iterating a function within itself

Hey everyone. I have a function which performs adjustment based on 2 indices NPI and CPI as shown below.

[A,B,C,D,NPI,CPI]=Adjustingcode2(fdrpath1,sheetname,RepCondpath);

CPI and NPI are a 1 by 7 matrices. The optimal value in any of the index of CPI and NPI must be less than 1. So to do this, I need to iterate the function by itself conditionally - only stopping at the first instance where CPI and aNPI are <1. I have thus created an iterate function to do this as follows. However, the final values I obtain for the NPI are not less than 1. Anyone can point out where the problem is?

function [A1,B1,C1,D1,NPI_new,CPI_new] =CSS_iterate(fdrpath1,sheetname,RepCondpath)
[A,B,C,D,NPI,CPI]=Adjustingcode2(fdrpath1,sheetname,RepCondpath);
for i=1:length(R)
    if NPI_updated(i) > 1
       [A,B,C,D,NPI,CPI]=Adjustingcode2(fdrpath1,sheetname,RepCondpath);
    end
    if CPI_updated(i) > 1
        [A,B,C,D,NPI,CPI]=Adjustingcode2(fdrpath1,sheetname,RepCondpath);
    end
end

[Walter Roberson]

Every time you call Adjustingcode2 you are calling it with exactly the same pararameters. You are looping over NPI_updated values but you run the same thing every time the condition is met. You might even run the exact same thing twice, if both conditions are met.

[Lewis Waswa]

Thank you @Walter Roberson and @dpb. Would you please have a look at the following - which explains what the Adjustingcode2 function. Please also see the NCL_recon function attached.

function[NewRC,NewAmpC,NewR,NewAmp,NPI,CPI,R]=Adjustingcode3(fdrpath1,sheetname,RepCondpath)
tic; % the first 4 outputs could be A, B,C and D.

fname1 =  fdrpath1; sheet = sheetname; ResRange1 = "G6:G12"; AmpRange="AG6:AG12"; 
fdrcon_range="AK6:AM60";fdrcon = xlsread(fdrpath1, sheetname,fdrcon_range);
[alphaN,betaN,V_feeder,fdr_HC,VI_out,VA,VB,VC,VUB,IA,IB,IC]= start_f7(1,1,1,1,fdrcon,sheetname,fdrpath1);
R=xlsread(fname1,sheet,ResRange1); Amp=xlsread(fname1,sheet,AmpRange);
% Branch and node performance - calculating replacement

[Vmax,Imax]=BNPerf(VA,VB,VC,IA,IB,IC);
[NPI,CPI,OptiCR_1, OptiVR_2] = NCL_recon(Vmax,Imax,fdrpath1,sheetname);
fname2 = RepCondpath; sheet1 = 3;Range = "G2:G15";
TAmpRange = "I2:I15";  TR=xlsread(fname2,sheet1,Range); TAmp= xlsread(fname2,sheet1,TAmpRange);
for i=1: size(R, 1)
    for j= 1:size(TR, 1)
        if CPI(i) < 1
            NewRC(i) = R(i); NewAmpC(i) = Amp(i);
            
        elseif CPI(i) > 1
            if TR(j) <= OptiCR_1(i)
                NewRC(i) = TR(j);NewAmpC(i) = TAmp(j);
                break;
            end
        end
    end
end
for i=1: size(R,1)
    for j=1:size(TR,1)
        if NPI(i)<=1
            NewR (i)= NewRC(i); NewAmp (i) =NewAmpC(i);
        elseif NPI(i)>1
            if TR(j)<=OptiVR_2(i)                     % original was R
                NewR(i)= TR(j); NewAmp(i) =TAmp(j);
                break
            end
        end
    end
end
xlswrite(fname1,NewR.',sheet,ResRange1)
xlswrite(fname1,NewAmp.',sheet,AmpRange)
toc

A few things to note about your earlier statement: the argument values, that is the values on the fdrpath1 are updated everytime I write /update the excel file with the NewR and NewAmp.

