# ml_taj: internal In mgpd: mgpd: Functions for multivariate generalized Pareto distribution (MGPD of Type II)

## Description

internal use only

## Usage

 `1` ```ml_taj(param, dat, mlmax = 1e+15, fixed = FALSE, ...) ```

## Arguments

 `param` `dat` `mlmax` `fixed` `...`

## Details

internal use only

## Value

internal use only

## Note

internal use only

P. Rakonczai

## References

internal use only

 ``` 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65``` ```##---- Should be DIRECTLY executable !! ---- ##-- ==> Define data, use random, ##-- or do help(data=index) for the standard data sets. ## The function is currently defined as function (param, dat, mlmax = 1e+15, fixed = FALSE, ...) { loglik = mlmax hxy = NA x = dat[, 1] y = dat[, 2] error = FALSE mux = param[1] muy = param[4] sigx = param[2] sigy = param[5] gamx = param[3] gamy = param[6] a = param[7] b = param[8] mu = function(x, y) ((1/x)^(2 * a) + 2 * (1 + b) * (1/x/y)^(a) + (1/y)^(2 * a))^(1/2/a) dxdymu = function(x, y) (1/4) * ((1/x)^(2 * a) + (2 * (1 + b)) * (1/(x * y))^a + (1/y)^(2 * a))^(1/(2 * a)) * (-(2 * (1 + b)) * (1/(x * y))^a * a/y - 2 * (1/y)^(2 * a) * a/y) * (-2 * (1/x)^(2 * a) * a/x - (2 * (1 + b)) * (1/(x * y))^a * a/x)/(a^2 * ((1/x)^(2 * a) + (2 * (1 + b)) * (1/(x * y))^a + (1/y)^(2 * a))^2) + ((1/x)^(2 * a) + (2 * (1 + b)) * (1/(x * y))^a + (1/y)^(2 * a))^(1/(2 * a)) * a * (1 + b) * (1/(x * y))^a/(y * x * ((1/x)^(2 * a) + (2 * (1 + b)) * (1/(x * y))^a + (1/y)^(2 * a))) - (1/2) * ((1/x)^(2 * a) + (2 * (1 + b)) * (1/(x * y))^a + (1/y)^(2 * a))^(1/(2 * a)) * (-2 * (1/x)^(2 * a) * a/x - (2 * (1 + b)) * (1/(x * y))^a * a/x) * (-(2 * (1 + b)) * (1/(x * y))^a * a/y - 2 * (1/y)^(2 * a) * a/y)/(a * ((1/x)^(2 * a) + (2 * (1 + b)) * (1/(x * y))^a + (1/y)^(2 * a))^2) if (sigx < 0 | sigy < 0 | a < 1 | b <= -1 | (b > (2 * a - 2))) error = TRUE if (fixed == TRUE) { mux = 0 } if (error) loglik = mlmax if (!error) { tx = (1 + gamx * (x - mux)/sigx)^(1/gamx) ty = (1 + gamy * (y - muy)/sigy)^(1/gamy) tx0 = (1 + gamx * (-mux)/sigx)^(1/gamx) ty0 = (1 + gamy * (-muy)/sigy)^(1/gamy) dtx = (1/sigx) * pmax((1 + gamx * (x - mux)/sigx), 0)^(1/gamx - 1) dty = (1/sigy) * pmax((1 + gamy * (y - muy)/sigy), 0)^(1/gamy - 1) c0 = -mu(tx0, ty0) hxy = 1/c0 * dxdymu(tx, ty) * dtx * dty hxy = as.numeric(hxy * (1 - ((x < 0) * (y < 0)))) loglik = -sum(log(hxy)) } if (min(1 + gamx * (x - mux)/sigx) < 0) loglik = mlmax if (min(1 + gamy * (y - muy)/sigy) < 0) loglik = mlmax loglik } ```