There are a lot of poker solvers out there approximating the Nash Equilibrium (NE), probably all of them using CFR or deviations of it.
However, they are also giving you the "target proximity to a global optimum" as a parameter.
So you can choose: 0.5% (of the pot size) - not so much; or 0.1% - still not much, but it takes some time - to find a solution that is apparently 0.1% close to the (or one) global optimum.
How is this approximation distance between approximated solution and real solution estimated, leave alone calculated?
I am coming from a classic numerical background: And there you had some algorithms to find a solution. But never has there been a fairy who told me how close to the real solution I was.
I tried to reach the Nash Equilibrium with a Poker Solver which uses Counterfactual Regret Minimization. I was expecting an error, but not an error in the sense that the algorithmic solution differs from the perfect solution by THIS margin, because nobody knows the perfect solution.