fds2s: Mass fractal dimension of sampling 2D clusters

In SECP: Statistical Estimation of Cluster Parameters

Mass fractal dimension of sampling 2D clusters

Description

fds2s() function uses a linear regression model for statistical estimation of the mass fractal dimension of sampling clusters on 2D square lattice with iso- & anisotropic sets cover.

Usage

fds2s(rfq=fssi20(x=95), bnd=isc2s(k=12, x=dim(rfq)))

Arguments

rfq	relative sampling frequencies for sites of the percolation lattice.
bnd	bounds for the iso- or anisotropic set cover.

Details

The mass fractal dimension for sampling clusters is equal to the coefficient of linear regression between log(w) and log(r), where w is a relative sampling frequency of the total sites which are bounded elements of iso- & anisotropic sets cover.

The isotropic set cover on 2D square lattice is formed from scalable squares with variable sizes 2r+1 and a fixed point in the lattice center.

The anisotropic set cover on 2D square lattice is formed from scalable rectangles with variable sizes r+1 and a fixed edge along the lattice boundary.

The percolation is simulated on 2D square lattice with uniformly weighted sites and the constant parameter p.

The isotropic cluster is formed from the accessible sites connected with initial sites subset.

Each element of the matrix rfq is equal to the relative frequency with which the 2D square lattice site belongs to a cluster sample.

Value

A linear regression model for statistical estimation of the mass fractal dimension of sampling clusters on 2D square lattice with iso- & anisotropic sets cover.

Author(s)

Pavel V. Moskalev

References

Moskalev P.V., Grebennikov K.V. and Shitov V.V. (2011) Statistical estimation of percolation cluster parameters. Proceedings of Voronezh State University. Series: Systems Analysis and Information Technologies, No.1 (January-June), pp.29-35, arXiv:1105.2334v1; in Russian.

See Also

fds3s, fdc2s, fdc3s

Examples

# # # # # # # # # # # # # # # # #
# Example 1: Isotropic set cover
# # # # # # # # # # # # # # # # #
pc <- .592746
p1 <- pc - .03
p2 <- pc + .03
lx <- 33; ss <- (lx+1)/2
rf1 <- fssi20(n=100, x=lx, p=p1)
rf2 <- fssi20(n=100, x=lx, p=p2)
bnd <- isc2s(k=9, x=dim(rf1))
fd1 <- fds2s(rfq=rf1, bnd=bnd)
fd2 <- fds2s(rfq=rf2, bnd=bnd)
w1 <- fd1$model[,"w"]; w2 <- fd2$model[,"w"]
r1 <- fd1$model[,"r"]; r2 <- fd2$model[,"r"]
rr <- seq(min(r1)-.2, max(r1)+.2, length=100)
ww1 <- predict(fd1, newdata=list(r=rr), interval="conf")
ww2 <- predict(fd2, newdata=list(r=rr), interval="conf")
s1 <- paste(round(confint(fd1)[2,], digits=3), collapse=", ")
s2 <- paste(round(confint(fd2)[2,], digits=3), collapse=", ")
x <- y <- seq(lx)
par(mfrow=c(2,2), mar=c(3,3,3,1), mgp=c(2,1,0))
image(x, y, rf1, zlim=c(0, .7), cex.main=1,
      main=paste("Isotropic set cover and\n",
                 "a 2D clusters frequency with\n",
                 "(1,0)-neighborhood and p=",
                 round(p1, digits=3), sep=""))
rect(bnd["x1",], bnd["y1",], bnd["x2",], bnd["y2",])
abline(h=ss, lty=2); abline(v=ss, lty=2)
image(x, y, rf2, zlim=c(0, .7), cex.main=1,
      main=paste("Isotropic set cover and\n",
                 "a 2D clusters frequency with\n",
                 "(1,0)-neighborhood and p=",
                 round(p2, digits=3), sep=""))
rect(bnd["x1",], bnd["y1",], bnd["x2",], bnd["y2",])
abline(h=ss, lty=2); abline(v=ss, lty=2)
plot(r1, w1, pch=3, ylim=range(c(w1,w2)), cex.main=1,
     main=paste("0.95 confidence interval for the mass\n",
                "fractal dimension is (",s1,")", sep=""))
matlines(rr, ww1, lty=c(1,2,2), col=c("black","red","red"))
plot(r2, w2, pch=3, ylim=range(c(w1,w2)), cex.main=1,
     main=paste("0.95 confidence interval for the mass\n",
                "fractal dimension is (",s2,")", sep=""))
matlines(rr, ww2, lty=c(1,2,2), col=c("black","red","red"))

## Not run: 
# # # # # # # # # # # # # # # # # #
# Example 2: Anisotropic set cover, dir=2
# # # # # # # # # # # # # # # # # #
pc <- .592746
p1 <- pc - .03
p2 <- pc + .03
lx <- 33; ss <- (lx+1)/2
ssy <- seq(lx+2, 2*lx-1)
rf1 <- fssi20(n=100, x=lx, p=p1, set=ssy, all=FALSE)
rf2 <- fssi20(n=100, x=lx, p=p2, set=ssy, all=FALSE)
bnd <- asc2s(k=9, x=dim(rf1), dir=2)
fd1 <- fds2s(rfq=rf1, bnd=bnd)
fd2 <- fds2s(rfq=rf2, bnd=bnd)
w1 <- fd1$model[,"w"]; w2 <- fd2$model[,"w"]
r1 <- fd1$model[,"r"]; r2 <- fd2$model[,"r"]
rr <- seq(min(r1)-.2, max(r1)+.2, length=100)
ww1 <- predict(fd1, newdata=list(r=rr), interval="conf")
ww2 <- predict(fd2, newdata=list(r=rr), interval="conf")
s1 <- paste(round(confint(fd1)[2,], digits=3), collapse=", ")
s2 <- paste(round(confint(fd2)[2,], digits=3), collapse=", ")
x <- y <- seq(lx)
par(mfrow=c(2,2), mar=c(3,3,3,1), mgp=c(2,1,0))
image(x, y, rf1, zlim=c(0, .7), cex.main=1,
      main=paste("Anisotropic set cover and\n",
                 "a 2D clusters frequency with\n",
                 "(1,0)-neighborhood and p=",
                 round(p1, digits=3), sep=""))
rect(bnd["x1",], bnd["y1",], bnd["x2",], bnd["y2",])
abline(v=ss, lty=2)
image(x, y, rf2, zlim=c(0, .7), cex.main=1,
      main=paste("Anisotropic set cover and\n",
                 "a 2D clusters frequency with\n",
                 "(1,0)-neighborhood and p=",
                 round(p2, digits=3), sep=""))
rect(bnd["x1",], bnd["y1",], bnd["x2",], bnd["y2",])
abline(v=ss, lty=2)
plot(r1, w1, pch=3, ylim=range(c(w1,w2)), cex.main=1,
     main=paste("0.95 confidence interval for the mass\n",
                "fractal dimension is (",s1,")", sep=""))
matlines(rr, ww1, lty=c(1,2,2), col=c("black","red","red"))
plot(r2, w2, pch=3, ylim=range(c(w1,w2)), cex.main=1,
     main=paste("0.95 confidence interval for the mass\n",
                "fractal dimension is (",s2,")", sep=""))
matlines(rr, ww2, lty=c(1,2,2), col=c("black","red","red"))

## End(Not run)

SECP documentation built on May 11, 2022, 9:05 a.m.