Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Is there some way of describing the co-domain of probability density functions? Does it relate in some way to something physically meaningful? I was given that question today - and I was at a loss. Density for me, is the co-domain of pdfs - a scalar dimension with values from zero to infinity.

For instance, I fit a normal distribution to a probability vector on N values (a probability mass function). This vector contains normalised values of the actual frequency counts in a histogram with N bins. The normalisation is done by dividing each frequency count by the discrete integral "area" of the pmf - so that the pmf values sum to 1. The pmfs now seems to be scaled similarly to the estimated pdfs.

Is probability density a measure of probability? I am sure that the necessary definitions must be hidden somewhere deep down in the guts of measure theory - which is why I have included that as a tag.

share|cite|improve this question
Just a note. The relevant object is really the measure $p(x)\,dx$ (that you integrate with respect to), not the pdf $p(x)$ itself. For example, under a change of variable $y(x)$, $p(x)$ can be changed almost arbitrarily (multiplied by $1/\frac{dy}{dx}$ which can be almost any function you want), but $p(x)dx=p(y)dy$ stays the same. – Kirill Jul 2 '13 at 22:32
The units of the codomain of the probability density function must be reciprocal of the units of the domain so that upon integration, we obtain the unitless probability. – Sammy Black Jul 2 '13 at 22:35
up vote 1 down vote accepted

To add to Kirill's comment, you can write $P_0 = P(x_0\leq X \leq x_0+{\rm d}x) = p(x_0)dx$, so the value of the pdf $p(x)$ at $x_0$ is the height of an infinitesimally narrow interval ('slice') of width d$x$ fitting below this function at $x_0$ that will give you a probablity value $P_0$ to find the random variable lying within this interval d$x$. You could also think of it, perhaps more intuitively, in terms of the derivative of the cdf: the pdf at $x_0$ is then the 'speed' or rate, d$P/$d$x$, at which the probability will increase when going from $x_0$ to $x_0+dx$ (for positive d$x$). A "physical" meaning is more difficult, because probability is mathematical... Perhaps if you think of the analogy of "mass density" (~pdf) as a measure of concentration of "mass" (~cdf), that may help?

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.