fixed minor bugs in post

Author: vlabatut

src/dynamic/narr_smooth.R

@@ -301,7 +301,7 @@ ns.graph.extraction <- function(char.stats, scene.chars, scene.stats, volume.sta
 		}
 		tlog.end.loop(4,"Finished the second character loop")
 	}
-	tlog.end.loop(6,"Finished the first character loop")
+	tlog.end.loop(2,"Finished the first character loop")
 
 	return(res)
 }

src/post/description/avgdist_evol.R

@@ -65,7 +65,7 @@ dist.vals[[2]] <- future_sapply(gs.filt, function(g) mean_distance(graph=g, dire
 # loop over unfiltered/filtered
 tlog(0, "Loop over unfiltered/filtered graphs")
 natures <- c("unfiltered", "filtered")
-pal <- ATT_COLORS_FILT
+pal <- ATT_COLORS_FILT[c("Discard","Keep")]
 for(i in 1:2)
 {	if(i==1)
 	{	filt.txt <- "unfiltered"
@@ -165,7 +165,7 @@ for(fformat in PLOT_FORMAT)
 		x=x, y=y, 
 		xlab=TeX(paste0("Number of vertices $n$")),
 		ylab=TeX("Average distance $<d>$"),
-		las=1, col=pal[1],
+		las=1, col=pal["Discard"],
 		type="l"
 	)
 	# plot fitted line
@@ -201,7 +201,7 @@ for(fformat in PLOT_FORMAT)
 	plot(
 		x=x, y=y, 
 		xlab=NA, ylab=NA,
-		las=1, col=pal[2],
+		las=1, col=pal["Keep"],
 		type="l",
 		cex.lab=0.75, cex.axis=0.75, cex=0.75
 	)

src/post/description/centr_vs_centr.R

@@ -19,7 +19,7 @@ start.rec.log(text="CentrVsCentr")
 tlog(0, "Plotting centrality vs. centrality")
 
 # plot parameters
-pal <- ATT_COLORS_FILT
+pal <- ATT_COLORS_FILT[c("Discard","Keep")]
 
 # measure names
 centr.names <- c(MEAS_DEGREE, MEAS_BETWEENNESS, MEAS_CLOSENESS, MEAS_EIGENCNTR)

src/post/description/centr_vs_occ.R

@@ -22,7 +22,7 @@ tlog(0, "Plotting centrality vs. occurrences")
 data <- read.corpus.data()
 
 # plot parameters
-pal <- get.palette(2)
+pal <- ATT_COLORS_FILT
 
 # measure names
 centr.names <- c(MEAS_DEGREE, MEAS_BETWEENNESS, MEAS_CLOSENESS, MEAS_EIGENCNTR)
@@ -110,7 +110,7 @@ for(centr.name in centr.names)
 #		summary(fit)
 
 		# plot unfiltered data
-		col <- pal[1]
+		col <- pal["Discard"]
 		col.sec <- combine.colors(col, "WHITE", transparency=20)
 		xlab <- paste0("Number of ",occ.proper.names[occ.name],"s")
 		ylab <- NODE_MEASURES[[centr.name]]$cname
@@ -183,7 +183,7 @@ for(centr.name in centr.names)
 #		summary(fit)
 
 		# plot as an inset
-		col <- pal[2]
+		col <- pal["Keep"]
 		col.sec <- combine.colors(col, "WHITE", transparency=20)
 		par(
 			fig=inlay.coords[paste0(centr.name,"_",occ.name),], 

src/post/description/char_distr.R

@@ -48,6 +48,8 @@ laws["Filtered-characters-arc"] <- "moderate"
 # distribution plots
 tlog(0,"Producing distribution plots")
 
+pal <- ATT_COLORS_FILT[c("Discard","Keep")]
+
 # loop params
 object <- "characters"
 counts <- c("panel","page","scene","volume","arc")
@@ -72,7 +74,6 @@ for(count in counts)
 	# set params
 	file <- get.path.stats.corpus(object=object, subfold="both", pref=paste0("distrib_",count,"s-by-char"))
 	tlog(4,"Producing files ",file)
-	pal <- get.palette(length(data))
 	ml <- paste0("Distribution of ",count," number over characters")
 	xl <- paste0("Number of ",count,"s by character")
 

src/post/description/char_sim.R

@@ -18,7 +18,7 @@ start.rec.log(text="CharSim")
 ################################################################################
 # main parameters
 wide <- TRUE				# wide plots showing volumes as rectangles
-narr.smooth <- TRUE			# whether to use narrative smoothing or cumulative scene-based networks
+narr.smooth <- FALSE			# whether to use narrative smoothing or cumulative scene-based networks
 weighted <- TRUE			# whether to use the graph weights
 sc.lim <- NA				# limit on the considered scenes (NA for no limit)
 pub.order <- TRUE			# whether to use the volume publication vs. story order
@@ -50,17 +50,17 @@ kept <- which(V(g)$Filter=="Keep")
 
