# Can the problem of unification of Quantum Mechanics and General Relativity be converted into a pure Mathematical problem?

I mean to say, can you make someone understand the problem who only knows Mathematics, even if a Physics description is absent? I mean, describing the problem with unification using pure Mathematical formulation.

Can we make the problem seem as if it has nothing to do with the real world and describe it only as a problem of mathematics?

• The problem of unification is based on a fundamental clash between the principles of GR and QFT. The former is perfectly well understood from the mathematical point of view, but the latter is not. Therefore, the problem of unification cannot be stated purely in mathematical terms. In order to do so, we would need to have a precise, mathematically rigorous definition of QFT, which we don't. – AccidentalFourierTransform Apr 5 '17 at 12:46
• arxiv.org/abs/1407.4748 – Count Iblis Apr 7 '17 at 0:46
• I think that the way you phrased the question led to off-topic answers, particularly the phrased emphasis on "no physics". Are you trying to find the current mathematical formulation of what the clash between QM and GR consists of? (or something like this...) – Helen Apr 9 '17 at 15:28
• @Helen Yeah, I've got the explanations but I've still not got any mathematics. – Dove Apr 10 '17 at 1:39
• @Helen In this question, I asked about the possibility of a pure mathematical formulation. That's why I think explanations came in. I've edited it again to make the answers fit in. And, I've asked this new question demanding only mathematics: physics.stackexchange.com/questions/326493/… – Dove Apr 14 '17 at 1:10

Your other question was marked as a duplicate of this one, so even though I don't believe they're the same question I'll answer that one here.

For reference, the new question was:

Mathematically show that Quantum Mechanics and General Relativity are inconsistent

Whilst it can't be expressed purely mathematically (it relies on dimensionality arguments), it can be shown mathematically. The trouble is, the actual maths itself is pretty hard to follow at the best of times. I'll try to explain what mathematically goes wrong when you move GR into QM, but it might not look like there's much maths because you just won't understand if I start throwing around propagation integrals.

Firstly, though, you need to realise that quantum mechanics isn't a physical theory in the same way that general relativity is. QM is just a mathematical framework in which one can formulate a theory. Without any physical grounding, QM is just the abstract mathematics of infinite dimensional complex Hilbert spaces. The mathematical framework used for general relativity is the one of classical fields, which is the same framework used for classical electromagnetism.

The approach that has traditionally worked with moving a theory formulated in the framework of classical fields into the framework of QM is called 'canonical quantisation', and you end up with a theory formulated in terms of quantum fields (which is just a sub-framework of QM). When you do this you find that to calculate any probability amplitude (which are the things that a measurable in QM) you have to do a Taylor expansion.

This is because in quantum mechanics time evolution is given by exponentiation of the Hamiltonian (group theoretically, the Hamiltonian is the generator of time evolution). So for some initial state, $|i\rangle$:

$$|i\rangle \to |i(t)\rangle = U(t)|i\rangle = e^{-i\int_{0}^{t}H(t')dt'}|i\rangle,$$

The definition of such a thing is its Taylor series:

$$e^{A} \equiv I + A + \frac{A^2}{2} + \cdots + \frac{A^k}{k!} + \cdots$$

In QFT you need to be able to calculate this so you can work out the probability that your initial state evolves into some final state $|f\rangle$:

$$\mathcal{P}(|i\rangle \to |f\rangle) = \langle f|U(t)|i \rangle^2$$

Typically, the Hamiltonian ($H$) looks something like:

$$H(t) = \int d^3x\ g\phi_1(t)\phi_2(t)\cdots$$

where the $\phi_n(t)$ are different fields all interacting and $g$ is coupling constant, which describes the strength of their interaction. This means each term in the expansion looks like:

$$U_n \sim g^n\ \left(\int d^4x\ \phi_1(t)\phi_2(t)\cdots\right)^n$$

Now, $U(t)$ as a whole must be dimensionless (because it is an exponentiation). This means each and every term in the expansion must be dimensionless. Therefore if $g$ has energy dimension $[g]$ (we're in natural units so all units can be expressed as powers of energy), the integral in the above expression must have dimensions $-[g]$. In general, we don't know if these integrals are going to converge, so we introduce a cutoff high energy scale, $\Lambda$, above which we don't bother integrating. (This is equivalent to choosing a small distance cutoff $L\propto\frac{1}{\Lambda}$ below which we don't integrate.) Then we can examine the behaviour as $\Lambda\to\infty$. The finite cutoff means that the integral will scale like $\Lambda^{-[g]}$.

