It would be odd if we had come so far in academia without being passionate about science. Unfortunately, this means that after a drink or two some of us (i.e. me) find it difficult to keep from talking physics and data analysis. While I'm working to avoid random outbursts of science, especially in mixed company, I can't do it without at least one cocktail memorializing my some aspects of my research.
How do you make the optimal cocktail? One approach would be to find your single favorite recipe, the selection and ratio of ingredients, and stop there. Another approach goes further, considering not just your favorite recipe but also perturbations away from that recipe. You may prefer a 6:1 martini, but how much more do you prefer it to the 5:1 or the 7:1? Or the 1:1 for that matter? The second approach would advocate serving an infinite number of cocktails, the amount of each serving weighted by the preference for that ratio. A similar approach would be taken towards ingredients.
This approach is not as foreign as it might immediately appear. Stripping a cocktail down to a single ingredient, someone with no previous experience would try the same amount of each different variety. But this is just the flight commonly used to taste spirits, beer, and wine.
In practice, preparing more than one cocktail is often prohibitive. When trying to optimize a recipe, however, it may not be a bad idea to compare the different possibilities directly. Consider the choice of bitters in the Monte Carlo,
Markov Chain Monte Carlo
In four shot glasses prepare
1 oz rye whiskey
0.5 oz benedictine
Add to individual glasses
1 dashes Angostura bitters
1 dashes Whisky Barrel Aged bitters
1 dashes Peychauds bitters
1 dashes Xocolatl Mole bitters
Now let's see how long I can go without talking about work...