Renormalization That ClicksEpisode 6 / The two places it leaks

Modern physics' two greatest puzzles have exactly the same shape ── coarse-graining fails here and only here

The two places it leaks The hierarchy problem and the cosmological constant problem.
These two demand cancellations of 10⁻³⁴ and 10⁻¹²⁰, and that is nothing but the flip side of
the Standard Model having exactly two relevant operators.

Tools you'll need: Episode 4's relevant / irrelevant, a feel for orders of magnitude, exponents The heart of this episode: a relevant operator cannot shield the floor above

In Episode 4 we saw the magic of coarse-graining ── details from the floor above die as \((E/\Lambda)^n\), and the floor below closes in its own language. Hence chemists need not know about quarks. But the magic has exactly two holes. The Higgs mass and the cosmological constant. These two alone cannot shield themselves from the far-off floor of the Planck scale and take the hit square on. And the observed values are \(10^{34}\) and \(10^{120}\) times smaller than that hit ── as if the bare value and the quantum correction had agreed in advance to cancel to the 34th digit, to the 120th digit. These are modern physics' two greatest puzzles. And the important thing is ── they have the same shape. This episode reads that shape.

01Why are most quantities protected?

Start with why things are usually fine. In quantum theory every quantity receives corrections from the world at higher energies. Most of the time this is harmless because a symmetry protects it.

QuantityProtecting symmetryHow the correction entersDanger
electron masschiral symmetry\(\delta m \propto m\ln(\Lambda/m)\) ── multiplicativelysafe (zero stays zero)
photon / gluon massgauge symmetryforbidden outright (zero)safe
couplings such as charge(dimension-4 operators)slowly, as \(\ln\Lambda\)safe
Higgs mass²none\(\delta m_H^2 \propto \Lambda^2\) ── additivelydangerous
cosmological constant (vacuum energy)none\(\delta\rho \propto \Lambda^4\)most dangerous

The electron row is the crucial one. If the correction is proportional to \(m\) (multiplicative), then a small starting value keeps a small correction ── so it can stay small. That is called being technically natural. If instead the correction enters additively as \(\Lambda^2\), a small starting value is swamped and smallness cannot be maintained.

02The Higgs and its 34 digits

Order of magnitude ── the Higgs mass cancellation

As long as you trust the Standard Model, the quantum correction enters as the square of the cutoff \(\Lambda\). Taking \(\Lambda\) to be the Planck scale, \(1.22\times10^{19}\ \mathrm{GeV}\):

$$\delta m_H^2 \sim \Lambda^2 \approx 1.5\times10^{38}\ \mathrm{GeV^2}$$

But the measured Higgs mass is 125 GeV, i.e.

$$m_H^2 \approx 1.6\times10^{4}\ \mathrm{GeV^2}$$

The ratio is \(1.6\times10^{4}/1.5\times10^{38}\approx10^{-34}\). The bare value and the correction must therefore agree and cancel to the 34th digit. A trillionth of a trillionth of another hundred-millionth. Too painful to call a coincidence ── hence the hierarchy problem.

03The cosmological constant and its 120 digits

Order of magnitude ── this one is worse by orders of magnitude

The vacuum has quantum-fluctuation energy too, naively of order \(\Lambda^4\):

$$\rho_{\text{theory}}\sim \Lambda^4 \approx (1.22\times10^{19})^4 \approx 2\times10^{76}\ \mathrm{GeV^4}$$

The value inferred from the observed accelerating expansion is

$$\rho_{\text{observed}}\sim 10^{-47}\ \mathrm{GeV^4}$$

The ratio is about \(10^{-123}\) (often quoted as roughly \(10^{-120}\), depending on how you estimate). It is sometimes called the worst theoretical prediction in the history of physics. This is the cosmological constant problem.

04These two have the same shape

Here is the heart. Translated into Episode 4's language, why it is only these two becomes a single line.

