Tom,
To continue the dialogue on the modeling of two kinetic minerals, I have been successful in modeling glass dissolving and a mineral phase precipitating. This is not totally satisfactory as I am unable to prevent dissolution even of a ng of mineral until saturation. However, for the time being, I am willing to live with that. However, as the glass dissolves and saturation of the second mineral is approached, the code begins to decrease the step size until it can converge. Another several small steps (5 or 6 at 0.00001 or so) are calculated before convergence is a problem again. This continues and, most of the time, the calculation completes. It often helps to relax the value of epsilon by 3 to 5 orders of magnetude, maybe as low as 10^7. Even with this, the calculation proceeds very slowly, often taking several hours to complete. In the output, I can often see spikes in the reaction rate and other output that suggests residual instability in the calculation. I am a bit concerned that the calculation is accurate.
In an email, I'll send you the script and the database. When you get the time to take a squint at it, I'd appreciate your feed back, especially if you have a suggestion on how to make the code run better.
Denis