 # compute the sequence of scene-based graphs (possibly one for each scene)
 gs <- list()
-w.name <- NA
+w.name <- "none"
 if(narr.smooth)
 {	net.type <- "narr_smooth"
-	gs[["FALSE"]] <- ns.read.graph(filtered=FALSE, remove.isolates=TRUE, pub.order=pub.order)
-	gs[["TRUE"]] <- ns.read.graph(filtered=TRUE, remove.isolates=TRUE, pub.order=pub.order)
+	gs[["unfiltered"]] <- ns.read.graph(filtered=FALSE, remove.isolates=TRUE, pub.order=pub.order)
+	gs[["filtered"]] <- ns.read.graph(filtered=TRUE, remove.isolates=TRUE, pub.order=pub.order)
 	if(weighted)
 		w.name <- "normalized"
 }else
 {	net.type <- "cumulative"
 	tlog(2,"Extracting the sequence of graphs")
-	gs[["FALSE"]] <- extract.static.graph.scenes(
+	gs[["unfiltered"]] <- extract.static.graph.scenes(
 		inter.df=data$inter.df, 
 		char.stats=char.stats, 
 		volume.stats=volume.stats, 
@@ -69,13 +69,13 @@ if(narr.smooth)
 	)
 	# possibly set weights
 	if(weighted)
-	{	gs[["FALSE"]] <- future_lapply(gs[["FALSE"]], function(g) E(g)$weight <- E(g)$Occurrences)
+	{	gs[["unfiltered"]] <- future_lapply(gs[["unfiltered"]], function(g) {E(g)$weight <- E(g)$Occurrences;return(g)})
 		w.name <- "occurrences"
 	}
 
 	# compute the filtered version
 	tlog(2,"Same thing for filtered graphs")
-	gs[["TRUE"]] <- future_lapply(gs[["FALSE"]], function(g) delete_vertices(g, v=intersect(filt.names,V(g)$name)))
+	gs[["filtered"]] <- future_lapply(gs[["unfiltered"]], function(g) delete_vertices(g, v=intersect(filt.names,V(g)$name)))
 }
 
 # scene range
@@ -99,29 +99,29 @@ tlog(0,"Evolution of similarity between pairs of characters")
 
 # similarity measures
 sim.meas <- list()
-sim.meas[["cosine"]] <- list(
+sim.meas[["sim-cosine"]] <- list(
 	bounds=c(0,1),
 	cname="Cosine Similarity",
 	foo=function(a,idx) {sapply(1:nrow(idx), function(r) sum(a[idx[r,1],]*a[idx[r,2],])/sqrt(sum(a[idx[r,1],]^2)*sum(a[idx[r,2],]^2)))}
 )
-sim.meas[["pearson"]] <- list(
+sim.meas[["sim-pearson"]] <- list(
 	bounds=c(-1,1),
 	cname="Pearson Coefficient",
 	foo=function(a,idx) {sapply(1:nrow(idx), function(r) cor(x=a[idx[r,1],], y=a[idx[r,2],]))}
 )
-sim.meas[["euclidean"]] <- list(
+sim.meas[["sim-euclidean"]] <- list(
 	bounds=c(0,NA),
 	cname="Euclidean Distance",
 	foo=function(a,idx) {sapply(1:nrow(idx), function(r) sqrt(sum((a[idx[r,1],]-a[idx[r,2],])^2)))}
 )
-#sim.meas[["regequiv"]] <- list(
+#sim.meas[["sim-regular"]] <- list(					# BEWARE: this one is extremely slow
 #	bounds=c(0,NA),
 #	cname="Regular Equivalence",
 #	foo=function(a,idx) {tmp <- REGE.for(M=a,E=0)$E; sapply(1:nrow(idx), function(r) tmp[idx[r,1],idx[r,2]])}
 #)
 
 # plot parameters
-pal <- ATT_COLORS_FILT
+pal <- ATT_COLORS_FILT[c("Discard","Keep")]
 if(wide)
 {	pw.pdf <- 15; ph.pdf <- 5
 	pw.png <- 2400; ph.png <- 800
@@ -236,18 +236,20 @@ for(m in 1:length(sim.meas))
 
 	#####
 	# plot unfiltered and filtered values on the same plot
-	tlog(4,"Looping over the pairs of vertices")
+	tlog(4,"Plotting both unfiltered and filtered networks at once")
+	
+	tlog(5,"Looping over the pairs of vertices")
 	for(p in 1:nrow(pairs))
-	{	tlog(5,"Processing pair ",pairs[p,1],"--",pairs[p,2]," (",p,"/",nrow(pairs),")")
+	{	tlog(6,"Processing pair ",pairs[p,1],"--",pairs[p,2]," (",p,"/",nrow(pairs),")")
 
