Signal Injection Test: What, How, & Expectations

  • Context: Undergrad 
  • Thread starter Thread starter Alkass
  • Start date Start date
  • Tags Tags
    Injection Signal Test
Click For Summary

Discussion Overview

The discussion revolves around the concept of a "signal injection test" in the context of particle physics, particularly related to Higgs boson studies at CERN. Participants explore the methodology of conducting such tests, the implications of the results, and the expectations regarding statistical significance and hypothesis testing.

Discussion Character

  • Exploratory
  • Technical explanation
  • Debate/contested

Main Points Raised

  • One participant inquires about the nature and execution of a signal injection test, referencing its application in Higgs studies.
  • Another participant explains that a signal injection test involves simulating data without a Higgs signal and then adding a Higgs signal to observe changes in measurements.
  • A follow-up question addresses the implications of observing or not observing a difference in measurements, particularly regarding the statistical significance of the null hypothesis and the potential for a biased search.
  • A participant responds that if no difference is observed, it allows for setting upper limits on signal strength and discusses the interpretation of results in terms of confidence levels and the implications for Standard Model predictions.
  • There is a mention of colorful plots that represent the results of such tests, with a specific example of a point at 150 GeV related to the Standard Model signal strength.
  • Concerns are raised about the validity of exclusion probabilities and the need for modifications to account for search sensitivity.

Areas of Agreement / Disagreement

The discussion contains multiple competing views regarding the interpretation of results from signal injection tests and the implications for hypothesis testing. There is no consensus on the best approach to quantify differences or the validity of certain statistical interpretations.

Contextual Notes

Participants express uncertainty about the implications of observed differences, the definitions of hypotheses involved, and the statistical methods used to interpret results. Limitations regarding the sensitivity of searches and the potential for biased conclusions are also noted.

Alkass
Messages
19
Reaction score
0
hi

I would like to ask, what is all about a "signal injection test" ? How is it done and what one expects to see ? I ve read about it in some Higgs study/plots CERN has published

thanks

Alex
 
Physics news on Phys.org
You can simulate what you expect without a Higgs, and then you can add a Higgs signal and see how the measurements change - is it possible to see the difference? As far as I know, "signal injection" is just that.
The same is possible for all other particle searches, of course.
 
Ok, I get it - So what happens if you see the difference or not and how you quantify it ? Does it change the statistical significance of the null hypothesis in the case you can "see" a differnece? Or one should think about it, like a "biased" search, thus you should consider an H1 hypothesis in such a case instead of a null one ? At the end, why one would want to do such a test ?

thanks
-a
 
So what happens if you see the difference or not and how you quantify it ?
If you do not see a difference between data and "simulation without signal", you can set an upper limit on the signal strength. If a simulated signal with some specific signal strength would lead to a significant deviation (usually: 10% or 5% probability that a downwards fluctuation gives fewer events), this signal strength is excluded at 90% (95%) confidence level.

That leads to those colorful plots like this one.

Consider a point at 150 GeV (just at the left edge of "LHC excluded"), for example, it is close to 1, and 1 is exactly the Standard Model signal strength. The point has the message "if there is a Standard Model Higgs at 150 GeV, we should have seen more events with a probability of 95%".

Actually, this is not true. Following my description, every point would have a probability of 5% to exclude everything, even if there is a Higgs and the data is not sufficient to see it yet. Therefore, the method is modified a bit to include the sensitivity of the search, but I think this is an irrelevant detail here.[/size]
 

Similar threads

Undergrad ATLAS and CMS see signs of Higgs->muons decay
  • · Replies 17 ·
Replies
17
Views
6K
Graduate Signal strength parameter (interpretation)
  • · Replies 5 ·
Replies
5
Views
4K
Undergrad The purpose of trigonometric axes in plot of electron clusters, ATLAS
  • · Replies 1 ·
Replies
1
Views
4K
Undergrad Queries on the operation of the LHC with pilot beams before Run 3
  • · Replies 1 ·
Replies
1
Views
2K
Undergrad Concentrated arcs in bubble chamber photos
  • · Replies 2 ·
Replies
2
Views
2K
Graduate Signal signficancies low expected events
  • · Replies 4 ·
Replies
4
Views
2K
Is It Safe to Inject a COVID Booster Before the Alcohol Prep Pad Dries?
  • · Replies 18 ·
Replies
18
Views
3K
CE Mark Surge Testing: Protecting Wireless Products
  • · Replies 5 ·
Replies
5
Views
2K
Graduate Confused about reflected signal in a bow-tie optical cavity
  • · Replies 5 ·
Replies
5
Views
4K
Discovering a Total Injection from R to [0,1]
  • · Replies 13 ·
Replies
13
Views
3K