkProgram to compute the set of feature flags that are compatible with a given program.
static maxProofsVerified: 0 | 1 | 2;
static publicInputType: FlexibleProvablePure<any>;
static publicOutputType: FlexibleProvablePure<any>;
get static provable(): ProvableProof<DynamicProof<any, any>, any, any>
ProvableProof
<DynamicProof
<any
, any
>, any
, any
>
publicFields(): {
"input": Field[];
"output": Field[];
}
{
"input": Field[];
"output": Field[];
}
input: Field[];
output: Field[];
toJSON(): JsonProof
verify(vk: VerificationKey): void
Verifies this DynamicProof using a given verification key
• vk: VerificationKey
The verification key this proof will be verified against
void
verifyIf(vk: VerificationKey, condition: Bool): void
• vk: VerificationKey
• condition: Bool
void
static dummy<S>(
this: S,
publicInput: InferProvable<S["publicInputType"]>,
publicOutput: InferProvable<S["publicOutputType"]>,
maxProofsVerified: 0 | 1 | 2,
domainLog2: number): Promise<InstanceType<S>>
• S extends Subclass
<typeof DynamicProof
>
• this: S
• publicInput: InferProvable
<S
["publicInputType"
]>
• publicOutput: InferProvable
<S
["publicOutputType"
]>
• maxProofsVerified: 0
| 1
| 2
• domainLog2: number
= 14
Promise
<InstanceType
<S
>>
static fromJSON<S>(this: S, __namedParameters: JsonProof): Promise<DynamicProof<InferProvable<S["publicInputType"]>, InferProvable<S["publicOutputType"]>>>
• S extends Subclass
<typeof DynamicProof
>
• this: S
• __namedParameters: JsonProof
Promise
<DynamicProof
<InferProvable
<S
["publicInputType"
]>, InferProvable
<S
["publicOutputType"
]>>>
static fromProof<S>(this: S, proof: Proof<InferProvable<S["publicInputType"]>, InferProvable<S["publicOutputType"]>>): InstanceType<S>
Converts a Proof into a DynamicProof carrying over all relevant data. This method can be used to convert a Proof computed by a ZkProgram into a DynamicProof that is accepted in a circuit that accepts DynamicProofs
• S extends Subclass
<typeof DynamicProof
>
• this: S
• proof: Proof
<InferProvable
<S
["publicInputType"
]>, InferProvable
<S
["publicOutputType"
]>>
InstanceType
<S
>
static publicFields(value: ProofBase<any, any>): {
"input": Field[];
"output": Field[];
}
• value: ProofBase
<any
, any
>
{
"input": Field[];
"output": Field[];
}
input: Field[];
output: Field[];
static tag(): {
"name": string;
}
{
"name": string;
}
name: string;
ProofBase.tag