Infer decay while preserving background and unused records

State the event and measurement model

Independent unstable nuclei have a statistical survival law. An exponential expectation does not specify which individual nucleus decays next or promise that an actual sample contains exactly half its nuclei after one half-life. Activity is λN in decays per time, whereas a detector rate additionally depends on efficiency, branching and geometry. The synthetic source contribution here begins at 640 counts/min; that rate alone does not specify an initial number of nuclei.

Reconcile the quantitative evidence

The indicated expectation is R(t)=S 2^(−t/T)+B with unknown amplitudeS, half-lifeT and constant backgroundB. At equally spaced times 0,4,8 min write R0=S+B,R1=Sr+B,R2=Sr²+B. Then r=(R1−R2)/(R0−R1), T=−4 ln2/lnr and B=(R1−rR0)/(1−r). The three records are all calibration inputs. A fourth at 12 min must remain unused until the prediction is frozen; reusing the8 min point would falsely label fitting as validation.

Preserve limits and alternative explanations

Raw-rate ratios do not cancel an additive background and can suggest a source is decaying more slowly than it is. Real noisy differences can be small or even give an inadmissible ratio, which is a reason to revisit uncertainty or model assumptions, not silently clip the answer. A mixture of nuclides, a changing baseline or detector dead time can invalidate a single exponential. A separate dating inference also requires an initial composition or another justified reference; a half-life formula alone does not establish the age of an unknown object.

Worked original case

For T=4 min and B=20 counts/min, indications at 0,4,8 min are660,340,180 counts/min. Difference ratio160/320=.5 recoversT4 andB20. The unused12 min prediction is100 counts/min, made of80 source plus20 background.

Investigation

Fit first three expectation records without reading the hidden half-life or baseline, then freeze the12 min prediction. Compare a raw-ratio estimate with the difference method. State what independent efficiency or source-composition evidence would be needed to infer activity or age.

Changed-case practice

Sources and boundaries

Original cases checked 31 October 2026 against the primary explanation. These are bounded virtual models, not instructions to illuminate, accelerate, collect or expose real material. No real laboratory participation, detector certification, personal medical conclusion or independent mastery is claimed. Keep first work, critique, revision and genuinely new assigned evidence separately from public practice.