A Natural Nuclear Reactor — Question 10

Section 43-4 · Physics pages 7230–7231

Question 10 Independent isotope signatures strengthen a hypothesis quantitatively For a hypothetical evidence exercise, let H mean ancient fission and A an alternative geological process. Priors are P(H) = .010 and P(A) = .990. A uranium anomaly has likelihood ratio 10 in favor of H; a stable-product anomaly has ratio 20. Assume conditional independence under both hypotheses. uranium signature product signature combined evidence Schematic; labels, not drawing lengths, specify the data. (a) Update prior odds with both observations and calculate the posterior probability of H. (b) Calculate the posterior using only the uranium anomaly. Why can correlated observations not automatically be multiplied?
Show solutionHide solution
Question 10 - Solution Independent isotope signatures strengthen a hypothesis quantitatively uranium signature product signature combined evidence (a) Method and calculation Oprior = P(H)/P(A) = 1/99, Opost = (1/99)(10)(20) = 200/99 . P(H | E1, E2) = 200 200 + 99 = .668896 . (b) Interpretation and further calculation P(H | E1) = 10 10 + 99 = .0917431 . P(E1, E2 | H) P(E1, E2 | A) = P(E1 | H)P(E2 | E1, H) P(E1 | A)P(E2 | E1, A) . Physical check The likelihoods are invented teaching data, not a historical confidence estimate for Oklo. Multiple signatures require attention to dependence and geological context.

Question 10

Independent isotope signatures strengthen a hypothesis quantitatively

For a hypothetical evidence exercise, let H mean ancient fission and A an alternative geological process. Priors are P(H)=.010P(H)=.010 and P(A)=.990P(A)=.990. A uranium anomaly has likelihood ratio 10 in favor of H; a stable-product anomaly has ratio 20. Assume conditional independence under both hypotheses.

Diagram shown in the original SVG worksheet.

  1. Update prior odds with both observations and calculate the posterior probability of H.

  2. Calculate the posterior using only the uranium anomaly. Why can correlated observations not automatically be multiplied?

Question 10 - Solution

Independent isotope signatures strengthen a hypothesis quantitatively

Diagram shown in the original SVG worksheet.

(a) Method and calculation

Oprior=P(H)/P(A)=1/99,Opost=(1/99)(10)(20)=200/99.O_{\mathrm{prior}}=P(H)/P(A)=1/99,\quad\boxed{O_{\mathrm{post}}=(1/99)(10)(20)=200/99}.P(H∣E1,E2)=200200+99=.668896.\boxed{P(H\mid E_1,E_2)=\frac{200}{200+99}=.668896}.

(b) Interpretation and further calculation

P(H∣E1)=1010+99=.0917431.\boxed{P(H\mid E_1)=\frac{10}{10+99}=.0917431}.P(E1,E2∣H)P(E1,E2∣A)=P(E1∣H)P(E2∣E1,H)P(E1∣A)P(E2∣E1,A).\frac{P(E_1,E_2\mid H)}{P(E_1,E_2\mid A)}=\frac{P(E_1\mid H)P(E_2\mid E_1,H)}{P(E_1\mid A)P(E_2\mid E_1,A)}.

Physical check

The likelihoods are invented teaching data, not a historical confidence estimate for Oklo. Multiple signatures require attention to dependence and geological context.