mix.EM: Function to mix sample spectres.
In EMMAgeo: End-Member Modelling of Grain-Size Data

Description Usage Arguments Details Value Author(s) See Also Examples

This functions allows to mix grain-size distributions with specified proportions and defined noise levels, for example to test the goodness of the EMMA algorithm.

1	mix.EM(EM, proportion, noise, autocorrelation)

`EM`	`Numeric` matrix, grain-size distribution definitions. Each definition is in a separate row with variable contributions in columns.
`proportion`	`Numeric` vector, relative proportions of each distribution per sample.
`noise`	`Numeric` scalar, optional relative white noise levels.
`autocorrelation`	`Numeric` scalar, the degree of autocorrelation among classes. Autocorrelation is realised as running mean of the specified length. Only odd values are allowed.

The function multiplies each end-member with the respective proportion value, sums the resulting variables, adds uniform noise and normalises the resulting mixed sample to 100 %.

Numeric vector, a sample composed of known proportions of end-members.

Michael Dietze, Elisabeth Dietze

create.EM

## define end-member loadings and phi vector
EMa.1 <- create.EM(p1 = c(2, 8), p2 = c(1, 0.8), s = c(0.7, 0.3), 
                   boundaries = c(0, 11), n = 80)
EMa.2 <- create.EM(p1 = c(4, 7), p2 = c(1.1, 1.4), s = c(0.5, 0.5),
                   boundaries = c(0, 11), n = 80)
EMa   <- rbind(EMa.1, EMa.2)

phi   <- seq(0, 11, length.out = 80)

## mix end-member loadings
sample1 <- mix.EM(EMa, proportion = c(0.3, 0.7))
sample2 <- mix.EM(EMa, proportion = c(0.5, 0.5), noise = 0.1,
                  autocorrelation = 3)

## plot end-member loadings (grey) and resulting samples (black)
plot(phi, EMa.1, type="l", col = "grey")
lines(phi, EMa.2, col = "grey")
lines(phi, sample1)
lines(phi, sample2)