Question: how to capture the error of api.AssertIsEqual() func ?

[George551556]

How can I intercept and handle the error that occurs from the function api.AssertIsEqual(0, 1) without the program terminating with an error?

[readygo67]

api.Assertxx funcitos will assert if the condition fails.
If you want just test the condition, I guess api.IsZero(x-y), or api.Cmp(x, y) can be used.

[George551556]

I know, but the functionality I want to implement is: in the main function, I execute groth16.Verify(proof, vk, publicWitness) for verification, check for any errors, and then perform corresponding operations. It's something like the following code:

err := groth16.Verify(proof, vk, publicWitness)
if err == nil {
    fmt.Println("verify success!!!")
} else {
    fmt.Println("other op")
}

could id come true?

[readygo67]

Please check whether the following code is what you want.
In this case, 1 correct and 1 incorrect witness is provided, the latter one trig an error in roth16.Verify.

func TestCubic_Groth16_BN254(t *testing.T) {
	field := ecc.BN254.ScalarField()
	assert := test.NewAssert(t)
	var circuit Circuit
	assignment := &Circuit{
		X: 3,
		Y: 35,
	}

	_r1cs, err := frontend.Compile(field, r1cs.NewBuilder, &circuit)
	assert.NoError(err)

	pk, vk, err := groth16.Setup(_r1cs)
	assert.NoError(err)

	wit, err := frontend.NewWitness(assignment, ecc.BN254.ScalarField())
	assert.NoError(err)
	proof, err := groth16.Prove(_r1cs, pk, wit)
	assert.NoError(err)

	//rebuild the pubWitness, verify pass
	values1 := make(chan interface{})
	go func() {
		defer close(values1)
		values1 <- fr_bn254.NewElement(35)
	}()

	pubWit, _ := witness.New(ecc.BN254.ScalarField())
	pubWit.Fill(1, 0, values1)

	err = groth16.Verify(proof, vk, pubWit)
	if err != nil {
		fmt.Printf("err: %v\n", err)
	} else {
		fmt.Printf("verification pass\n")
	}

	//rebuild the pubWitness, verify fail
	values2 := make(chan interface{})
	go func() {
		defer close(values2)
		values2 <- fr_bn254.NewElement(36)
	}()
	
	pubWit, _ = witness.New(ecc.BN254.ScalarField())
	pubWit.Fill(1, 0, values2)

	err = groth16.Verify(proof, vk, pubWit)
	if err != nil {
		fmt.Printf("err: %v\n", err)
	} else {
		fmt.Printf("verification pass\n")
	}

}

[George551556]

Basically the same, thank you for the components

[ivokub]

Are you trying to verify the proof in-circuit or out-circuit? For out-circuit implementation you can switch on err to see if the proof verified or not. For in-circuit implementation we only provide assertion that a valid proof verifies, but do not provide an invalidity proof. I think in general it is possible to provide it by replacing AssertIsEqual here with IsEqual. We have currently implemented IsEqual only for BN254, but can possibly extend the interface in std/algebra to provide the method for all curves.

[ivokub]

Converted to discussion as seems to be question about gnark usage, not an issue.