a problem: restart run could not repeat previous run, with the CESM 1.0.3

[mai]

Let me make two suggestions:1. Read the paper What Every Computer Scientist Should Know About Floating-Point Arithmetic 2. Try two 30-day runs changing only the least significant bit in a single input number; save history daily; plot some fields on the last day of the month What you will see is what you have already seen in your two simulations when you changed only the PE layout: The smallest difference propagates through the system in about three weeks until the output fields become as different as they will ever be.