This is not going to be easy. 1. Undergrad courses are definitely helpful but they are baby stuff compared to research level study. 2. Taking relevant grad co

This is not going to be easy. 1. Undergrad courses are definitely helpful but they are baby stuff compared to research level study. 2. Taking relevant grad courses on OCW might be helpful but less helpful than you think. For instance, my phd is in theoretical quantum optics. And I took a quantum optics course in grad school. I don't think I ever used anything I learned in that course in my research. Physics grad courses are often precisely designed to work in that prof's group, and not useful otherwise. 3. But you will have to take those courses anyway to up your base knowledge level and mathematical skills. 4. Read papers and work backwards and learn the stuff mentioned in the background sections. Do this for 6 months for few dozen papers. At which point you might to start to understand roughly what one paper has to say. 5. Open problems are often mentioned in papers and talks by field leaders - because they want others to build on their work. 6. But honestly this is not going to be easy. What you are trying to do is working at the home gym alone - except no one would ever see your muscles. PhD students working in labs feel lost. I can't imagine doing this on your own.