It’s hard to say which mathematical concepts occur in reality; imaginary numbers are critical in describing quantum particles, while even integers are ultimately abstractions. But I do think you have a point there. Pure mathematics is all about assuming particular systems and studying them.
It’s usually most useful to work with axioms inspired by our observed reality, which is where we get our numbers and ordinary Euclidean geometry from. But you can make up ones that seem more elegant, more unusual, more illustrative of some principle. There are other number systems and geometries, and more abstract things that are not so clearly either, and all sorts of constructs that are easily thought of as invented.
But if axioms are your choice, their consequences are not. They may inescapably explode with self-contradiction, or only have simple results, or result in a large body of non-intuitive theorems. People started looking at defining square roots for negative numbers, and then iterating functions over them, without ever expecting the Mandelbrot set. I think it’s fair to say these kind of things are easily thought of as discovered.
So I think it would be better to say something between the two – math is discovering what abstractions can be invented, or that kind of thing.
I wanted to offer my intuition about it, because I’m curious if it actually helps anyone else. The function y = et is unusual in that it is its own derivative, and sometimes that is even taken as part of its definition. More general exponential functions can be written as y = eat, and then we get dy/dt = ay.
In other words, they represent functions where the rate of change is some constant multiple a of the current position. It’s often handy to think of that as a speed, so it’s like an ever-increasing repulsion from zero. Or rather it is when a is positive. When a is negative, it falls asymptotically into zero instead.
And for imaginary a we do something between those. We can look at this in terms of the complex plane. Then multiplying by a positive number scales everything around zero, and a negative number inverts it, but an imaginary number gives a rotation. In particular multiplying by i means turning 1 into i, which we get by rotating everything 90º to the left.
If you appreciate that you can maybe see how dy/dt = iy will act in the plane. The function is not going to travel outward or inward, but instead the rate of change is always going to be perpendicular to the current position. In other words, we are going to be going in a circle. Since y = eit starts from 1, we go to i then -1 then -i and then back to 1.
And since our absolute distance from zero stays 1, the magnitude of our rate will stay 1 as well – we’re circling with constant speed. At which point it’s not so surprising that a full circle needs t = 2π. Or, if you want to travel to -1, you only go half the circumference; we plug in t = π and so end up at eit = -1.
So writing that out, I see it relies on a lot of other concepts to try to form a picture, though for me it gives a more satisfying one than just a numerical proof. Does any of it work for you?