yeah, it's annoying that author talks about RPN notation, but only gives found formulas in form of images
looks like it computes ln(1)=0, then computes e-ln(0)=+inf, then computes e-ln(+inf)=-inf