This article includes a list ofgeneral references, butit lacks sufficient correspondinginline citations. Please help toimprove this article byintroducing more precise citations.(June 2010) (Learn how and when to remove this message)

Probability density function Probability density function for the truncated normal distribution for different sets of parameters. In all cases,a = −10 andb = 10. For the black:μ = −8,σ = 2; blue:μ = 0,σ = 2; red:μ = 9,σ = 10; orange:μ = 0,σ = 10.
Cumulative distribution function Cumulative distribution function for the truncated normal distribution for different sets of parameters. In all cases,a = −10 andb = 10. For the black:μ = −8,σ = 2; blue:μ = 0,σ = 2; red:μ = 9,σ = 10; orange:μ = 0,σ = 10.
Notation	${\displaystyle \xi =\frac{x-\mu }{\sigma },\alpha =\frac{a-\mu }{\sigma },\beta =\frac{b-\mu }{\sigma }}$ ${\displaystyle Z=\mathrm{\Phi }(\beta )-\mathrm{\Phi }(\alpha )}$
Parameters	${\displaystyle \mu \in \mathbb{R}}$ ${\displaystyle {\sigma }^2\ge 0}$ (but see definition) ${\displaystyle a\in \mathbb{R}}$ — minimum value of${\displaystyle x}$ ${\displaystyle b\in \mathbb{R}}$ — maximum value of${\displaystyle x}$ (${\displaystyle b>a}$)
Support	${\displaystyle x\in [a,b]}$
PDF	${\displaystyle f(x;\mu ,\sigma ,a,b)=\frac{\phi (\xi )}{\sigma Z}}$[1]
CDF	${\displaystyle F(x;\mu ,\sigma ,a,b)=\frac{\mathrm{\Phi }(\xi )-\mathrm{\Phi }(\alpha )}{Z}}$
Mean	${\displaystyle \mu +\frac{\phi (\alpha )-\phi (\beta )}{Z}\sigma }$
Median	${\displaystyle \mu +{\mathrm{\Phi }}^{-1}\left(\frac{\mathrm{\Phi }(\alpha )+\mathrm{\Phi }(\beta )}{2}\right)\sigma }$
Mode
Variance	${\displaystyle {\sigma }^2\left[1-\frac{\beta \phi (\beta )-\alpha \phi (\alpha )}{Z}-{\left(\frac{\phi (\alpha )-\phi (\beta )}{Z}\right)}^2\right]}$
Entropy	${\displaystyle \ln (\sqrt{2\pi e}\sigma Z)+\frac{\alpha \phi (\alpha )-\beta \phi (\beta )}{2Z}}$
MGF	${\displaystyle e^{\mu t+{\sigma }^2{t}^2/2}\left[\frac{\mathrm{\Phi }(\beta -\sigma t)-\mathrm{\Phi }(\alpha -\sigma t)}{\mathrm{\Phi }(\beta )-\mathrm{\Phi }(\alpha )}\right]}$

In probability and statistics, thetruncated normal distribution is the probability distribution derived from that of anormally distributed random variable by bounding the random variable from either below or above (or both). Thetruncated normal distribution has wide applications in statistics andeconometrics.

Suppose${\displaystyle X}$ has a normal distribution with mean${\displaystyle \mu }$ and variance${\displaystyle {\sigma }^2}$ and lies within the interval. Then${\displaystyle X}$ conditional on has a truncated normal distribution.

Itsprobability density function,${\displaystyle f}$, for${\displaystyle a\le x\le b}$, is given by

$${\displaystyle f(x;\mu ,\sigma ,a,b)=\frac{1}{\sigma }\frac{\phi (\frac{x-\mu }{\sigma })}{\mathrm{\Phi }(\frac{b-\mu }{\sigma })-\mathrm{\Phi }(\frac{a-\mu }{\sigma })}}$$

and by${\displaystyle f=0}$ otherwise.

