Is there a way to evaluate Eigen information without using numbers/values in R?
Here is my example, I am working with a three equation system from Bairagi et al 2007, Role of infection on the stability of a predator–prey system with several
response functions—A comparative study. I would like to know if there is a way to analyze a jacobian without using numbers/values (using variables, instead). Following is my example:
Equations
dsdtau = expression(b*s-b*s^2-b*s*i-s*i-m*s*p)
didtau = expression(s*i - d*i*p - e*i)
dpdtau = expression(-theta*d*i*p - g*p + theta*m*s*p)
Partial derivatives
dsds <- D(dsdtau,"s")
dids <- D(dsdtau,"i")
dpds <- D(dsdtau,"p")
dsdi <- D(didtau,"s")
didi <- D(didtau,"i")
dpdi <- D(didtau,"p")
dsdp <- D(dpdtau,"s")
didp <- D(dpdtau,"i")
dpdp <- D(dpdtau,"p")
Jacobian
j1 <- matrix(c(
dsds,dids,dpds,
dsdi,didi,dpdi,
dsdp,didp,dpdp),
nrow = 3, byrow = TRUE)
First, is there a way to coerce R into giving me the full details in matrix form? For example, when I run
j1
I get
[,1] [,2] [,3]
[1,] Expression Expression Expression
[2,] ? Expression Expression
[3,] Expression Expression Expression
but when I run j1[1], I get:
b - b * (2 * s) - b * i - i - m * p
which is what I want.
Second, is there a way to analyze this jacobian using the variables rather than plugging in values? For example, at (s,i,p) = (0,0,0) there is a trivial equilibrium with lambda = (b,-e,-g) which is an unstable saddle point (as all variables are greater than 0) and not particularly exciting. But for non-trivial equilibria the math is more labor intensive, and if R knows a way, I'd love help figuring it out!
If you're open to storing j1 as a list, here's a way to explicitly show each expression. As far as evaluating goes, that's something I would need more time to investigate. For now, the following list of functions seems promising: match.call(), do.call(), eval(), and get().
# store j1
# as a list object
# naming each set of equations
j1 <-
list(
didtau = c( dsds,dids,dpds )
, dpdtau = c( dsdi,didi,dpdi )
, dsdtau = c( dsdp,didp,dpdp )
)
# view the list
j1
# $didtau
# $didtau[[1]]
# b - b * (2 * s) - b * i - i - m * p
#
# $didtau[[2]]
# -(b * s + s)
#
# $didtau[[3]]
# -(m * s)
#
#
# $dpdtau
# $dpdtau[[1]]
# i
#
# $dpdtau[[2]]
# s - d * p - e
#
# $dpdtau[[3]]
# -(d * i)
#
#
# $dsdtau
# $dsdtau[[1]]
# theta * m * p
#
# $dsdtau[[2]]
# -theta * d * p
#
# $dsdtau[[3]]
# -theta * d * i - g + theta * m * s
# end of script #
Related
I'm given a question in R language to find the 30th term of the recurrence relation x(n) = 2*x(n-1) - x(n-2), where x(1) = 0 and x(2) = 1. I know the answer is 29 from mathematical deduction. But as a newbie to R, I'm slightly confused by how to make things work here. The following is my code:
loop <- function(n){
a <- 0
b <- 1
for (i in 1:30){
a <- b
b <- 2*b - a
}
return(a)
}
loop(30)
I'm returned 1 as a result, which is way off.
In case you're wondering why this looks Python-ish, I've mostly only been exposed to Python programming thus far (I'm new to programming in general). I've tried to check out all the syntax in R, but I suppose my logic is quite fixed by Python. Can someone help me out in this case? In addition, does R have any resources like PythonTutor to help visualise the code execution logic?
Thank you!
I guess what you need might be something like below
loop <- function(n){
if (n<=2) return(n-1)
a <- 0
b <- 1
for (i in 3:n){
a_new <- b
b <- 2*b - a
a <- a_new
}
return(b)
}
then
> loop(30)
[1] 29
If you need a recursion version, below is one realization
loop <- function(n) {
if (n<=2) return(n-1)
2*loop(n-1)-loop(n-2)
}
which also gives
> loop(30)
[1] 29
You can solve it another couple of ways.