The heart of this episode ── a relevant operator cannot shield the floor above
Operator dimension dWhat coarse-graining doesExamples in the Standard Model
\(d>4\): irrelevantshrinks as \((E/\Lambda)^{d-4}\)almost everything (which is why the world has layers)
\(d=4\): marginaldrifts logarithmicallygauge couplings, Yukawa couplings
\(d<4\): relevantgrows as you head to low energyexactly two of them

Those two are ── the identity operator with \(d=0\) (the cosmological constant) and \(H^\dagger H\) with \(d=2\) (the Higgs mass²). Every other potentially relevant term (a fermion mass \(\bar\psi\psi\) with \(d=3\), a gauge-field mass \(A^2\) with \(d=2\)) is forbidden by a symmetry. So the universe has exactly two unprotected relevant operators ── and those two are precisely modern physics' two greatest puzzles.

Put differently: coarse-graining works only in the irrelevant directions. In the relevant directions, coarse-graining does not discard information ── it magnifies the floor above and brings it down. So it cannot be shielded. The hierarchy problem and the cosmological constant problem are not "new puzzles"; they are the flip side of the very fact that the world has layers. Why the magic works and why it fails in exactly two places come from the same theorem.

05Try it ── what does matching 34 digits actually look like?

The easiest way to feel how strange "agreement to 34 digits" is, is to lay the digits out. The figure below places two enormous numbers ── the bare value and the quantum correction ── side by side, with a slider for "how many digits agree."

Matching digits cancel and vanish; the leftovers are the physical quantity we observe. For the Higgs you need about 34 digits, for the cosmological constant about 120, before you reach the observed value ── switch between the two problems with the button and compare how far away the mark is.

Figure: the digits of the bare value and the quantum correction laid side by side. Grey = digits that cancelled, coloured = digits that survive. Moving the slider changes the residue (the observed physical quantity). The point is to see how strange the required number of matching digits is
digits that cancelled surviving digits (= the observed quantity)

06So why is it like that?

No answer yet. There are three leading positions.

PositionClaimWhere it stands
① a new symmetry protects itsupersymmetry or similar makes the \(\Lambda^2\) correction cancel against contributions from partner particles ── i.e. "no protecting symmetry" is simply wrongthe most beautiful solution, but no superpartners have shown up at the LHC in the expected mass range. Optimism has cooled
② anthropics / multiversethere are countless universes and the value varies; observers only exist where galaxies can formWeinberg used this reasoning in 1987 to predict an upper bound on the cosmological constant, and the 1998 observation fell within it. But testing it is hard
③ "naturalness" is the wrong criterionthe intuition that "matched digits are unnatural" may itself be a groundless aesthetica growing number of researchers take this line. But its predictive power is weak
The single most important sentence in this episode Coarse-graining leaks in exactly two places. And those two places come from precisely the same theorem that makes coarse-graining work.
"The world has layers" and "there is a hierarchy problem" are two sides of one coin. If there were no relevant operators at all, there would be no cosmological constant and no Higgs mass ── no scale of mass would be set, and therefore no matter and no atoms. The two holes that trouble us are also the two things that give the world a size.
◇ ◇ ◇
The honest line ── the reach of the phrase "matching digits"

Established: that quantum corrections to the Higgs mass squared enter as the square of the cutoff (scalars have no symmetry protecting their mass); that fermion masses are protected by chiral symmetry and gauge-field masses by gauge symmetry; the classification of operators as relevant/marginal/irrelevant by dimension; that the unprotected relevant operators of the Standard Model are the identity operator (cosmological constant) and \(H^\dagger H\); that the ratios between the measured values (\(m_H\approx125\) GeV, the cosmological constant energy density) and naive Planck-scale estimates are around \(10^{-34}\) and \(10^{-120}\); and that Weinberg (1987) derived an anthropic upper bound on the cosmological constant which later observations fell within ── all established physics and standard understanding.

Things to be careful about: (1) The language of "\(\Lambda^2\) corrections cancelling against a bare value" is tied to cutoff regularisation. In other schemes such as dimensional regularisation the \(\Lambda^2\) terms never appear and the problem instead presents itself as a hierarchy between scales (the masses of new particles, say). The figure's "match the digits" picture is a visualisation for intuition, not the literal calculation. (2) The \(10^{-120}\) for the cosmological constant swings between roughly \(10^{-120}\) and \(10^{-123}\) depending on where you cut. (3) The very premise that "unnatural means it needs explaining" (naturalness) is not an established physical principle but a guideline, and its validity is currently under debate ── that is position ③ above. (4) This episode is therefore an organisation of an unsolved puzzle in the language of the renormalization group, not a solution.

