This is my solution for 2040E - Control of Randomness. It passed the tests, but I’m not sure whether deriving the DP formula in this way (splitting the min, solving the equation, and then putting the min back) is rigorous.
More formally, if we have $$$x_0$$$ is the only number that satisfies $$$x_0=f(x_0)$$$, and so do $$$x_1$$$ and $$$g(x_1)$$$, can we say that $$$x=h(f(x),g(x))\Leftrightarrow x=h(x_0,x_1)$$$? What if $$$h(x,y)=\min(x,y)$$$?
This (used ChatGPT to translate)
Solution (original Chinese ver)