Friday, March 1, 2013

MIstakes happen

I was disappointed to find a mistake in one of the figures of my recent paper on hydrogen bonding. Fortunately, the mistake has no implications for the results in the paper. I just made a basic mistake when using Mathematica(!) to produce the figure. (Rather ironic and noteworthy given recent discussions on this blog about the dangers of Mathematica).

The upper two plots in Figure 2 of the paper should be replaced with those below.

The mistake was kindly pointed out to me by Sai Ramesh and Bijyalaxmi Athokpam during my recent visit to Bangalore. They found they could not reproduce the figure and were wondering why.

In hindsight, it is obvious that there is something wrong with the original plots. Consider the energy gap between the two adiabatic curves at the co-ordinate at which the diabatic curves cross. This energy is 2 times Delta, the off-diagonal matrix element which couples the two diabatic states. A major idea/assumption in the paper is that Delta decreases exponentially with increasing R, the donor-acceptor distance. However, in the original figure this gap does not vary with R! The new curves above clearly show this trend.

It is amazing to me that this basic point was missed by me, and by several colleagues and referees who read the paper. But I must take full responsibility.

What should be learn from this?

1. It does not matter how many times we check something, we can still make basic mistakes. The easiest person to fool is yourself.
Co-authors need to be particularly diligent in checking each others work.
A fresh set of eyes may reveal problems.
Trying to reproduce results from scratch is often a good check.

2. Just because something is published does not mean it is correct!
When we find something in a published paper that is not quite right, we should not assume it is correct.

Mistakes happen.

No comments:

Post a Comment