ProbXiv
Explorersign in
Problem archiveProblem record

Statement

Let N=2N=2 and suppose the conditional densities of the observation YY given hypotheses H=0H=0 and H=1H=1 are

fY∣H(y∣0)=12πσ2exp⁡(−y22σ2),f_{Y\mid H}(y\mid 0)=\frac{1}{\sqrt{2\pi\sigma^2}}\exp\left(-\frac{y^2}{2\sigma^2}\right), fY∣H(y∣1)=12πσ2exp⁡(−(y−h1)22σ2),f_{Y\mid H}(y\mid 1)=\frac{1}{\sqrt{2\pi\sigma^2}}\exp\left(-\frac{(y-h_1)^2}{2\sigma^2}\right),

where σ\sigma and h1h_1 are arbitrary positive numbers, and let λ\lambda be a real number. The Gaussian integral inequality. If λ<h12\lambda<\frac{h_1}{2}, then

∫−∞λexp⁡(−y22+λh1) dy∫λ∞exp⁡(−y22+λh1) dy<∫−∞λexp⁡(−y22+yh1) dy∫λ∞exp⁡(−y22+yh1) dy.\int_{-\infty}^{\lambda}\exp\left(-\frac{y^2}{2}+\lambda h_1\right)\,dy\int_{\lambda}^{\infty}\exp\left(-\frac{y^2}{2}+\lambda h_1\right)\,dy < \int_{-\infty}^{\lambda}\exp\left(-\frac{y^2}{2}+y h_1\right)\,dy\int_{\lambda}^{\infty}\exp\left(-\frac{y^2}{2}+y h_1\right)\,dy.

This inequality is used in the analysis of the prior belief that minimizes the second agent's Bayes risk in a two-agent sequential decision problem with additive Gaussian observation noise. The surrounding discussion indicates that optimal prior beliefs can differ from the true prior probability, but the supplied text does not establish whether this asserted inequality is proved or remains open.

Record

Comments

Nothing has been published against this problem yet, and nothing has been said about it.

Sign in with an institutional address to take part in the discussion. Reading every thread stays open to everyone.

Sign in

Solve with an agent

Open the statement in a chat, with the problem and the ground rules already written into the prompt.

This opens a third-party site. Nothing is posted back to ProbXiv and nothing you write there is recorded here — what a model gives you is an attempt, which a person still has to check.