Solve the linear homogeneous recurrence relation, let
x(n) = r^n
plugging into the recurrence relation, you get the quadratic
r^n-2*r^(n-1)+r^(n-2) = 0
, i.e.,
r^2-2*r+1=0
, i.e.,
r = 1, 1
leading to general solution
x(n) = c1 * 1^n + c2 * n * 1^n = c1 + n * c2
and with x(1) = 0 and x(2) = 1, you get c2 = 1, c1 = -1, s.t.,
x(n) = n - 1
=> x(30) = 29
Hence, R code to compute x(n) as a function of n is trivial, as shown below:
x <- function(n) {
return (n-1)
}
x(30)
#29
Use matrix powers (first find the following matrix A from the recurrence relation):
(The matrix A has algebraic / geometric multiplicity, its corresponding eigenvectors matrix is singular, otherwise you could use spectral decomposition yourself for fast computation of matrix powers, here we shall use the library expm as shown below)
library(expm)
A <- matrix(c(2,1,-1,0), nrow=2)
A %^% 29 %*% c(1,0) # [x(31) x(30)]T = A^29.[x(2) x(1)]T
# [,1]
# [1,] 30 # x(31)
# [2,] 29 # x(30)
# compute x(n)
x <- function(n) {
(A %^% (n-1) %*% c(1,0))[2]
}
x(30)
# 29
You're not using the variable you're iterating on in the loop, so nothing is updating.
loop <- function(n){
a <- 0
b <- 1
for (i in 1:30){
a <- b
b <- 2*i - a
}
return(a)
}
You could define a recursive function.
f <- function(x, n) {
n <- 1:n
r <- function(n) {
if (length(n) == 2) x[2]
else r({
x <<- c(x[2], 2*x[2] - x[1])
n[-1]
})
}
r(n)
}
x <- c(0, 1)
f(x, 30)
# [1] 29
I wonder if there's any function that can solve for a variable in an equation in terms of other values.
For example, I have:
(a-1/3)/(a+b-2/3)==0.3
The output of this function should be:
(0.3*b+0.1333333)/(0.7)
Or something along these lines.
Thank you
You can use Yacas through the Ryacas package:
library(Ryacas)
a <- Sym("a")
b <- Sym("b")
Solve((a-1/3)/(a+b-2/3)==0.3, a)
# Yacas vector:
# [1] a == -((-0.3 * b - 0.1333333333)/0.7)
To get the solution as an expression, do:
solution <- Solve((a-1/3)/(a+b-2/3)==0.3, a)
yacas(paste0("a Where ", solution))
# expression(-((-0.3 * b - 0.1333333333)/0.7))
You can define a function returning the solution in function of b as follows:
f <- function(b) {}
body(f) <- yacas(paste0("a Where ", solution))$text
f
# function (x)
# -((-0.3 * b - 0.1333333333)/0.7)
Also note that you do rational calculus to get exact values:
q1 <- Sym(1)/Sym(3)
q2 <- Sym(2)/Sym(3)
solution <- Solve((a-q1)/(a+b-q2)==0.3, a)
solution
# [1] a == -(3 * (-0.3 * b - 1.2/9)/2.1)
Simplify(solution)
# [1] a + (-0.9 * b/2.1 - 3.6/18.9) == 0
Suppose I have the following system of equations:
a * b = 5
sqrt(a * b^2) = 10
How can I solve these equations for a and b in R ?
I guess this problem can be stated as an optimisation problem, with the following function... ?
fn <- function(a, b) {
rate <- a * b
shape <- sqrt(a * b^2)
return(c(rate, shape) )
}
In a comment the poster specifically asks about using solve and optim so we show how to solve this (1) by hand, (2) using solve, (3) using optim and (4) a fixed point iteration.
1) by hand First note that if we write a = 5/b based on the first equation and substitute that into the second equation we get sqrt(5/b * b^2) = sqrt(5 * b) = 10 so b = 20 and a = 0.25.
2) solve Regarding the use of solve these equations can be transformed into linear form by taking the log of both sides giving:
log(a) + log(b) = log(5)
0.5 * (loga + 2 * log(b)) = log(10)
which can be expressed as:
m <- matrix(c(1, .5, 1, 1), 2)
exp(solve(m, log(c(5, 10))))
## [1] 0.25 20.00
3) optim Using optim we can write this where fn is from the question. fn2 is formed by subtracting off the RHS of the equations and using crossprod to form the sum of squares.
