y = Lgamma(scope_, x);
        break;
      case ERF:
        y = Erf(scope_, x);
        break;
      case ERFINV:
        y = Erfinv(scope_, x);
        break;
      case NDTRI:
        y = Ndtri(scope_, x);
        break;
    }

    float max_error;
    TF_ASSERT_OK((ComputeGradientError<X_T, Y_T, float>(scope_, x, x_data, y,

    // CHECK-NOT: tf.Erfinv
    %0 = "tf.Erfinv"(%input) : (tensor<4xf32>) -> tensor<4xf32>
    func.return %0 : tensor<4xf32>
  }

  // CHECK-LABEL: @ndtri
  func.func @ndtri(%input: tensor<4xf32>) -> tensor<4xf32> {
    // CHECK-NOT: tf.Ndtri
    %0 = "tf.Ndtri"(%input) : (tensor<4xf32>) -> tensor<4xf32>
    func.return %0 : tensor<4xf32>
  }

  // CHECK-LABEL: @fake_param
  func.func @fake_param() -> tensor<4xf32> {

  auto x = ConjugateHelper(grad_scope, op.input(0));
  // grad * sqrt(2 * pi) * exp(ndtri(x) ** 2 / 2)
  auto dx = Mul(
      grad_scope, Mul(grad_scope, grad, root_two_pi),
      Exp(grad_scope, Div(grad_scope, Square(grad_scope, op.output(0)), two)));
  grad_outputs->push_back(dx);
  return grad_scope.status();
}
REGISTER_GRADIENT_OP("Ndtri", NdtriGrad);

Status LgammaGrad(const Scope& scope, const Operation& op,

            "Atan", "Atanh", "Ceil", "Cos", "Cosh", "Sin", "Exp", "Expm1",
            "Floor", "IsFinite", "IsInf", "IsNan", "Inv", "Reciprocal", "Log",
            "Log1p", "Invert", "LogicalNot", "Ndtri", "Neg", "Rint", "Round",
            "Rsqrt", "Sigmoid", "Sign", "Sinh", "Softplus", "Softsign", "Sqrt",
            "Square", "Tan", "Tanh", "Real", "Imag", "Erf", "Erfc", "Erfinv",
            "Lgamma", "Digamma",

	"𝓌":                            "\U0001d4cc",
	"⋂":                            "\u22c2",
	"◯":                           "\u25ef",
	"⋃":                            "\u22c3",
	"▽":                           "\u25bd",
	"𝔵":                             "\U0001d535",
	"⟺":                           "\u27fa",
	"⟷":                           "\u27f7",

		"wscr;":                            '\U0001D4CC',
		"xcap;":                            '\U000022C2',
		"xcirc;":                           '\U000025EF',
		"xcup;":                            '\U000022C3',
		"xdtri;":                           '\U000025BD',
		"xfr;":                             '\U0001D535',
		"xhArr;":                           '\U000027FA',
		"xharr;":                           '\U000027F7',

    static bool isCompatibleReturnTypes(TypeRange inferred, TypeRange actual) {
      return ArraysAreCastCompatible(inferred, actual);
    }
  }];
}

def TF_NdtriOp : TF_Op<"Ndtri", [Pure]> {
  let summary = "";

  let arguments = (ins
    TF_FloatTensor:$x
  );

  let results = (outs
    TF_FloatTensor:$y
  );

        eagerly.
    *   Added complex128 support to `RFFT`, `RFFT2D`, `RFFT3D`, `IRFFT`,
        `IRFFT2D`, and `IRFFT3D`.
    *   Add `pfor` converter for `SelfAdjointEigV2`.
    *   Add `tf.math.ndtri` and `tf.math.erfinv`.
    *   Add `tf.config.experimental.enable_mlir_bridge` to allow using MLIR
        compiler bridge in eager model.
    *   Added support for MatrixSolve on Cloud TPU / XLA.