-
Notifications
You must be signed in to change notification settings - Fork 0
/
xiaolinzhuo_2router.R
251 lines (190 loc) · 6.8 KB
/
xiaolinzhuo_2router.R
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
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
### stat221 pset 5
### Xiaolin Zhuo
source("xiaolinzhuo_functions.R")
###### 1.9 fit locally iid model to 2-router data ######
###### ONLY THIS PART RUN ON ODYSSEY ######
df2 <- read.csv("2router_linkcount.dat")
unique(df2$nme)
nodes <- c("router5", "r4-local", "switch", "r4-others", "gw1", "gw2", "gw3", "gw-others")
# create matrix 'names' in the same order as the labels in figure 9
names <- matrix(paste("router5 >", nodes), nrow=1)
for (n in nodes[-1]) {
names <- rbind(paste(n, ">", nodes), names)
}
names
# A is 15 x 64 matrix
if (file.exists("matrixA_2router.csv")) {
A <- as.matrix(read.csv("matrixA_2router.csv"))
} else {
A <- matrix(c(rep(0, 56), rep(1, 8),
rep(0, 48), rep(1, 8), rep(0, 8),
rep(0, 40), rep(1, 8), rep(0, 16),
rep(0, 32), rep(1, 8), rep(0, 24),
rep(0, 24), rep(1, 8), rep(0, 32),
rep(0, 16), rep(1, 8), rep(0, 40),
rep(0, 8), rep(1, 8), rep(0, 48),
rep(1, 8), rep(0, 56),
as.numeric(1:64 %in% seq(1, 64, 8)),
as.numeric(1:64 %in% seq(2, 64, 8)),
as.numeric(1:64 %in% seq(3, 64, 8)),
as.numeric(1:64 %in% seq(4, 64, 8)),
as.numeric(1:64 %in% seq(5, 64, 8)),
as.numeric(1:64 %in% seq(6, 64, 8)),
as.numeric(1:64 %in% seq(7, 64, 8))), nrow=15, byrow=T)
write.csv(A, "matrixA_2router.csv", row.names=F)
}
# data Y
if (file.exists("matrixY_2router.csv")) {
Y <- as.matrix(read.csv("matrixY_2router.csv"))
} else {
Y <- matrix(df2[1:15, "value"], nrow=1)
for (i in seq(17, nrow(df2), 16)) {
Y <- rbind(Y, df2[i:(i+14), "value"])
}
write.csv(Y, "matrixY_2router.csv", row.names=F)
}
dim(Y)
head(Y)
c <- 2
w <- 11
h <- 5 # w = 2*h + 1
# get slurm job task ID
id <- as.integer(Sys.getenv('SLURM_ARRAY_TASK_ID'))
#id <- 0
set.seed(id)
# fits model for 10 time points per job
# store parameters (lambdas, phi)
theta <- matrix(NA, nrow=10, ncol=ncol(A)+1)
for (i in 1:10) {
cat(i, " ")
t <- id + i
# ensure the time point corresponds to a valid 11-unit time window
if (t >= h+1 & t <= nrow(Y)-h) {
data <- Y[(t-h):(t+h), ] # determine time window
res <- locally_iid_EM(data, c, A)
theta[i, ] <- c(res$lambda, res$phi)
}
}
# remove empty rows in theta - the first job and the last job will have some empty rows
theta <- theta[rowSums(is.na(theta)) == 0, ]
# export theta to csv
write.csv(theta, paste("theta_2router_", id, ".csv", sep=""), row.names=F)
###### 1.9 fit refined model to 2-router data ######
###### ONLY THIS PART RUN ON ODYSSEY ######
# A is 15 x 64 matrix
if (file.exists("matrixA_2router.csv")) {
A <- as.matrix(read.csv("matrixA_2router.csv"))
} else {
A <- matrix(c(rep(0, 56), rep(1, 8),
rep(0, 48), rep(1, 8), rep(0, 8),
rep(0, 40), rep(1, 8), rep(0, 16),
rep(0, 32), rep(1, 8), rep(0, 24),
rep(0, 24), rep(1, 8), rep(0, 32),
rep(0, 16), rep(1, 8), rep(0, 40),
rep(0, 8), rep(1, 8), rep(0, 48),
rep(1, 8), rep(0, 56),
as.numeric(1:64 %in% seq(1, 64, 8)),
as.numeric(1:64 %in% seq(2, 64, 8)),
as.numeric(1:64 %in% seq(3, 64, 8)),
as.numeric(1:64 %in% seq(4, 64, 8)),
as.numeric(1:64 %in% seq(5, 64, 8)),
as.numeric(1:64 %in% seq(6, 64, 8)),
