Someone emailed me recently asking if it's really possible to paint using a Cyan/Magenta/Yellow (CMY), or "process colour" model. He's run into skepticism from working artists, and I'm not surprised. So, let's start with the most important thing. The late great Douglas Adams once said
"Anything that’s invented after you’re 35 is against the natural order of things."
Artists couldn't paint with CMY until we had a stable, lightfast M to work with. I didn't see one on the market until the 1990s. So there aren't many practising painters who trained with M paint. We're practical people, artists, and when it comes to making work, we're surprisingly conservative. We use the materials we know. We want our paintings to last for centuries and we generally use only the tried and true materials with which we trained. Hey, we're experts at how to use two different reds, one for orange and one for purple, so that's what we do. As the saying goes, if it ain't broke, don't fix it. My father drives stick instead of automatic, and I send email instead of text messages. We'd both argue we've chosen the superior technology but you know it's also because we got good at it before we turned 35 and don't see any reason to change.
If you want to persuade a painter, you'll have the most luck with watercolourists, who are fond of a colour called Opera from Holbein, which is a very useful magenta that some people would call pink. It's not quite lightfast but it's still a favourite.
As for me, I trained in the classic RYB model and my thinking is still rooted there, but I have enough science education to be pretty comfortable talking about the relationship between additive and subtractive colour, the RYB model, the CMY model, and the RGB model we need to describe light.
The CMY colour model is about light, but paint relies on pigment. Pigments don't arise on the planet in optically pure colours conveniently corresponding to wavelengths of visible light. There are single-pigment paints that are close to being optically pure primaries (for magentas, you'll want to look at colours with quinacridone), but nothing perfect. There are manufactured paints that are damned close, usually gouache meant for (ephemeral) use in graphic design, but they won't be single pigment.
Here's a kickass web page that assesses a whack of magenta pigments for lightfastness. It's not a casual read (but then, upon reflection, neither is this blog entry) but it looks rock solid and matches all of my own experience with pigments like quinacridone. Looking around the site, Bruce Macavoy does an outstanding job on colour theory. Not for beginners, perhaps, but check it out.
If you want to learn to paint in a CMY model, I recommend the System 3 line of acrylics from Daler Rowney, which includes CM and Y (and sells a handy five-colour Process Set for just this purpose). It's a student grade rather than high end. It's reasonably but not thoroughly permanent. But the colours are very good primaries.
For your first exercise, take a small amount of Y and mix dabs of M into it, very gradually, painting samples onto a clean page, until you get to red. It's a mind-blowing experience and you'll want to make everyone you know try the same experiment. Especially the painters.