I thought of an easier way to explain this. An outward push by light is basically negating gravity. The light's intensity even even falls off as in inverse square, just like gravity. So for the same reason that the inward pull of gravity does not cause an inward spiral, the outward push of light does not cause an outward spiral.
One subtracts from the other, and basically leaves you with what acts like a less massive star.
-------------------------------------------------------------------------
In my examples from the previous posts, with an object orbiting at 0.1 AU, and accelerating away from the sun at 1 mm/s, it is no different than simply subtracting M=ar
2/G = 0.001 * 14959787340
2/6.672e-11= 3.354e+27 kg from the sun's mass of 1.989e30 kg, giving the sun a new mass of 1.986e30kg. Instantly subtracting this from the mass of the sun should produce a blip in the orbits of everything circling the sun. But... once subtracted, everything should immediately make a one-time jump to their new orbits, and there should be no additional fluctuations in sma and ecc.
So why does my graph show fluctuations? 2 reasons. One is that any numerical simulation is only an approximation, and with Excel's auto ranging graphs, it can zoom in to show the numerical artifacts. See the new graph for an illustration. But the biggest reason is that in my simulations, the acceleration is constant. It does not fall off as an inverse square.
The first graph shows the error produced by the numerical simulation. Throughout the course of this graph, the time step is increased from 1 to 2 to 4 seconds, The test particle was originally orbiting a full-mass sun, and then mass was subtracted off the sun to lessen the acceleration produced by the sun's gravity by 1 mm/s
2. This transition is not shown on this graph. So sma and ecc should not fluctuate at all. The fact that they do is from numerical simulation error. But as you can see, it is very small, well less than 1 km of sma at time step =1.
The next graph shows the transition jump at about 50 days, when the sun magically loses 3.354E+27 kilograms. The eccentricity flucuations are barely noticable at a time step of 1 second, and increase as the time step is raised to 2 and then to 4 seconds. The sma does as well, but it is not visible at the resolution of this graph.
