Probability Density Functions and Distributions

General Overview

• The lecture begins with an inquiry to the class about clarity and understanding of the previous material.
• Introduction of comments about the homework assignment related to probability density functions (PDFs).

Probability Density Function (PDF)

• Definition: A probability density function describes the likelihood of a random variable falling within a particular range of values, as opposed to taking on specific values.
• Integration of PDFs: Probabilities are calculated using integrals involving the PDF; it is essential to understand that the probability density function itself is not probability.
• Normalization Requirement: When constructing a PDF from experimental data, it must be normalized so that the area under the PDF curve equals one.
  • Normalization methods include:
  • Using histogram bins from the data.
  • Employing numerical integration techniques (e.g., trapezoidal rule).
  • If the integral yields a value greater than one after normalization, a reevaluation of the normalization is necessary.

Gaussian Probability Density Function

• The Gaussian PDF, a specific type of probability distribution, is described with its functional form:
• Formula:$p(x) = \frac{1}{\sqrt{2\pi \sigma^2}} e^{-\frac{(x - \mu)^2}{2\sigma^2}}$
  • Where:
  • $\mu$ is the mean.
  • $\sigma$ is the standard deviation.
• Sample Gaussian PDFs: Demo various forms indicating differences in mean and standard deviation effects on the curve.
• Normalization and Mean: The Gaussian distribution can also be normalized to have a mean of zero, allowing for comparative analysis.
• Characterization: The Gaussian distribution is symmetric concerning the mean.

Expected Value and Higher Moments

• Expected Value (Mean):
  • Defined as:
    $E[X] = \int_{-\infty}^{\infty} x p(x) \, dx$
• Higher Moments: Variance and higher moments measured as:
  • $\int_{-\infty}^{\infty} (x - \mu)^n p(x) \, dx$
  • For odd moments, notably, they equal zero, evidencing symmetry.

Cumulative Density Function (CDF)

• The cumulative PDF is defined using integration:
  $F(x) = \int_{-\infty}^{x} p(t) \, dt$
• Importance: Allows the calculation of the probability of the variable being less than a certain value

Standard Normal Distribution

• Standardization: A variable $z$ can be defined, with mean zero and standard deviation of one:
  $z = \frac{x - \mu}{\sigma}$
• Standard Normal PDF: For normalized function:
  $p(z) = \frac{1}{\sqrt{2\pi}} e^{-\frac{z^2}{2}}$
• Utilization of Normal Tables: For probabilities concerning a normalized Gaussian:
  • Proportions of data within specific standard deviation ranges:
  • 68.27% within $\mu \pm 1\sigma$.
  • Proportions get progressively smaller as you move away from the mean.

Example Problem

• Objective to find the z-range encompassing 68.27% of data yields z equals +1 and -1:
• Consequently, the standardized range of values $(x)$ can be calculated:
  \mu - \sigma < x < \mu + \sigma
• Questions regarding the understanding of the relevant tables are addressed, detailing column values and their significance.

Assessment of Departure from Gaussian Distribution

• Excess and Kurtosis: Critical values can be derived from experimental data to assess Gaussianity using:
  $\delta = \frac{E[X^4] - 3\sigma^4}{\sigma^4}$
• Skewness Assessment: Procedures to statistically analyze the data's distribution.

Time Resolved Data Statistics

• Approach on analyzing data measured in time intervals.
• Derivation of mean and variance based on discretized measurements, leading to an accumulated understanding of temporal data behavior.

Multi-Variable Probability Density Functions

• Introduction to statistical measures in multi-dimensional spaces,
  • Joint Probability PDF: For two variables $X<em>1$ and $Y</em>1$, the PDF is defined analogous to single-variable PDF:
  • Integral properties to determine multidimensional probabilities encompass volumes instead of areas, similar concepts apply to covariance and variance.

Additional Distributions: Chi-Squared, Student's T, and F Distribution

• Further introduction to essential sampling distributions:
  • Chi-Squared Distribution: $\chi^2$ used for hypothesis testing, defined through degrees of freedom:
    $\chi^2 \sim Gamma(\frac{\nu}{2}, 2)$
  • Student's T Distribution: Represents estimates of means when variance is unknown, approaches normality with higher degrees of freedom:
    $P(t) = \frac{Γ(\frac{\nu + 1}{2})}{√{\nu \pi} Γ(\frac{\nu}{2}) (1 + \frac{t^2}{\nu})^{\frac{\nu + 1}{2}}}$
  • F Distribution: Relates two variances from samples, noted for degrees of freedom:
    $P(F) = \frac{Γ(\frac{\nu<em>1 + \nu</em>2}{2})}{Γ(\frac{\nu<em>1}{2})Γ(\frac{\nu</em>2}{2})}F^{\frac{\nu<em>1}{2}-1} (1 + \frac{F}{\nu</em>2})^{-(\frac{<br />\nu<em>1 + \nu</em>2}{2})}$

Conclusion

• Importance of understanding both single and multidimensional statistical distributions.
• Encouragement to practice utilization of tables and normalization in statistical analysis.
• Issues of constraints and degrees of freedom framed as key considerations in probability modeling.