Here,$${\displaystyle \phi (\xi )=\frac{1}{\sqrt{2\pi }}\exp \left(-\frac{1}{2}{\xi }^2\right)}$$is the probability density function of thestandard normal distribution and${\displaystyle \mathrm{\Phi }(\cdot )}$ is itscumulative distribution function$${\displaystyle \mathrm{\Phi }(x)=\frac{1}{2}\left(1+\mathrm{erf}(x/\sqrt{2})\right).}$$ By definition, if${\displaystyle b=\mathrm{\infty }}$, then${\displaystyle \mathrm{\Phi }\left(\frac{b-\mu }{\sigma }\right)=1}$, and similarly, if${\displaystyle a=-\mathrm{\infty }}$, then${\displaystyle \mathrm{\Phi }\left(\frac{a-\mu }{\sigma }\right)=0}$.

The above formulae show that when the scale parameter${\displaystyle {\sigma }^2}$ of the truncated normal distribution is allowed to assume negative values. The parameter${\displaystyle \sigma }$ is in this case imaginary, but the function${\displaystyle f}$ is nevertheless real, positive, and normalizable. The scale parameter${\displaystyle {\sigma }^2}$ of the untruncated normal distribution must be positive because the distribution would not be normalizable otherwise. The doubly truncated normal distribution, on the other hand, can in principle have a negative scale parameter (which is different from the variance, see summary formulae), because no such integrability problems arise on a bounded domain. In this case the distribution cannot be interpreted as an untruncated normal conditional on, of course, but can still be interpreted as amaximum-entropy distribution with first and second moments as constraints, and has an additional peculiar feature: it presentstwo local maxima instead of one, located at${\displaystyle x=a}$ and${\displaystyle x=b}$.

The truncated normal is one of two possiblemaximum entropy probability distributions for a fixed mean and variance constrained to the interval [a,b], the other being the truncatedU.^[2] Truncated normals with fixed support form an exponential family.Nielsen^[3] reported closed-form formula for calculating the Kullback-Leibler divergence and the Bhattacharyya distance between two truncated normal distributions with the support of the first distribution nested into the support of the second distribution.

If the random variable has been truncated only from below, some probability mass has been shifted to higher values, giving afirst-order stochastically dominating distribution and hence increasing the mean to a value higher than the mean${\displaystyle \mu }$ of the original normal distribution. Likewise, if the random variable has been truncated only from above, the truncated distribution has a mean less than${\displaystyle \mu .}$

Regardless of whether the random variable is bounded above, below, or both, the truncation is amean-preserving contraction combined with a mean-changing rigid shift, and hence the variance of the truncated distribution is less than the variance${\displaystyle {\sigma }^2}$ of the original normal distribution.

Source:^[4]

Let${\displaystyle \alpha =(a-\mu )/\sigma }$ and${\displaystyle \beta =(b-\mu )/\sigma }$. Then:and

Care must be taken in the numerical evaluation of these formulas, which can result incatastrophic cancellation when the interval${\displaystyle [a,b]}$ does not include${\displaystyle \mu }$. There are better ways to rewrite them that avoid this issue.^[5]

Sources:^[6][7]

In this case${\displaystyle b=\mathrm{\infty },\phi (\beta )=0,\mathrm{\Phi }(\beta )=1,}$ then

$${\displaystyle \mathrm{E}(X\mid X>a)=\mu +\sigma \phi (\alpha )/Z,}$$

and

$${\displaystyle \mathrm{Var}(X\mid X>a)={\sigma }^2[1+\alpha \phi (\alpha )/Z-(\phi (\alpha )/Z{)}^2],}$$

where${\displaystyle Z=1-\mathrm{\Phi }(\alpha ).}$

In this case${\displaystyle a=\alpha =-\mathrm{\infty },\phi (\alpha )=0,\mathrm{\Phi }(\alpha )=0,}$ then

Barr & Sherrill (1999) give a simpler expression for the variance of one sided truncations. Their formula is in terms of the chi-square CDF, which is implemented in standard software libraries.Bebu & Mathew (2009) provide formulas for (generalized) confidence intervals around the truncated moments.

