Doing 'what if' experiments is a lot easier in a simulator than it is in real life. Also you wont let out any 'magic' smoke or melt any breadboards. It can be a big help towards understanding the operation of the circuit, as long as you remember that a simulation is only as good as the models used, the parts are all ideal with no stray capacitance, lead inductance or resistance, and 0% tolerance. Some circuits dont simulate well and some that do are impractical in real life.
However you want to gain understanding of a LM317 + external pass transistor circuit under low supply conditions. That's certainly easier to simulate than to investigate in real life, unless your bench is set up with a programmable PSU, programmable load and a few bench DMMs that support datalogging, + the PC software to control it all.
I posted a LTSPICE sim of a LM317 + external pass transistor a while back.
https://www.eevblog.com/forum/beginners/variable-regulator-pass-transistor/msg1059349/#msg1059349Its a bit more complex than the one you've been experimenting with (as it uses a Sziklai pair to avoid the need for a PNP power transistor and have half the Vbe of a PNP Darlington, + has over-current protection for the pass transistor, and over-heating protection if its on the same heatsink), but once you have it up and running, it would be trivial to rip out the extra parts and substitute a PNP power transistor so you can simulate your circuit.