Quantum Gravity signatures high sigma?

Question

As an Amazon Associate, I earn from qualifying purchases. For more practice problems like this, see Schaum’s Outline of College Physics, 12th Edition.

1. What the question is really asking

The original post is a frustrated rant about “quantum‑gravity signatures” and the claim that some experiments have reported a high‑σ (high‑sigma) detection of such a signature.
In plain language the student wants to know:

What does “high σ” mean?
How do you calculate the sigma (σ) significance of a possible quantum‑gravity effect?
What would be considered a convincing (i.e., statistically robust) result?

Below is a step‑by‑step guide that shows how a physicist turns raw data into a “σ‑level” statement, illustrated with a generic quantum‑gravity‑type measurement (e.g., a tiny deviation in the speed of light for high‑energy photons, a modification of the dispersion relation, or an unexpected noise feature in a gravitational‑wave interferometer).

2. Detailed Worked Solution

2.1 Set up the hypothesis test

Symbol	Meaning
(H_0)	Null hypothesis – “no quantum‑gravity effect; the data are described entirely by the Standard Model (or classical GR)”.
(H_1)	Alternative hypothesis – “there is a genuine quantum‑gravity signal of magnitude (S)”.
(x)	Measured quantity (e.g., a time‑delay, a phase shift, a spectral index).
(\mu_0)	Expected value of (x) under (H_0).
(\sigma_{\text{tot}})	Total (one‑sigma) uncertainty on the measurement, including statistical + systematic contributions.
(\hat S)	Best‑fit signal strength obtained from the data (difference between the observed value and the null expectation).

The goal is to quantify how far the observed data lie from the null‑hypothesis expectation, measured in units of the total uncertainty. This distance is the sigma (σ) significance.

2.2 Combine all sources of error

Statistical error ((\sigma_{\text{stat}})):
Comes from the finite number of events, photon counts, etc.
Usually obtained from the square‑root of the number of counts or from a fit covariance matrix.
Systematic error ((\sigma_{\text{sys}})):
Calibration, timing offsets, detector alignment, theoretical model uncertainties, etc.
Total error (assuming the two are independent):

[ \sigma_{\text{tot}} = \sqrt{\sigma_{\text{stat}}^{2} + \sigma_{\text{sys}}^{2}} . ]

If the systematic uncertainties are correlated, one must build the full covariance matrix and invert it; the simple quadrature formula above is a good first‑order approximation.

2.3 Compute the test statistic

The most common test statistic for a single measured quantity is the standardised residual (also called the “z‑score”):

[ z \;=\; \frac{x - \mu_0}{\sigma_{\text{tot}}}. ]

Interpretation:

(z = 0) → perfect agreement with the null hypothesis.
(z = +1) → the measurement is 1‑σ above the null prediction.
(z = -2) → the measurement is 2‑σ below the null prediction, etc.

When the measurement is a fit parameter (e.g., the coefficient of a Lorentz‑invariance‑violating term), the same formula holds, but (x) is replaced by the best‑fit value (\hat S) and (\mu_0 = 0) (because under (H_0) the signal strength is zero).

[ z \;=\; \frac{\hat S}{\sigma_{\hat S}} . ]

2.4 Convert sigma to a p‑value (optional)

Physicists often quote the p‑value (probability that a fluctuation at least as extreme as observed would occur under (H_0)). For a two‑sided Gaussian:

[ p = 2\,\bigl[1-\Phi(|z|)\bigr], ]

where (\Phi) is the cumulative distribution function of the standard normal distribution.

Typical benchmarks:

σ (one‑sided)	Two‑sided p‑value	Common jargon
1	0.317	“not significant”
2	0.0455	“evidence” (≈ 2 σ)
3	0.0027	“strong evidence”
5	(5.7\times10^{-7})	“discovery” (5 σ)

In high‑energy physics and quantum‑gravity searches, a 5‑σ result is the community standard for claiming a discovery, because it reduces the chance of a statistical fluke to less than one in a million.

2.5 Example: Time‑of‑flight delay of high‑energy photons

Suppose a space‑based gamma‑ray telescope measures the arrival times of two photons from a distant gamma‑ray burst (GRB).

Observed delay: (\Delta t_{\text{obs}} = 0.42 \pm 0.12) ms (stat) (\pm 0.08) ms (sys).
Null‑hypothesis prediction: (\Delta t_{0}=0) (no quantum‑gravity dispersion).