fn2 <- function(x) crossprod( fn(x[1], x[2]) - c(5, 10))
optim(c(1, 1), fn2)
giving:
$par
[1] 0.2500805 19.9958117
$value
[1] 5.51508e-07
$counts
function gradient
97 NA
$convergence
[1] 0
$message
NULL
4) fixed point For this one rewrite the equations in a fixed point form, i.e. in the form c(a, b) = f(c(a, b)) and then iterate. In general, there will be several ways to do this and not all of them will converge but in this case this seems to work. We use starting values of 1 for both a and b and divide both side of the first equation by b to get the first equation in fixed point form and we divide both sides of the second equation by sqrt(a) to get the second equation in fixed point form:
a <- b <- 1 # starting values
for(i in 1:100) {
a = 5 / b
b = 10 / sqrt(a)
}
data.frame(a, b)
## a b
## 1 0.25 20
Use this library.
library("nleqslv")
You need to define the multivariate function you want to solve for.
fn <- function(x) {
rate <- x[1] * x[2] - 5
shape <- sqrt(x[1] * x[2]^2) - 10
return(c(rate, shape))
}
Then you're good to go.
nleqslv(c(1,5), fn)
Always look at the detailed results. Numerical calculations can be tricky. In this case I got this:
Warning message:
In sqrt(x[1] * x[2]^2) : NaNs produced
That just means the procedure searched a region that included x[1] < 0 and then presumably noped the heck back to the right hand side of the plane.
I have some trouble in order to solve my set of linear equations.
I have three 3D points (A, B, C) in my example and I want to automate the solving of my system. I want to create a plane with these 3 points.
It's very simple manually (mathematically) but I don't see why I don't solve my problem when I code...
I have a system of cartesian equation which is the equation of a plane : ax+by+cz+d=0
xAx + yAy + zA*z +d = 0 #point A
xBx + yBy + zB*z +d = 0 #point B
etc
I use a matrix, for example A=(0,0,1) ; B=(4,2,3) and C=(-3,1,0).
With manual solving, I have for this example this solution : x+3y-5z+5=0.
For resolving it in R : I wanted to use solve().
A <- c(0,0,1)
B <- c(4,2,3)
C <- c(-3,1,0)
res0 <- c(-d,-d,-d) #I don't know how having it so I tried c(0,0,0) cause each equation = 0. But I really don't know for that !
#' #param A vector 3x1 with the 3d coordinates of the point A
carteq <- function(A, B, C, res0) {
matrixtest0 <- matrix(c(A[1], A[2], A[3], B[1], B[2], B[3],C[1], C[2], C[3]), ncol=3) #I tried to add the 4th column for solving "d" but that doesn't work.
#checking the invertibility of my matrix
out <- tryCatch(determinant(matrixtest0)$modulus<threshold, error = function(e) e)#or out <- tryCatch(solve(X) %*% X, error = function(e) e)
abcd <- solve(matrixtest0, res0) #returns just 3 values
abcd <- qr.solve(matrixtest0, res0) #returns just 3 values
}
That's not the good method... But I don't know how I can add the "d" in my problem.
The return that I need is : return(a, b, c, d)
I thing that my problem is classical and easy, but I don't find a function like solve() or qr.solve() which can solve my problem...
Your solution is actually wrong:
A <- c(0,0,1)
B <- c(4,2,3)
C <- c(-3,1,0)
CrossProduct3D <- function(x, y, i=1:3) {
#http://stackoverflow.com/a/21736807/1412059
To3D <- function(x) head(c(x, rep(0, 3)), 3)
x <- To3D(x)
y <- To3D(y)
Index3D <- function(i) (i - 1) %% 3 + 1
return (x[Index3D(i + 1)] * y[Index3D(i + 2)] -
x[Index3D(i + 2)] * y[Index3D(i + 1)])
}
N <- CrossProduct3D(A - B, C - B)
#[1] 4 2 -10
d <- -sum(N * B)
#[1] 10
#test it:
crossprod(A, N) + d
# [,1]
#[1,] 0
crossprod(B, N) + d
# [,1]
#[1,] 0
crossprod(C, N) + d
# [,1]
#[1,] 0
I have a 2396x34 double matrix named y wherein each row (2396) represents a separate situation consisting of 34 consecutive time segments.
I also have a numeric[34] named x that represents a single situation of 34 consecutive time segments.
Currently I am calculating the correlation between each row in y and x like this:
crs[,2] <- cor(t(y),x)
What I need now is to replace the cor function in the above statement with a weighted correlation. The weight vector xy.wt is 34 elements long so that a different weight can be assigned to each of the 34 consecutive time segments.
I found the Weighted Covariance Matrix function cov.wt and thought that if I first scale the data it should work just like the cor function. In fact you can specify for the function to return a correlation matrix as well. Unfortunately it does not seem like I can use it in the same manner because I cannot supply my two variables (x and y) separately.
