wave_filter=function(ts, sp, lp){
########## ts is the time series to be filtered. input as a vector - 1 column
########## sp is the smallest period you want to include in the filtered ts - e.g. 14 day period...
########## lp is the largest period you want to include in the filtered ts e.g. 45 day period...
########## note: lp must be larger than sp

source("wavelet_function_r.txt")


XD=ts-mean(ts)

n=length(ts)

DJ=.25	#spacing between discrete frequencies - defaulty 0.25
DT=1	#sampling intterval of time series - for monthly it will be 1/12 = 0.08333
PAD = 1 # default 0, but if set to 1 it will pad with zeros to get Nn upto next higher power of 2
S0 = 2*DT	#smallest scale of wavelet - default is 2*DT  
param = 6

fin=wavelet(XD, DT,PAD,DJ,param)

#run after wavelet function

L=length(fin$scale)

scale_mat=cbind((1:L), fin$scale)

scales=subset(scale_mat[,1], scale_mat[,2]<=lp & scale_mat[,2]>=sp)


ls=length(scales)

##############revisit how many bands to include in the "significant ts"


# frequency band


utd=matrix(ncol=n, nrow=ls)

I=0

for(i in scales[1]:scales[ls]){

I=I+1

utd[I,]=(sqrt(DT)*DJ)*as.real(fin$wave[(i),])/((sqrt(fin$scale[i]))*.776*(pi^(-1/4)))
}


xu=1:n
 for(i in 1:n){
 xu[i]=sum(utd[,i])
 }



LF_T=xu


#write(LF_T, file="Filtered_ts.txt", ncol=1)
return(LF_T)

}