This code runs well. However, the objective is to get an NPI and CPI values that are less than 1. This means that I have to iterate this program to achieve that. This is how I came to my trying to do it using the CSS iterate code above.

I can manually run this code while observing the NPI and CPI and get the right values. How do I automate that process?

[Voss]

@Lewis Waswa: Can you also upload the functions start_f7 and BNPerf (and any other functions necessary for us to run the code) as well as the necessary data files?

[Lewis Waswa]

@Voss, here is the BNPerf function, The input file is updated based on the valued from the ACSR file, which is also pointed by the RepCondpath. I was avoiding providing the whole sets of files associated with start function since they are many - but please see the zip with the rest.

[Voss]

@Lewis Waswa: How about start_f7?

Adjustingcode3('Inputs_JP.xlsx','JP_fdr','ACSRtechnicalData.xlsx')

Unrecognized function or variable 'start_f7'.

Error in solution>Adjustingcode3 (line 9)
[alphaN,betaN,V_feeder,fdr_HC,VI_out,VA,VB,VC,VUB,IA,IB,IC]= start_f7(1,1,1,1,fdrcon,sheetname,fdrpath1);

function[NewRC,NewAmpC,NewR,NewAmp,NPI,CPI,R]=Adjustingcode3(fdrpath1,sheetname,RepCondpath)
tic; % the first 4 outputs could be A, B,C and D.

fname1 =  fdrpath1; sheet = sheetname; ResRange1 = "G6:G12"; AmpRange="AG6:AG12"; 
fdrcon_range="AK6:AM60";fdrcon = xlsread(fdrpath1, sheetname,fdrcon_range);
[alphaN,betaN,V_feeder,fdr_HC,VI_out,VA,VB,VC,VUB,IA,IB,IC]= start_f7(1,1,1,1,fdrcon,sheetname,fdrpath1);
R=xlsread(fname1,sheet,ResRange1); Amp=xlsread(fname1,sheet,AmpRange);
% Branch and node performance - calculating replacement

[Vmax,Imax]=BNPerf(VA,VB,VC,IA,IB,IC);
[NPI,CPI,OptiCR_1, OptiVR_2] = NCL_recon(Vmax,Imax,fdrpath1,sheetname);
fname2 = RepCondpath; sheet1 = 3;Range = "G2:G15";
TAmpRange = "I2:I15";  TR=xlsread(fname2,sheet1,Range); TAmp= xlsread(fname2,sheet1,TAmpRange);
for i=1: size(R, 1)
    for j= 1:size(TR, 1)
        if CPI(i) < 1
            NewRC(i) = R(i); NewAmpC(i) = Amp(i);
            
        elseif CPI(i) > 1
            if TR(j) <= OptiCR_1(i)
                NewRC(i) = TR(j);NewAmpC(i) = TAmp(j);
                break;
            end
        end
    end
end
for i=1: size(R,1)
    for j=1:size(TR,1)
        if NPI(i)<=1
            NewR (i)= NewRC(i); NewAmp (i) =NewAmpC(i);
        elseif NPI(i)>1
            if TR(j)<=OptiVR_2(i)                     % original was R
                NewR(i)= TR(j); NewAmp(i) =TAmp(j);
                break
            end
        end
    end
end
xlswrite(fname1,NewR.',sheet,ResRange1)
xlswrite(fname1,NewAmp.',sheet,AmpRange)

toc
end

[dpb]

...
for i=1: size(R, 1)
    for j= 1:size(TR, 1)
        if CPI(i) < 1
            NewRC(i) = R(i); NewAmpC(i) = Amp(i);
            
        elseif CPI(i) > 1
            if TR(j) <= OptiCR_1(i)
                NewRC(i) = TR(j);NewAmpC(i) = TAmp(j);
                break;
            end
        end
    end
end
...