Does anyone know of a way I can get a weighted correlation in the manner I described without sacrificing much speed?
Edit: Perhaps some mathematical function could be applied to y prior to the cor function in order to get the same results that I'm looking for. Maybe if I multiply each element by xy.wt/sum(xy.wt)?
Edit #2 I found another function corr in the boot package.
corr(d, w = rep(1, nrow(d))/nrow(d))
d
A matrix with two columns corresponding to the two variables whose correlation we wish to calculate.
w
A vector of weights to be applied to each pair of observations. The default is equal weights for each pair. Normalization takes place within the function so sum(w) need not equal 1.
This also is not what I need but it is closer.
Edit #3
Here is some code to generate the type of data I am working with:
x<-cumsum(rnorm(34))
y<- t(sapply(1:2396,function(u) cumsum(rnorm(34))))
xy.wt<-1/(34:1)
crs<-cor(t(y),x) #this works but I want to use xy.wt as weight
Unfortunately the accepted answer is wrong when y is a matrix of more than one row. The error is in the line
vy <- rowSums( w * y * y )
We want to multiply the columns of y by w, but this will multiply the rows by the elements of w, recycled as necessary. Thus
> f(x, y[1, , drop = FALSE], xy.wt)
[1] 0.103021
is correct, because in this case the multiplication is performed element-wise, which is equivalent to column-wise multiplication here, but
> f(x, y, xy.wt)[1]
[1] 0.05463575
gives a wrong answer due to the row-wise multiplication.
We can correct the function as follows
f2 <- function( x, y, w = rep(1,length(x))) {
stopifnot(length(x) == dim(y)[2] )
w <- w / sum(w)
# Center x and y, using the weighted means
x <- x - sum(x * w)
ty <- t(y - colSums(t(y) * w))
# Compute the variance
vx <- sum(w * x * x)
vy <- colSums(w * ty * ty)
# Compute the covariance
vxy <- colSums(ty * x * w)
# Compute the correlation
vxy / sqrt(vx * vy)
}
and check the results against those produced by corr from the boot package:
> res1 <- f2(x, y, xy.wt)
> res2 <- sapply(1:nrow(y),
+ function(i, x, y, w) corr(cbind(x, y[i,]), w = w),
+ x = x, y = y, w = xy.wt)
> all.equal(res1, res2)
[1] TRUE
which in itself gives another way that this problem could be solved.
You can go back to the definition of the correlation.
f <- function( x, y, w = rep(1,length(x))) {
stopifnot( length(x) == dim(y)[2] )
w <- w / sum(w)
# Center x and y, using the weighted means
x <- x - sum(x*w)
y <- y - apply( t(y) * w, 2, sum )
# Compute the variance
vx <- sum( w * x * x )
vy <- rowSums( w * y * y ) # Incorrect: see Heather's remark, in the other answer
# Compute the covariance
vxy <- colSums( t(y) * x * w )
# Compute the correlation
vxy / sqrt(vx * vy)
}
f(x,y)[1]
cor(x,y[1,]) # Identical
f(x, y, xy.wt)
Here is a generalization to compute the weighted Pearson correlation between two matrices (instead of a vector and a matrix, as in the original question):
matrix.corr <- function (a, b, w = rep(1, nrow(a))/nrow(a))
{
# normalize weights
w <- w / sum(w)
# center matrices
a <- sweep(a, 2, colSums(a * w))
b <- sweep(b, 2, colSums(b * w))
# compute weighted correlation
t(w*a) %*% b / sqrt( colSums(w * a**2) %*% t(colSums(w * b**2)) )
}
Using the above example and the correlation function from Heather, we can verify it:
> sum(matrix.corr(as.matrix(x, nrow=34),t(y),xy.wt) - f2(x,y,xy.wt))
[1] 1.537507e-15
In terms of calling syntax, this resembles the unweighted cor:
> a <- matrix( c(1,2,3,1,3,2), nrow=3)
> b <- matrix( c(2,3,1,1,7,3,5,2,8,1,10,12), nrow=3)
> matrix.corr(a,b)
[,1] [,2] [,3] [,4]
[1,] -0.5 0.3273268 0.5 0.9386522
[2,] 0.5 0.9819805 -0.5 0.7679882
> cor(a, b)
[,1] [,2] [,3] [,4]
[1,] -0.5 0.3273268 0.5 0.9386522
[2,] 0.5 0.9819805 -0.5 0.7679882