Debian Codegraph @

man/timelags.Rd

raw
1 \name{timelags}
2 \alias{timelags}
3 \alias{lagvalue}
4 \alias{lagderiv}
5 \title{
6 Time Lagged Values of State Variables and Derivatives.
7 }
8 \description{
9 Functions \code{lagvalue} and \code{lagderiv} provide access to past
10 (lagged) values of state variables and derivatives.
11
12 They are to be used with function \code{dede}, to solve delay differential
13 equations.
14 }
15 \usage{
16 lagvalue(t, nr)
17 lagderiv(t, nr)
18 }
19 \arguments{
20 \item{t }{the time for which the lagged value is wanted; this should
21 be no larger than the current simulation time and no smaller than the
22 initial simulation time.
23 }
24 \item{nr }{the number of the lagged value; if \code{NULL} then all state
25 variables or derivatives are returned.
26 }
27 }
28
29 \value{
30 a scalar (or vector) with the lagged value(s).
31 }
32 \author{Karline Soetaert <[email protected]>}
33
34 \details{
35 The \code{lagvalue} and \code{lagderiv} can only be called during the
36 integration, the lagged time should not be smaller than the initial
37 simulation time, nor should it be larger than the current simulation
38 time.
39
40 Cubic Hermite interpolation is used to obtain an accurate interpolant
41 at the requested lagged time.
42 }
43 \seealso{
44 \link{dede}, for how to implement delay differential equations.
45 }
46 \examples{
47 ## =============================================================================
48 ## exercise 6 from Shampine and Thompson, 2000
49 ## solving delay differential equations with dde23
50 ##
51 ## two lag values
52 ## =============================================================================
53
54 ##-----------------------------
55 ## the derivative function
56 ##-----------------------------
57 derivs <- function(t, y, parms) {
58 History <- function(t) c(cos(t), sin(t))
59 if (t < 1)
60 lag1 <- History(t - 1)[1]
61 else
62 lag1 <- lagvalue(t - 1)[1] # returns a vector; select first element
63
64 if (t < 2)
65 lag2 <- History(t - 2)[2]
66 else
67 lag2 <- lagvalue(t - 2,2) # faster than lagvalue(t - 2)[2]
68
69 dy1 <- lag1 * lag2
70 dy2 <- -y[1] * lag2
71
72 list(c(dy1, dy2), lag1 = lag1, lag2 = lag2)
73 }
74
75 ##-----------------------------
76 ## parameters
77 ##-----------------------------
78
79 r <- 3.5; m <- 19
80
81 ##-----------------------------
82 ## initial values and times
83 ##-----------------------------
84
85 yinit <- c(y1 = 0, y2 = 0)
86 times <- seq(0, 20, by = 0.01)
87
88 ##-----------------------------
89 ## solve the model
90 ##-----------------------------
91
92 yout <- dede(y = yinit, times = times, func = derivs,
93 parms = NULL, atol = 1e-9)
94
95 ##-----------------------------
96 ## plot results
97 ##-----------------------------
98
99 plot(yout, type = "l", lwd = 2)
100
101 ## =============================================================================
102 ## The predator-prey model with time lags, from Hale
103 ## problem 1 from Shampine and Thompson, 2000
104 ## solving delay differential equations with dde23
105 ##
106 ## a vector with lag valuess
107 ## =============================================================================
108
109 ##-----------------------------
110 ## the derivative function
111 ##-----------------------------
112 predprey <- function(t, y, parms) {
113 tlag <- t - 1
114 if (tlag < 0)
115 ylag <- c(80, 30)
116 else
117 ylag <- lagvalue(tlag) # returns a vector
118
119 dy1 <- a * y[1] * (1 - y[1]/m) + b * y[1] * y[2]
120 dy2 <- c * y[2] + d * ylag[1] * ylag[2]
121 list(c(dy1, dy2))
122 }
123
124 ##-----------------------------
125 ## parameters
126 ##-----------------------------
127
128 a <- 0.25; b <- -0.01; c <- -1 ; d <- 0.01; m <- 200
129
130 ##-----------------------------
131 ## initial values and times
132 ##-----------------------------
133
134 yinit <- c(y1 = 80, y2 = 30)
135 times <- seq(0, 100, by = 0.01)
136
137 #-----------------------------
138 # solve the model
139 #-----------------------------
140
141 yout <- dede(y = yinit, times = times, func = predprey, parms = NULL)
142
143 ##-----------------------------
144 ## display, plot results
145 ##-----------------------------
146
147 plot(yout, type = "l", lwd = 2, main = "Predator-prey model", mfrow = c(2, 2))
148 plot(yout[,2], yout[,3], xlab = "y1", ylab = "y2", type = "l", lwd = 2)
149
150 diagnostics(yout)
151
152 ## =============================================================================
153 ##
154 ## A neutral delay differential equation (lagged derivative)
155 ## y't = -y'(t-1), y(t) t < 0 = 1/t
156 ##
157 ## =============================================================================
158
159 ##-----------------------------
160 ## the derivative function
161 ##-----------------------------
162 derivs <- function(t, y, parms) {
163 tlag <- t - 1
164 if (tlag < 0)
165 dylag <- -1
166 else
167 dylag <- lagderiv(tlag)
168
169 list(c(dy = -dylag), dylag = dylag)
170 }
171
172 ##-----------------------------
173 ## initial values and times
174 ##-----------------------------
175
176 yinit <- 0
177 times <- seq(0, 4, 0.001)
178
179 ##-----------------------------
180 ## solve the model
181 ##-----------------------------
182
183 yout <- dede(y = yinit, times = times, func = derivs, parms = NULL)
184
185 ##-----------------------------
186 ## display, plot results
187 ##-----------------------------
188
189 plot(yout, type = "l", lwd = 2)
190 }
191
192 \keyword{utilities}