As for the non-truncated case, there is a recursive formula for the truncated moments.^[8]

In particular, for${\displaystyle n\ge 0}$, we have

$${\displaystyle \mathrm{E}\left[{\left(\frac{x-\mu }{\sigma }\right)}^{n+2}\right]=\frac{{\alpha }^{n+1}\phi (\alpha )-{\beta }^{n+1}\phi (\beta )}{\mathrm{\Phi }(\beta )-\mathrm{\Phi }(\alpha )}+(n+1)\mathrm{E}\left[{\left(\frac{x-\mu }{\sigma }\right)}^n\right].}$$

By the change of variables${\displaystyle \xi =(x-\mu )/\sigma }$, one obtains$${\displaystyle \mathrm{E}\left[{\left(\frac{x-\mu }{\sigma }\right)}^{n+2}\right]={\int }_{\alpha }^{\beta }\frac{{\xi }^{n+2}\phi (\xi )}{\mathrm{\Phi }(\beta )-\mathrm{\Phi }(\alpha )}d\xi .}$$Using${\displaystyle {\phi }^{\prime }(\xi )=-\xi \phi (\xi ),}$integration by parts yields$${\displaystyle \mathrm{E}\left[{\left(\frac{x-\mu }{\sigma }\right)}^{n+2}\right]={\left[\frac{-{\xi }^{n+1}\phi (\xi )}{\mathrm{\Phi }(\beta )-\mathrm{\Phi }(\alpha )}\right]}_{\alpha }^{\beta }+(n+1){\int }_{\alpha }^{\beta }\frac{{\xi }^n\phi (\xi )}{\mathrm{\Phi }(\beta )-\mathrm{\Phi }(\alpha )}d\xi ,}$$which gives the equation to be proven.

Computing the moments of a multivariate truncated normal is harder.

This section'suse ofexternal links may not follow Wikipedia's policies or guidelines. Pleaseimprove this article by removingexcessive orinappropriate external links, and converting useful links where appropriate intofootnote references.(May 2022) (Learn how and when to remove this message)

A random variate${\displaystyle x}$ defined as${\displaystyle x={\mathrm{\Phi }}^{-1}(\mathrm{\Phi }(\alpha )+U\cdot (\mathrm{\Phi }(\beta )-\mathrm{\Phi }(\alpha )))\sigma +\mu }$ with${\displaystyle \mathrm{\Phi }}$ the cumulative distribution function of the normal distribution to be sampled from (i.e. with correct mean and variance) and${\displaystyle {\mathrm{\Phi }}^{-1}}$ its inverse,${\displaystyle U}$ a uniform random number on${\displaystyle (0,1)}$, follows the distribution truncated to the range${\displaystyle (a,b)}$. This is simply theinverse transform method for simulating random variables. Although one of the simplest, this method can either fail when sampling in the tail of the normal distribution,^[9] or be much too slow.^[10] Thus, in practice, one has to find alternative methods of simulation.

One such truncated normal generator (implemented inMatlab andinR (programming language) astrandn.R ) is based on an acceptance rejection idea due to Marsaglia.^[11] Despite the slightly suboptimal acceptance rate ofMarsaglia (1964) in comparison withRobert (1995), Marsaglia's method is typically faster,^[10] because it does not require the costly numerical evaluation of the exponential function.

For more on simulating a draw from the truncated normal distribution, seeRobert (1995),Lynch (2007, Section 8.1.3 (pages 200–206)),Devroye (1986). TheMSM package in R has a function,rtnorm, that calculates draws from a truncated normal. Thetruncnorm package in R also has functions to draw from a truncated normal.

Chopin (2011) proposed (arXiv) an algorithm inspired from theZiggurat algorithm of Marsaglia and Tsang (1984, 2000), which is usually considered as the fastest Gaussian sampler, and is also very close to Ahrens's algorithm (1995). Implementations can be found inC,C++,Matlab andPython.