This means that $U_n \sim g^n \Lambda^{-n[g]}$, and so the rate of change of $U_n$ with respect to $\Lambda$ is given by:

$$\frac{dU_n}{d\Lambda} \sim -n[g] g^n \Lambda^{-n[g]-1}$$

This rate of change absolutely cannot be positive for quantum fields to be an effective framework to describe the theory. If it is positive, every single term in the Taylor expansion will diverge and you scattering amplitude just becomes infinite. Since $g$, $n$, and $\Lambda$ are all positive, this means:

$$[g] \geq 0$$

Note that everything I've said so far is generally true about the entire framework of quantum fields. Therefore, in any theory described in terms of quantum fields, all the coupling constants must have non-negative energy dimension. Note that [g] being zero is also not brilliant: in this case you're highly dependent on the specifics of the theory to save you.

Now let's look at general relativity. The field in question for GR is the metric tensor for a 4-dimensional Lorentzian manifold, $g_{\mu\nu}$ (nothing to do with $g$ the coupling constant). The 'free' case of GR is special relativity, in which case:

$$g_{\mu\nu} = \eta_{\mu\nu}$$

where $\eta_{\mu\nu} = \pm\mathrm{diag}(-1,1,1,1)$ is the standard Minkowski metric for a flat spacetime. Let's only consider small perturbations, $\delta_{\mu\nu}$ around the free case:

$$g_{\mu\nu} = \eta_{\mu\nu} + \delta_{\mu\nu}$$

The action (time integral of the Lagrangian) for general relativity is given by the Einstein-Hilbert action:

$$S = \frac{1}{2G}\int d^4x\ \sqrt{-g}R$$

where $G$ is Newton's gravitational constant, $g$ is the determinant of $g_{\mu\nu}$, and $R$ is the Ricci scalar. Expanding this out you get:

$$S = \frac{1}{2G}\int d^4x\ \left((\partial\delta)^2 + (\partial\delta)^2\delta + \cdots\right)$$

For this to fit in with the existing framework for QFT we need to rescale $\delta$, so we say:

$$\delta \to \delta' = \frac{1}{\sqrt{G}}\delta$$

which gives:

$$S = \frac{1}{2}\int d^4x\ \left((\partial\delta')^2 + \sqrt{G}(\partial\delta')^2\delta' + \cdots\right)$$

Then our Hamiltonian for quantum mechanics is given by:

$$H = \int d^3x\ \left(\sqrt{G}(\partial\delta')^2\delta' + \cdots\right)$$

Which means our coupling constants are successive powers of $G^{1/2}$, which has energy units of $-1$. Oh dear, all of our coupling constants have negative energy dimension! This is precisely what we wanted to avoid, as it means that the theory is more dependant on higher energies than lower ones; hence, we can't simply ignore high energies. $G \sim [\mathrm{energy}]^{-2}$ because the gravitational potential, $V$, is dimensionless and depends on mass:

$$V = -\frac{GM}{r}$$

So ultimately general relativity doesn't work as a quantum field theory because its strength scales up with mass and energy, which makes it impossible to apply a high energy cutoff.

That's why quantum fields aren't a good mathematical framework for general relativity. This is telling us that we need a different way to move GR into the framework of quantum mechanics as this naive direct approach yields the wrong theory. Or, maybe we need to develop a new framework that we can move all the theories currently described by QM into along with GR? In any case, such a framework must have a finite 'smallest possible length scale' that provides a natural cutoff for our integral (hello string theory).

A search for a theoretical model of any physics discipline starts by looking at the data/experiments and trying to find a mathematical description that is predictive of new data/experiments. When this is successful the problem is solved because the goal was achieved.

Example: The successful unification of electricity and magnetism, laws and all, by Maxwell's equations.