Step 1 – total error

[ \sigma_{\text{tot}} = \sqrt{0.12^{2} + 0.08^{2}} = \sqrt{0.0144 + 0.0064} = \sqrt{0.0208} \approx 0.144\;\text{ms}. ]

Step 2 – sigma significance

[ z = \frac{0.42\ \text{ms} - 0}{0.144\ \text{ms}} \approx 2.92 . ]

Step 3 – p‑value (two‑sided)

[ p = 2\bigl[1-\Phi(2.92)\bigr] \approx 2(1-0.9982) \approx 0.0036 . ]

Interpretation: The observed delay corresponds to a ~3‑σ effect. It is intriguing, but not yet a discovery; further independent bursts and a careful treatment of possible astrophysical systematics are required.

2.6 General recipe for any quantum‑gravity search

Step	Action
1	Write down the observable that a quantum‑gravity model predicts (e.g., a modification of the dispersion relation (E^{2}=p^{2}c^{2}+m^{2}c^{4} + \eta \frac{p^{3}}{M_{\text{Pl}}})).
2	Obtain the null expectation ((\eta=0)) and the signal expectation for a given model parameter.
3	Collect the data and perform a fit that returns the best‑fit value (\hat\eta) and its uncertainty (\sigma_{\hat\eta}).
4	Compute the sigma: (z = \hat\eta/\sigma_{\hat\eta}).
5	If several independent data sets exist, combine them by adding inverse variances (i.e., weighted average) or by constructing a global likelihood.
6	Quote the resulting σ and, optionally, the corresponding p‑value.
7	Discuss look‑elsewhere effect (multiple tests on different energies, directions, etc.) – correct the σ accordingly.
8	State whether the result meets the community threshold (≥ 5 σ for a claim of discovery).

3. Final Answer

The “high‑σ” statement is simply a way of saying:

*The measured quantity differs from the prediction of ordinary physics by z standard deviations, where

[ z = \frac{\text{observed – expected}}{\text{total uncertainty}} . ]

If (z \ge 5) (≈ (p < 6\times10^{-7})), the physics community would normally call the result a discovery; lower values (2–3 σ) are regarded as hints that need confirmation.

Thus, to evaluate any claimed quantum‑gravity signature you:

Quantify all sources of error and combine them into a single (\sigma_{\text{tot}}).
Compute the residual (observed minus null expectation).
Divide the residual by (\sigma_{\text{tot}}) to obtain the sigma level.
Interpret the sigma using the Gaussian‑tail table (or convert to a p‑value).

Only after a result consistently reaches the 5‑σ benchmark, survives all systematic checks, and is reproduced by independent experiments, can one claim a genuine quantum‑gravity signal.

4. Common Mistakes

Mistake	Why it’s wrong	How to avoid it
Confusing σ with p‑value – quoting “3 σ” but interpreting it as a 3 % probability.	σ is a distance in units of standard deviations; the associated p‑value is the probability of such a fluctuation (≈ 0.27 % for 3 σ, two‑sided).	Always convert σ → p if you need a probability; keep the two concepts distinct.
Neglecting systematic uncertainties or adding them linearly instead of in quadrature.	Systematics can dominate; linear addition over‑estimates the error, quadrature under‑estimates if correlated.	Build a full covariance matrix; if you must combine them, add in quadrature only when they are independent.
Ignoring the look‑elsewhere effect (multiple independent tests inflate the chance of a fluke).	Reporting a 4 σ excess found after scanning many energies/directions is less significant than 4 σ in a single pre‑defined test.	Pre‑define the test, or apply a trials‑factor correction to the σ/p‑value.
Using a Gaussian approximation for very low counts (e.g., < 10 events).	Poisson statistics are not Gaussian in the low‑count regime, leading to biased σ.	Use exact Poisson or likelihood‑ratio methods; only use Gaussian σ when the count > ~ 30.
Double‑counting data (e.g., treating the same burst’s two photons as independent measurements).	Correlated data artificially reduce σ, making the result appear more significant.	Identify correlations; if data are not independent, treat them as a single measurement or include the correlation in the covariance.
Reporting a one‑sided σ for a two‑sided test (or vice‑versa).	A 2 σ one‑sided corresponds to a 2.5 % p‑value, not 4.5 % as for a two‑sided test.	State explicitly whether the test is one‑ or two‑sided and use the appropriate conversion.

By staying vigilant about these pitfalls, you can correctly assess whether a claimed quantum‑gravity signature truly reaches the “high‑σ” (i.e., discovery‑level) threshold.

Original question: Quantum Gravity signatures high sigma? on Physics Stack Exchange, licensed CC BY-SA.