I reckon there is value in only partially simulating the environment and grow lots of potatoes at scale. Take those potatoes that succeed and breed them and repeat. With every iteration, change the environmental conditions to better simulate Mars. Over successive generations, I reckon we can get to potatoes (and other plants) that fair well.

The big question is how you optimize this process to assist evolution in moving faster.

Given an inexpensive transparent material with very high tensile strength and resistance to weathering in the Martian environment, I could see a passive pressurization/thermal regulation scheme for greenhouses on the surface of Mars that relies only on, say, a heat source, some valves, and some consumable gasses. I could imagine such schemes using mainly water as an input.

That said, such a material would be asking for a lot. Then, there's also the UV and other radiation to consider.