From the broad mathematical field of second order differential equations , a choice was made, that described data and is accurately predictive:

Maxwell's equations are a set of partial differential equations that, together with the Lorentz force law, form the foundation of classical electromagnetism, classical optics, and electric circuits. They underpin all electric, optical and radio technologies such as power generation, electric motors, wireless communication, cameras, televisions, computers etc. Maxwell's equations describe how electric and magnetic fields are generated by charges, currents and changes of each other. One important consequence of the equations is the demonstration of how fluctuating electric and magnetic fields can propagate at the speed of light

Can we make the problem seem as if it has nothing to do with the real-world and describe it only as a problem of Mathematics?

If there is no description of boundary constraints ( imposed by data) there is no self-contained reason to pick the specific subset of differential equations and their solutions which will describe the physical situation.

The laws and postulates are imposed so as the pick the appropriate subset .

Can the problem of unification of Quantum Mechanics and General Relativity be converted into a pure Mathematical problem?

In a sense it is, with string theories, a lot of mathematics was developed afaik, and it is still at the mathematical stage because the laws or postulates that would pick the correct mathematical subset of the huge number of mathematical string theories so as to have predictive solutions has not been discovered yet.

So I guess my reply is no, because it would make no sense without the postulates/laws.

I mean to say, can you make someone understand the problem who only knows Mathematics but no Physics ? I mean, describing the problem only using the terms of Mathematics and not using a single word related to Physics.

No.

It makes no sense to even try to do that.

Put it this way : I can write down a series of equations which are perfectly sensible mathematically, but have no way of linking them with the real world.

A mathematical system of equations and relations without a connection to the real world is just not a physics model, it's just mathematics - an abstraction.

Let's take a simple case :

$$F = ma$$

It's a perfectly good equation, but if you have no physical meaning for $F$, $m$ and $a$ it's nothing more than a piece of mathematics.

This problems becomes far worse when you deal with GR and QFT because those physical models don't even connect to our common sense easily (it's very hard work just staying grounded in these worlds).

Can we make the problem seem as if it has nothing to do with the real-world and describe it only as a problem of Mathematics?

That's not physics.

The whole point of using mathematics to study the world is to be able to model the behavior of the real world. If you remove the real world, what do you have : just abstract mathematics.

I have to say that I was "brought up" to regard experimental physics as the cornerstone of physics. The goal is, as I was taught, always to find a theory that matches the real world to withing required experimental accuracy, not to construct complicated theories for the mathematical fun of it.

Without experimental evidence leading the way and pointing to flaws in existing theories, we would have no new theories. You cannot divorce the two - it's hand in glove.

And the process of experiment and theory is incremental (or evolutionary, if you prefer). They help each other.

So returning to your question :

Can the problem of unification of Quantum Mechanics and General Relativity be converted into a pure Mathematical problem ?

You cannot divorce physics (even theoretical physics) from the real world, as you have described it.

So a the problem of unification of QM and GR is not doable without reference to the real world.

This problem is much harder for the very simple reason that the process of building such a theory is on the leading edge of out knowledge, both theoretical and experimental. Developments in both will be required to develop a robust theory incorporating concepts from both QM and GR which matches the real world.

• I don't think the OP meant it this way. S/he didn't want to abandon the real world, s/he just emphasizes mathematical description (and its current state). – Helen Apr 9 '17 at 15:25
• Agreed with @Helen, this answer does not answer any part of the question, and hence is completely off topic in my opinion – zzz Apr 10 '17 at 20:36

Sure, you can say Please find a mathematical structure which resembles QFT under certain circumstances and resembles GR under certain different circumstances. GR and QFT can be precisely defined mathematically (e.g. link), and the words "resembles" and "circumstances" can be made precise too.†

But I don't see how this is a "purely mathematical problem". Anyone working on this problem is by definition doing physics. They might be a mathematician, and they might be using mathematical methods and doing math, but they are also doing physics. Some physics problems are also math problems! You can state the problem without using any "physics words" ... for example instead of saying "energy" you can say "a certain real-valued function which we shall call χ" ... but what does that accomplish? It's just obfuscation.

BTW, as far as I understand, we have very good reasons to believe (1) String theory is one such mathematical structure, and (2) there are no other such structures (other than things that are essentially equivalent to string theory) ... But neither of those statements has been rigorously proven!