Sampling from themultivariate truncated normal distribution is considerably more difficult.^[12] Exact or perfect simulation is only feasible in the case of truncation of the normal distribution to a polytope region.^[12][13] In more general cases,Damien & Walker (2001) introduce a general methodology for sampling truncated densities within aGibbs sampling framework. Their algorithm introduces one latent variable and, within a Gibbs sampling framework, it is more computationally efficient than the algorithm ofRobert (1995).

• Folded normal distribution
• Half-normal distribution
• Modified half-normal distribution^[14] with the pdf on${\displaystyle (0,\mathrm{\infty })}$ is given as${\displaystyle f(x)=\frac{2{\beta }^{\frac{\alpha }{2}}{x}^{\alpha -1}\exp (-\beta x^2+\gamma x)}{\mathrm{\Psi }\left(\frac{\alpha }{2},\frac{\gamma }{\sqrt{\beta }}\right)}}$, where${\displaystyle \mathrm{\Psi }(\alpha ,z)={}_1{\mathrm{\Psi }}_1\left(\begin{array}{c}\left(\alpha ,\frac{1}{2}\right)\\ (1,0)\end{array};z\right)}$ denotes theFox–Wright Psi function.
• Normal distribution
• Rectified Gaussian distribution
• Truncated distribution
• PERT distribution

• Botev, Zdravko & L'Ecuyer, Pierre (2018). "Chapter 8: Simulation from the Tail of the Univariate and Multivariate Normal Distribution". In Puliafito, Antonio (ed.).Systems Modeling: Methodologies and Tools. EAI/Springer Innovations in Communication and Computing. Springer, Cham. pp. 115–132.doi:10.1007/978-3-319-92378-9_8.ISBN 978-3-319-92377-2.S2CID 125554530.
• Devroye, Luc (1986).Non-Uniform Random Variate Generation(PDF). New York: Springer-Verlag. Archived fromthe original(PDF) on 2014-08-18. Retrieved2012-04-12.
• Greene, William H. (2003).Econometric Analysis (5th ed.). Prentice Hall.ISBN 978-0-13-066189-0.
• Norman L. Johnson and Samuel Kotz (1970).Continuous univariate distributions-1, chapter 13. John Wiley & Sons.
• Lynch, Scott (2007).Introduction to Applied Bayesian Statistics and Estimation for Social Scientists. New York: Springer.ISBN 978-1-4419-2434-6.
• Robert, Christian P. (1995). "Simulation of truncated normal variables".Statistics and Computing.5 (2):121–125.arXiv:0907.4010.doi:10.1007/BF00143942.S2CID 15943491.
• Barr, Donald R.; Sherrill, E.Todd (1999). "Mean and variance of truncated normal distributions".The American Statistician.53 (4):357–361.doi:10.1080/00031305.1999.10474490.
• Bebu, Ionut; Mathew, Thomas (2009). "Confidence intervals for limited moments and truncated moments in normal and lognormal models".Statistics and Probability Letters.79 (3):375–380.doi:10.1016/j.spl.2008.09.006.
• Damien, Paul; Walker, Stephen G. (2001). "Sampling truncated normal, beta, and gamma densities".Journal of Computational and Graphical Statistics.10 (2):206–215.doi:10.1198/10618600152627906.S2CID 123156320.
• Chopin, Nicolas (2011-04-01)."Fast simulation of truncated Gaussian distributions".Statistics and Computing.21 (2):275–288.arXiv:1201.6140.doi:10.1007/s11222-009-9168-1.ISSN 1573-1375.
• Burkardt, John."The Truncated Normal Distribution"(PDF).Department of Scientific Computing website. Florida State University. Retrieved15 February 2018.

• Continuous distributions
• Normal distribution

• CS1 maint: location
• Articles with short description
• Short description is different from Wikidata
• Articles lacking in-text citations from June 2010
• All articles lacking in-text citations
• Wikipedia external links cleanup from May 2022