as.numeric(1:64 %in% seq(7, 64, 8))), nrow=15, byrow=T)
write.csv(A, "matrixA_2router.csv", row.names=F)
}
# data Y
if (file.exists("matrixY_2router.csv")) {
Y <- as.matrix(read.csv("matrixY_2router.csv"))
} else {
Y <- matrix(df2[1:15, "value"], nrow=1)
for (i in seq(17, nrow(df2), 16)) {
Y <- rbind(Y, df2[i:(i+14), "value"])
}
write.csv(Y, "matrixY_2router.csv", row.names=F)
}
dim(Y)
head(Y)
# empirically derive prior var-cov matrix V
if (file.exists("matrixV_2router.csv")) {
V <- as.matrix(read.csv("matrixV_2router.csv"))
} else {
res <- NA
for (i in seq(0, 287, 10)) {
if (i == 0) {
mat <- read.csv(paste("theta_2router_", i, ".csv", sep=""))
} else {
mat <- rbind(mat, read.csv(paste("theta_2router_", i, ".csv", sep="")))
}
}
V <- cov(mat)
write.csv(V, "matrixV_2router.csv", row.names=F)
}
c <- 2
w <- 11
h <- 5 # w = 2*h + 1
# get slurm job task ID
id <- as.integer(Sys.getenv('SLURM_ARRAY_TASK_ID'))
#id <- 0
set.seed(id)
# fit model to 10 time points per job
# store parameters (lambdas, phi)
theta <- matrix(NA, nrow=10, ncol=ncol(A)+1)
for (i in 1:10) {
cat(i, " ")
t <- id + i
# ensure the time point corresponds to a valid 11-unit time window
if (t >= h+1 & t <= nrow(Y)-h) {
data <- Y[(t-h):(t+h), ] # determine time window
res <- smoothed_EM(data, c, A, V)
theta[i, ] <- c(res$lambda, res$phi)
}
}
# remove empty rows in theta - the first job and the last job will have some empty rows
theta <- theta[rowSums(is.na(theta)) == 0, ]
# export theta to csv
write.csv(theta, paste("smoothed_theta_2router_", id, ".csv", sep=""), row.names=F)
###### ====================================================================
###### 1.9 replicate figure 6 to 2-router data ######
# estimates from refined model is unavailable
# plot only theta prior
thetaPrior <- NA
for (i in seq(0, 287, 10)) {
if (i == 0) {
thetaPrior <- read.csv(paste("theta_2router_", i, ".csv", sep=""))
} else {
thetaPrior <- rbind(thetaPrior, read.csv(paste("theta_2router_", i, ".csv", sep="")))
}
}
# add 5 empty rows to top and bottom (because I only estimated time points 6-282)
nrow(thetaPrior)
thetaPrior <- rbind(matrix(NA, nrow=5, ncol=65), thetaPrior, matrix(NA, nrow=5, ncol=65))
nrow(thetaPrior)
nodes <- c("router5", "r4-local", "switch", "r4-others", "gw1", "gw2", "gw3", "gw-others")
# create matrix 'names' in the same order as the labels in figure 9
names <- matrix(paste("router5 >", nodes), nrow=1)
for (n in nodes[-1]) {
names <- rbind(paste(n, ">", nodes), names)
}
names
pdf("xiaolinzhuo_fig6_2router.pdf")
par(mfrow=c(8, 8), mai=c(0, 0, 0, 0), oma=c(5, 5, 5, 4))
for (i in 1:64) {
plot(thetaPrior[, i], type="l", ylim=c(0, 1e5), main="", xlab="", ylab="", xaxt="n", yaxt="n")
# add x axis
if (i > 56) {
if (i %% 2 == 0) axis(1, at=seq(0, 288, 48), labels=seq(0, 24, 4))
else axis(1, at=seq(0, 288, 48), labels=NA)
}
# add y axis
if (i %% 8 == 1) {
if (i %in% c(9, 25, 41, 57)) axis(2, at=seq(0, 1e5, 2e4), las=2, labels=c("0", "20K", "40K", "60K", "80K", "100K"))
else axis(2, at=seq(0, 1e5, 2e4), las=2, labels=NA)
}
# add title
text(140, 0.95e5, as.vector(names)[i], font=2, cex=0.6)
}
# add main X axis
mtext("hour of day", side=1, outer=T, line=3, cex=1.1)
# add main Y axis
mtext("bytes/sex", side=2, outer=T, line=3, cex=1.1)
dev.off()