Gauss-Hermite Integration

One-dimensional Gauss-Hermite integration

Gauss-Hermite integration approximates a specific integral as follows

f(x)exp(x2)dxj=1Nwjf(xj)

in which wj and xj are weights and abscissas for N-point Gauss-Hermite integration, see Abramowitz and Stegun (1972). N-point integration is exact for all polynomials f(x) of degree 2N-1, see Dahlquist and Bjorck (1974). This can for instance be used to approximate the mean of a function F(Y) of a normally distributed random variable Y with mean μ and variance σ2:

F(x)12πσexp((yμ)22σ2)dy
=F(μ+2σx)1πexp(x2)dx
=1πj=1NwjF(μ+2σxj)

Two-dimensional Gauss-Hermite integration

One-dimensional Gauss-Hermite integration can readily be extended to two dimensions. The following principal result in two dimensions is more or less given in Jäckel (2005) for the standard bivariate normal distribution ϕ(x,y;ρ) with correlation parameter ρ :

F(x,y)ϕ(x,y;ρ)dxdy1πi=1Nj=1NwiwjF(2[axi+bxj],2[bxi+axj])

in which

a=1+ρ+1ρ2

and

b=1+ρ1ρ2

as given in Jäckel (2005) .

Jäckel (2005) discusses other Gauss-Hermite approximations to the two-dimensional integral, but found that the approximation given above generally gives the most accurate results. For the general bivariate normal distribution with means (μx,μy) and variances (σx2,σy2) the integral can be approximated by means of

1πi=1Nj=1NwiwjF(μx+σx2[axi+bxj],μy+σy2[bxi+axj])

The product wiwj can be very small, especially when many quadrature points are used, thus wasting possibly precious calculation time. This can be remedied by pruning, i.e. by dropping combinations of (i,j) with very small values of the product wiwj.

Maximum likelihood for the LNN model with two-dimensional Gauss-Hermite integration

Denote non-consumption on day j for individual i as Yij=0. The conditional likelihood, i.e. given random effects bi and vi, of a non-consumption on day j equals, with H() the inverse of the logit function

P(Yij=0|bi,vi)=1H(λ+vi).

The conditional likelihood of a positive intake Yij > 0 equals, with ϕ the density of the normal distribution

f(Yij=yij|yij>0,bi,vi)=H(λ+vi)ϕ(yijμbi;0,σw2)

The conditional likelihood contribution for individual i is the product of the individual contributions for each day. The marginal likelihood contribution for individual i is obtained by integrating over the possible values of bi and vi. Since the pair (bi,vi) follows a bivariate normal distribution, the likelihood contribution for individual i can be approximated by means of two-dimensional Gauss-Hermite integration. Individually based covariables, such as sex or age, imply that μi and λi must be used instead of μ and λ. The likelihood must be optimized by means of some general optimization routine.