†(Per @Void's comments, it may be impossible or very difficult to make it precise in an absolutely universal way. But once we have some candidate mathematical structure and its mappings to GR and to QFT, it can be made precise. But "stating the requirements precisely" is not really required for this question. Mathematicians work on plenty of problems which cannot be precisely defined in advance! Things like "Please find an explanation of a weird apparent-coincidence where the exact same number comes up in different branches of math" or "Please find a way to generalize such-and-such theorem using the framework of blah-blah theory" or whatever.

• Is axiomatic QFT considered so successful that QFT is precisely defined mathematically? – innisfree Apr 6 '17 at 13:04
• I do not agree that the "circumstances" and "resembles" are well defined. There is no well-defined criterion on when and where exactly we should see the quantum-gravitational deviations from general relativity and/or in which limit exactly we should recover GR from a full-blown quantum gravity. (Similarly to electromagnetism, there is no good "strength" of the relativistic gravitational field.) Also, objects such as the S-matrix and the measurement process might look extremely different mathematically and it is hard to pin down what precisely we expect from them. – Void Apr 6 '17 at 16:29
• @Void - There are plenty of experimental null results checking for deviations from GR or QFT. Each of these can be recast as a certain inequality which the mathematical structure must obey. There's no point in doing this -- it's much better to think of the experimental results as experimental results -- but if you really wanted to, I don't see why it should be impossible to do that, and in a very specific and rigorous way... – Steve Byrnes Apr 7 '17 at 0:29
• @SteveByrnes I am very well aware of that but I don't think you understand what I am saying. You have to formulate your criteria for mathematical objects, not physical experimental quantities. For instance, the electroweak unification was well posed in this way because it was asking for a quantum field theory which would show the behaviour of Fermi theory at these and these collision energies. I.e. we had the mathematical framework well connected to observations with a certain parameter and asymptotic properties at the value of this parameter. – Void Apr 7 '17 at 10:00
• However, this is not the case for quantum gravity because we aren't asking for a "quantum field theory of this and this behaviour at these and these collision energies". We have no mathematical framework in which we are asking for certain mathematical objects. Since the mathematical framework is undefined, we cannot convert our experimental requirements into mathematical statements. My second objection was more theoretical: it is hard to define where we should see quantum gravity, because in a sense, Einstein gravity is always "no-gravity", free-fall in some frame. – Void Apr 7 '17 at 10:05

It sounds like you want something like: If some equations or mathematical constraints have a solution, then GR+QM could work, else it cannot.

There are theorems called "no go" theorems which allow mathematically ruling out theories with certain properties. For example the Coleman-Mandula theorem, later extended by the Haag–Łopuszański–Sohnius theorem, greatly restricts how gravity could be added into a quantum field theory.

However this must be done within some "framework" where one is discussing the space of theories. Therefore even "no go" theorems can have loopholes if you break some of the underlying assumptions of what the theories under consideration can look like.

I feel this means that one would first have to choose a framework or approach to quantum gravity if one wished to formulate the existence of a quantum theory of gravity as a mathematical question. As one example, depending on what you consider an acceptable framework, string theory could already be considered a proof that a quantum theory can reduce to GR in the classical limit.

As another example, there is an approach to quantum gravity called Asymptotic Safety. The researchers are looking for a UV fixed point in the renormalization flow of GR. Currently the researchers believe they have evidence for such a point, but cannot fully solve it. If such a point is found with the right properties, then the divergences of quantum gravity can be tamed. So within this approach, the math question of finding or disproving the existence of a fixed point with certain properties would definitively answer yes or no for this particular approach to quantum gravity.

To the extent that a framework is mathematically well defined and consistent, I think any such approach could indeed be formulated as a mathematical existence conjecture.

If there's a single problem which has stood for the challenge of quantum gravity as a whole, it is probably the non-renormalizability of quantum gravity: The ultraviolet divergences of perturbatively quantized general relativity cannot be absorbed into a finite number of Lagrangian parameters.

Although it uses some physics terminology, that is essentially a mathematical proposition. It is saying that if you apply the usual algorithms of quantum field theory to the Lagrangian of general relativity, you end up with an infinite number of distinct divergent integrals that need to be renormalized.

See this lecture for more.

