I know that this is over 2 years too late, but 0.69314... is ln2, and 1.79175... is ln6 - just take the natural log of whatever to find the A given DesiredOutputPower.
For the other function, 1.15129... is ln(sqrt10), and 6.90775... is ln(1000). Notice that 1000 is equal to sqrt(1000000), which is equal to 10^6. Therefore, the general function for the upper limit of g is ln(sqrt(10^(MPG/10))).
I don't actually know what an AGC is or whether you're still interested in it. Hopefully this can be helpful.
