}
	i := new(big.Int).Lsh(h, 1)
	i.Mul(i, i)
	j := new(big.Int).Mul(h, i)

	s1 := new(big.Int).Mul(y1, z2)
	s1.Mul(s1, z2z2)
	s1.Mod(s1, curve.P)
	s2 := new(big.Int).Mul(y2, z1)
	s2.Mul(s2, z1z1)
	s2.Mod(s2, curve.P)
	r := new(big.Int).Sub(s2, s1)
	if r.Sign() == -1 {
		r.Add(r, curve.P)
	}
	yEqual := r.Sign() == 0
	if xEqual && yEqual {
		return curve.doubleJacobian(x1, y1, z1)
	}

	MUL	x2, x2, t0
	ADCS	t0, acc4, acc4
	UMULH	x2, x2, t1
	ADCS	t1, acc5, acc5

	MUL	x3, x3, t0
	ADCS	t0, acc6, acc6
	UMULH	x3, x3, t1
	ADC	t1, acc7, acc7
	// First reduction step
	MUL	acc0, hlp1, hlp0

	MUL	const0, hlp1, t0
	ADDS	t0, acc0, acc0
	UMULH	const0, hlp0, t1

	MUL	const1, hlp0, t0
	ADCS	t0, acc1, acc1
	UMULH	const1, hlp0, y0

	MUL	const2, hlp0, t0
	ADCS	t0, acc2, acc2

    // CHECK: [[ACCUM_MOMENTUM:%.*]] = "tf.Mul"([[ACCUM]], [[MOMENTUM]])
    // CHECK: [[ACCUM_NEW:%.*]] = "tf.AddV2"([[ACCUM_MOMENTUM]], [[GRAD]])
    // CHECK: "tf.AssignVariableOp"([[ACCUM_HANDLE]], [[ACCUM_NEW]])
    // CHECK: [[GRAD_LR:%.*]] = "tf.Mul"([[GRAD]], [[LR]])
    // CHECK: [[MOMENTUM_LR:%.*]] = "tf.Mul"([[MOMENTUM]], [[LR]])
    // CHECK: [[ACCUM_NEW_MOMENTUM_LR:%.*]] = "tf.Mul"([[ACCUM_NEW]], [[MOMENTUM_LR]])

        %cast = "tf.Cast"(%output) : (tensor<1024x24x24x3xi32>) -> tensor<1024x24x24x3xf32>
        %mul = "tf.Mul"(%cast, %scale) : (tensor<1024x24x24x3xf32>, tensor<f32>) -> tensor<1024x24x24x3xf32>
        func.return %mul : tensor<1024x24x24x3xf32>
      }
    }
  )mlir";

  OwningOpRef<ModuleOp> module_op_ref = ParseModuleOpString(kModuleCode);
  const auto test_func =

}

func nextTerm(k int64) {
	y2 := k*2 + 1
	accum.Add(accum, tmp1.Lsh(numer, 1))
	accum.Mul(accum, tmp1.SetInt64(y2))
	numer.Mul(numer, tmp1.SetInt64(k))
	denom.Mul(denom, tmp1.SetInt64(y2))
}

func eliminateDigit(d int64) {
	accum.Sub(accum, tmp1.Mul(denom, tmp1.SetInt64(d)))
	accum.Mul(accum, ten)
	numer.Mul(numer, ten)
}

func main() {
	i := 0
	k := int64(0)
	for {
		d := int64(-1)
		for d < 0 {

                        ZerosLike(scope, x));
    auto gy_1 = Mul(scope, Mul(scope, grad, z), log_x);
    return BinaryGradCommon(scope, op, grad_outputs, gx_1, gy_1);
  } else {
    // There's no sensible real value to return if x < 0, so return 0
    auto log_x = Where3(scope, Greater(scope, x, zero), Log(scope, x),
                        ZerosLike(scope, x));
    auto gy_1 = Mul(scope, Mul(scope, grad, z), log_x);

	t0 := new(fiat.P224Element).Mul(p1.x, p2.x)  // t0 := X1 * X2
	t1 := new(fiat.P224Element).Mul(p1.y, p2.y)  // t1 := Y1 * Y2
	t2 := new(fiat.P224Element).Mul(p1.z, p2.z)  // t2 := Z1 * Z2
	t3 := new(fiat.P224Element).Add(p1.x, p1.y)  // t3 := X1 + Y1
	t4 := new(fiat.P224Element).Add(p2.x, p2.y)  // t4 := X2 + Y2
	t3.Mul(t3, t4)                               // t3 := t3 * t4

	t0 := new(fiat.P521Element).Mul(p1.x, p2.x)  // t0 := X1 * X2
	t1 := new(fiat.P521Element).Mul(p1.y, p2.y)  // t1 := Y1 * Y2
	t2 := new(fiat.P521Element).Mul(p1.z, p2.z)  // t2 := Z1 * Z2
	t3 := new(fiat.P521Element).Add(p1.x, p1.y)  // t3 := X1 + Y1
	t4 := new(fiat.P521Element).Add(p2.x, p2.y)  // t4 := X2 + Y2
	t3.Mul(t3, t4)                               // t3 := t3 * t4

    %0 = "tf.Mul"(%arg1, %arg0) : (tensor<1xf32>, tensor<1xf32>) -> tensor<1xf32>
    func.return %0 : tensor<1xf32>
  }
// CHECK: func private @mul1(%arg0: tensor<1xf32>, %arg1: tensor<1xf32>) -> tensor<1xf32> attributes {tf.entry_function = {inputs = "mul1_y:0,mul1_x:0", outputs = "PartitionedCall:0"}}
// CHECK:   %[[MUL_0:.*]] = "tf.Mul"(%arg1, %arg0) : (tensor<1xf32>, tensor<1xf32>) -> tensor<1xf32>

loop:
	CBZ	R0, done

	LDP.P	16(R2), (R5, R6)
	LDP.P	16(R2), (R7, R8)

	LDP	(R1), (R9, R10)
	ADDS	R4, R9
	MUL	R6, R3, R14
	ADCS	R14, R10
	MUL	R7, R3, R15
	LDP	16(R1), (R11, R12)
	ADCS	R15, R11
	MUL	R8, R3, R16
	ADCS	R16, R12
	UMULH	R8, R3, R20
	ADC	$0, R20

	MUL	R5, R3, R13
	ADDS	R13, R9
	UMULH	R5, R3, R17
	ADCS	R17, R10
	UMULH	R6, R3, R21
	STP.P	(R9, R10), 16(R1)
	ADCS	R21, R11