• Good. Finally an answer that attempts to address some part of the question without philosophical BS. – zzz Apr 14 '17 at 1:52

Sure one can convert the problem of unification of Quantum Mechanics and G.R into a mathematical problem. A Mathematician can understand equations from the Q.M and G.R. For eg. A mathematician can understand Schrödinger's equation very easily but would not get physical insights about the atomic model. As said by Anna V "A theoretical model of any physics discipline starts by looking at the experiments and trying to find a mathematical description that is predictive of new experiments". So when such a theoretical model is made then our goal is achieved.

When a mathematician will solve this unification problem, he will take into consideration of all the possible dimensions (I mean he will solve this problem generally) using math tools and stuff. He will get to a solution for the same problem. So far good. But will see the solution just as a solution without any physical insights. He will not consider boundary conditions and not consider where his solution can be applied to physical world.

As the example of string theory given by Anna V we can see hat string theory contains lot of mathematics and still is at a mathematical age. So Mathematicians will enjoy solving the theory in generalized form and we will get many different solutions of the same theory but only some solutions can be explained in the real world. So Physicist might understand some of the solutions/theories but he would not get any information from the other solutions/theory. On the other hand Mathematician will not be able to use such solution/theory on real world.

A Physicist job is to observe phenomenon and derive equations out of it (Its not the only job of Physicist). So when a Physicist give such equations to a Mathematician he will solve those equations for Physicist but Mathematician might come out with some different theory which for which Physicist don't find any use in the real world.

As it is said that less than 5% of mathematics is used in Physics. Hence the remaining 95% of mathematics in my view is just waiting to find its uses in the Physical world. So, to your question Can we make the problem seem as if it has nothing to do with the real-world and describe it only as a problem of Mathematics? My view is that such a problem will be considered in that 95% of mathematics which is not used in Physical world but that mathematics might have uses in the coming future.

Yes Problem of unification of Q.M and G.R can be converted into a pure mathematical problem but will be just mathematical statements until we find the use of the solutions of the problem to be used in the real world.

25% Yes and 75% No.

Yes, since without experimental data on our side many theories are counting on mathematical consistency to carry on. String theory here fits very well this description.

No, since mathematics give us much more than what is needed to describe the real word. If you don't take insight from physics you would be lost into a (literally) infinite amount of mathematics.

SPECULATION ON: It is actually a very deep question to ask what are the minimum number of physical inputs needed in order to describe our universe. It would be nice if they could be only an handful. I would bet on the following minimal set: mass and charge of the electron and quarks, speed of light, fine structure constant and planck length. Notable predicted quantities: strength of gravity and other forces, number of non compact dimensions, all masses and charges of other particles, time asymmetry.

EDIT: considering your edit, I will rephrase your question into the following: Can I explain quantum gravity without ever using the proposition "as we know from experimental data". Again, yes and no and the above applies.

I suggest to shift the question to something of which we already know the answer. For instance, "Can the problem of unification of Electomagnetism and Weak Force be converted into a pure Mathematical problem?". If you consider enough the statement "a Gauge theory with symmetry group $U(1) \times SU(2)$ is consistent", then yes applies. But even no applies, since you cannot decide between that theory and a theory with $SU(5)$, which gives a consistent but not phenomenologically allowed theory (no proton decay in our universe).

There are various problems with QM which leads to incompatibility between QM and Relativity. You asked for a mathematical description, you can see it this way:

If you write the Schrodinger's equation we have:

$i\hbar \frac{\partial \psi}{\partial t} = -\frac{\hbar^{2}}{2m}\frac{\partial^2 \psi}{\partial^2 x}$

where, the LHS represents the energy operator in QM and RHS is just the Hamiltonian with $V(x)=0$, i.e. for a free particle. This equation clearly shows the incompatibility between QM and Relativity as we have $1^{st}$ derivative wrt $t (time)$ on LHS and $2^{nd}$ derivative wrt $x (space)$ on RHS. Whereas in relativity we have a completely different formulation of space and time not related by such condition. So this should be a mathematical formulation to understand the incompatibility of QM and Relativity.

• This problem is rectified in a way that's fully compliant with QM and SR through quantum field theory. OP's question was about general, not special, relativity. – gautampk Apr 15 '17 at 2:45