 		# set file name
 		pt <- paste0(names(sim.meas)[m],"_pair=", paste0(pairs[p,],collapse="--"), if(wide) "_wide" else "")
 		plot.file <- get.path.stats.topo(net.type=net.type, order=ord.fold, mode="scenes", weights=w.name, meas.name=MEAS_MULTI_NODEPAIRS, filtered="both", suf=pt)
-		tlog(6,"Creating file \"",plot.file,"\"")
+		tlog(7,"Creating file \"",plot.file,"\"")
 
 		# compute data ranges
 		xlim <- range(sc.rg)
-		ylim <- range(c(sim.vals[[as.character(FALSE)]][,p], sim.vals[[as.character(TRUE)]][,p]), na.rm=TRUE)
+		ylim <- range(c(sim.vals[["unfiltered"]][,p], sim.vals[["filtered"]][,p]), na.rm=TRUE)
 		ylim[2] <- ylim[2]*1.1	# add some space for volume names
 
 		# produce file
@@ -269,7 +271,7 @@ for(m in 1:length(sim.meas))
 				draw.volume.rects(ylim, volume.stats)
 			# add line
 			for(filt in c("unfiltered","filtered"))
-			{	if(filt="unfiltered")
+			{	if(filt=="unfiltered")
 					col <- pal["Discard"]
 				else
 					col <- pal["Keep"]
@@ -285,7 +287,7 @@ for(m in 1:length(sim.meas))
 			legend(
 				title="Characters",
 				x="bottomright",
-				fill=pal["Discard","Keep"],
+				fill=pal[c("Discard","Keep")],
 				legend=c("Unfiltered","Filtered")
 			)
 			# close file

src/post/description/comp_vs_edges.R

@@ -40,7 +40,7 @@ ylab <- "Largest component size (proportion of vertices)"
 
 # compute both types of weights
 tlog(0, "Compute values and plot basic figures")
-pal <- ATT_COLORS_FILT
+pal <- ATT_COLORS_FILT[c("Discard","Keep")]
 wts <- c("Duration","Occurrences")
 res <- list()
 for(wt in wts)

src/post/description/comp_vs_vertices.R

@@ -35,7 +35,7 @@ xlab <- "Number of vertices removed"
 
 # compute and plot
 tlog(0, "Compute values and plot basic figures")
-pal <- ATT_COLORS_FILT
+pal <- ATT_COLORS_FILT[c("Discard","Keep")]
 res <- list()
 # remove each vertex iteratively
 tlog(4, "Removing vertices iteratively for the unfiltered graph")
@@ -230,7 +230,7 @@ for(fformat in PLOT_FORMAT)
 		legend(
 			title="Characters",
 			x="topright",
-			fill=pal["Discard","Keep"],
+			fill=pal[c("Discard","Keep")],
 			legend=c("Unfiltered","Filtered")
 		)
 	dev.off()
@@ -261,7 +261,7 @@ for(fformat in PLOT_FORMAT)
 		legend(
 			title="Characters",
 			x="topleft",
-			fill=pal["Discard","Keep"],
+			fill=pal[c("Discard","Keep")],
 			legend=c("Unfiltered","Filtered")
 		)
 	dev.off()

src/post/description/deg_plots.R

@@ -44,6 +44,7 @@ idx.keep <- which(data$char.stats[,COL_FILTER]=="Keep")
 
 # loop params
 meass <- c(MEAS_DEGREE,MEAS_STRENGTH)
+pal <- ATT_COLORS_FILT[c("Discard","Keep")]
 
 # load numbers of occurrences of characters
 file <- get.path.stats.corpus(object="characters", subfold="unfiltered", pref="_char_stats.csv")
@@ -82,7 +83,6 @@ for(meas in meass)
 
 		# set params
 		file <- get.path.stats.topo(net.type="static", mode="scenes", meas.name=meas, weights=if(is.na(wt)) "none" else wt, filtered="both", suf="distrib")
-		pal <- get.palette(length(data))
 		ml <- paste0(ALL_MEASURES[[meas]]$cname, " distribution")
 		if(!is.na(wt))
 			ml <- paste0(ml," (",wt,")")

src/post/description/nei_vs_deg.R

@@ -55,7 +55,7 @@ fit <- nlsLM(cut.nei ~ c1*cut.deg^c2,
 summary(fit)
 
 # plot
-pal <- ATT_COLORS_FILT
+pal <- ATT_COLORS_FILT[c("Discard","Keep")]
 col <- pal["Discard"]
 col.sec <- combine.colors(col, "WHITE", transparency=20)
 xlab <- "Degree $k$"