What is the canonical way to check for approximate zeros in Catch2?

Question

What is the canonical way to compare to approximate zeros in Catch2?

I found this way given a tolerance of 1e-12, but it is not clear it is the best way:

TEST("a approx. equal to b", "[test]"){
    REQUIRE( a - b == (0_a).margin(1e-12) );
}

I am not asking how to compare floats in general. I know that is not a simple problem. I am asking how to use Catch2 given a certain tolerance known in advance.

What follows didn't work, because relative (epsilon) errors do not behave well near zero:

TEST("a approx. equal to b", "[test]"){
    REQUIRE( a - b == (0_a).epsilon(1e-5) );
}

---

Other possible (not so nice)( alternatives seem to be

TEST("a approx. equal to b", "[test]"){
    REQUIRE( std::abs( a - b ) < 1e-12 );
}

TEST("a approx. equal to b", "[test]"){
    REQUIRE_THAT( a - b, WithinULP(0., ???));
}

TEST("a approx. equal to b", "[test]"){
    REQUIRE_THAT( a, WithinULP(b, ???));
}

Answer 1

(a == Approx(b).margin(1e-12))

From the Catch2 GitHub