3
$\begingroup$

Question:

What is appropriate notation for calculating the mode of a function? In this case, I use the mode of a Gamma(a,b) distribution in a cost-function.

I am looking for something analogous to $\bar{x}$ or $\tilde{x}$ notation that is commonly used for the mean and medians, respectively, if something like $\bar{\text{G}(a,b)}$ would make sense.

What I have tried:

  1. Google results are confounded by mode $\simeq$ model and results for latex's 'math mode'

  2. It is not on the following wikipedia pages:

  • 0
    For 1: [tried this](http://www.google.com/search?q=%22mode%22+%22gamma%2Bdistribution%22)?2011-05-06
  • 0
    @J.M. that gives the analytical form of the mode, but that is not what I want. I have changed my question to focus on notation for mode rather than calculation of the mode of the gamma.2011-05-06
  • 1
    I was addressing the "Google results are confounded by" portion in #1; that is, use quote marks to tell Google that you meant what you said.2011-05-06
  • 0
    @J.M. Thanks for the hint. But I am still unsuccessful.2011-05-06

2 Answers 2

3

One possibility for the notation for a mode is $$\underset{x}{\operatorname{arg\,max}} f(x)$$ but I thought that was ugly, so here I used a hat $$\hat{X}$$ to indicate the peak in the probability density or mass function, even though this can also be used for other meanings.

1

In Bayesian estimation, the mode is given by the maximum a posteriori (MAP) estimate. Using the "MAP" subscript would suffice.

  • 0
    .. except that I am not using a posterior, I am using data to set the mode of an informed Gamma prior.2011-05-06
  • 0
    @David: I would just pick a symbol and explain it.2011-05-06
  • 0
    thanks ... right now I am using mode(G(a,b)); perhaps I could use $u()$ but I didn't want to reinvent the wheel2011-05-06