I would add to write done where you are when you are done for the day or switch to something else. I have an incredibly hard time picking up where I left off without something like this as apparently I'm able to compartmentalize more easily than some of my peers.
A trick I use is to start something "easy" towards the end of the day, but not finish it. That makes it easier to get the ball rolling the next morning.
To that end, Hemingway would stop writing for the day when it was going well, in the hope that resuming the next day would see a continuation of that ease.
That sounds like the difference between the J and P traits in the MBTI system. J-types (judgers) like me (and probably you) tend to plan ahead a lot and like to have matters settled, whereas P-types (perceptors) prefer to keep decisions open and decide how to do stuff when the time comes to act. I would guess that pmarreck is a P-type, or would at least lean towards P.
Wikipedia has more details on MBTI, and there are online self-tests for your own MBTI classification.
Disclaimer: Note the "Criticism" part of the Wikipedia page. MBTI is AFAIK not a tool that professional psychologists would use. Nonetheless, I find it useful in order to find similar people on the internet. For example, if I want to learn a new skill, a search query like "[thing] tutorial for INTJ" gives me way better results than just "[thing] tutorial" because I get recommendations from people who think in similar ways.
Basically I am able to (mentally and physically) walk away from work and let go of it and go home and game or work out or whatever. Then, the next day when I sit down again, I get my bearings and see where I'm at and then proceed!
There are a few exceptions to this, of course, but it's always with hard problems I become absolutely OBSESSED about. For example I couldn't find an XML parser in Elixir a couple weeks ago that I liked, so I was faced with writing my own... which I did... and then released as open source https://github.com/pmarreck/mega_xml but I was literally up until 4:30AM one night figuring out how to turn an event-based XML parser callback mechanism into an Elixir Map datastructure (something I had never had to do before, but which I somehow knew I was barely capable of figuring out, lol). It was extra tricky (for me) due to the lack of mutability