Exercises (solvable with this episode's ideas)
  1. Why can the electron mass stay small while the Higgs mass cannot?
    See the answer
    The electron mass is protected by chiral symmetry, so the correction enters multiplicatively as \(\delta m\propto m\ln(\Lambda/m)\) (small stays small). The Higgs has no such symmetry, and its correction enters additively as \(\delta m_H^2\propto\Lambda^2\), so smallness cannot be maintained.
  2. State the relation between "the world has layers" and "there is a hierarchy problem."
    See the answer
    Two sides of the same theorem. Coarse-graining cuts off the floor above only in the irrelevant directions (\(d>4\)); in relevant directions (\(d<4\)) the influence of the floor above is magnified on the way down. There are exactly two unprotected relevant operators (the cosmological constant and \(H^\dagger H\)), and they are the two great puzzles.
  3. Why is the cosmological constant problem "worse" than the Higgs hierarchy problem? Answer with orders of magnitude.
    See the answer
    The Higgs faces a \(\Lambda^2\) correction with a ratio of \(10^{-34}\); the cosmological constant faces \(\Lambda^4\) with about \(10^{-120}\). The higher dimension means a bigger correction and a matching requirement more than three times as severe.
  4. What would the universe be like with no relevant operators at all?
    See the answer
    Nothing would set a scale of mass, so neither particle masses nor vacuum energy would be determined and no structures with a size ── atoms ── could form. The two troublesome holes are also what give the world a size.

Episode 6 summaryWhy the magic works and why it fails twice come from one theorem

Quantum corrections are usually protected by symmetry ── electron mass by chiral symmetry, gauge-field mass by gauge symmetry. Protected quantities take their corrections multiplicatively and can stay small. But the Higgs mass² and the vacuum energy have no protecting symmetry and take direct additive hits of \(\Lambda^2\) and \(\Lambda^4\). Matching the observed values then requires cancellations of 34 and 120 digits.

And it is no accident that there are exactly two. In Episode 4's language, the Standard Model has exactly two unprotected relevant operators (\(d<4\)) ── the \(d=0\) identity operator (the cosmological constant) and the \(d=2\) \(H^\dagger H\) (the Higgs mass²). In relevant directions, coarse-graining doesn't shield the floor above; it magnifies and delivers it. So "the world has layers" and "there is a hierarchy problem" are two sides of one coin. No answer yet (a new symmetry / anthropics / naturalness being the wrong criterion, all three still in play).

This document is Episode 6 of the "Renormalization That Clicks" series, a reading piece for physics-loving high-schoolers and undergraduates. Quadratically divergent corrections to a scalar mass squared; protection of fermion masses by chiral symmetry and of gauge-field masses by gauge symmetry; 't Hooft's technical naturalness; the classification of operators as relevant/marginal/irrelevant by dimension; that the Standard Model's unprotected relevant operators are the identity operator and \(H^\dagger H\); the roughly \(10^{-34}\) tuning implied by a 125 GeV Higgs against the Planck scale; the roughly \(10^{-120}\) ratio between the theoretical estimate and observed value of the vacuum energy; and Weinberg's (1987) anthropic upper bound on the cosmological constant ── all established physics and standard understanding. That the phrase "a \(\Lambda^2\) correction cancelling against a bare value" is tied to cutoff regularisation (and appears differently in dimensional regularisation), that \(10^{-120}\) depends on how one estimates, and that naturalness is a guideline under debate rather than an established principle, is spelled out in the body's "honest line." The figure is a visualisation for intuiting the strangeness of the required matching, not a reproduction of the actual calculation. ── To print, use your browser's "Print" and "Save as PDF" (in the print version the slider and answers are frozen and hidden). Adjacent episodes: Episode 5, When coarse-graining breaks / Episode 7, Resolution as a dimension / Contents.

Print / make a PDF: ⌘+P (Ctrl+P on Windows). On screen the slider changes how many digits match, and the button switches between the Higgs and the cosmological constant. "Snap to the observed value" jumps to the required number of digits. "See the answer" opens each solution.