Just some comments w/o yet being able to run it...

1. the above if...else...end clause is incomplete in that it leaves out CPI==1. That may never occur, but there is no clause for it if it does.
2. NewRC(i) = R(i); NewAmpC(i) = Amp(i); is independent of j, it could be moved out of the loop on j and then only do the test for the alternative.
3. What if ~any(TR<=OptiCR)? Again it may not occur, but is it guaranteed to be true?

[Lewis Waswa]

I have added the rest of the code should any of you want to take run the same.

[Voss]

@Lewis Waswa: Thank you for the code. I've run Adjustingcode2 and Adjustingcode3, with my best guess of what the inputs should be, and both return NPI and CPI values that are all less than 1. See below:

[NewRC,NewAmpC,NewR,NewAmp,NPI,CPI,R] = Adjustingcode2('Inputs_JP.xlsx','JP_fdr','ACSRtechnicalData.xlsx');

Feeder under analysis: Excel_feeder
This feeder is Voltage Constrained!!
FeederMD = 6.183073e+01 kVA
Feeder under analysis: Excel_feeder
This feeder is Voltage Constrained!!
FeederMD = 6.183073e+01 kVA

Elapsed time is 2.440178 seconds.

close all
all(NPI < 1)

ans = logical
   1

all(CPI < 1)

ans = logical
   1

[NewRC,NewAmpC,NewR,NewAmp,NPI,CPI,R] = Adjustingcode3('Inputs_JP.xlsx','JP_fdr','ACSRtechnicalData.xlsx');

Feeder under analysis: Excel_feeder
This feeder is Voltage Constrained!!
FeederMD = 6.183073e+01 kVA

close all
all(NPI < 1)

ans = logical
   1

all(CPI < 1)

ans = logical
   1

What do I need to do to get some CPI and/or NPI values >= 1? In other words, how do I reproduce the problem you asked about?

Answer 1

[dpb]

Several issues here...

function [A1,B1,C1,D1,NPI_new,CPI_new] =CSS_iterate(fdrpath1,sheetname,RepCondpath)
[A,B,C,D,NPI,CPI]=Adjustingcode2(fdrpath1,sheetname,RepCondpath);
for i=1:length(R)
    if NPI_updated(i) > 1
       [A,B,C,D,NPI,CPI]=Adjustingcode2(fdrpath1,sheetname,RepCondpath);
    end
    if CPI_updated(i) > 1
        [A,B,C,D,NPI,CPI]=Adjustingcode2(fdrpath1,sheetname,RepCondpath);
    end
end

[A,B,C,D,NPI,CPI]=Adjustingcode2(fdrpath1,sheetname,RepCondpath);
for i=1:length(R)
    if NPI_updated(i) > 1

1. R is undefined
2. NPI/CPI_updated are never calculated; you return the (so it says) adjusted values in the NPI/CPI arrays, but then test a variable that never changes.
3. if() on and array is true, IF AND ONLY IF all elements of the array are true so the above tests will always be called until all elements in the array <=1. Whether that would be true if were testing the NPI and CPI values instead would all depend upon what the adjusting code actually does which we are not privy to.

That's for starters, at least. Undoubtedly there will be more if/when get enough to be able to test...

Answer 2

[Walter Roberson]

Suppose you have the task of moving one brick at a time to the left until you reach the edge, and you start at brick #4. Then consider the following potential recursive solution:

function move_left_ver1(bricknum)
   move_left_ver1(bricknum);
   bricknum = bricknum - 1;
end

now we call move_left_ver1(4) and the first thing it does is call move_left_ver1(4) . And the first thing that does is call move_left_ver1(4) ... It cannot ever get a chance to subtract 1 from the brick number until all of the calls to move_left_ver1 are finished... but they never finish.

If you have a recursive program and the function calls itself with the exact same parameter values that were passed into it, the only way the program can stop running is if the program tests something external, such as a random number generator, or such as looking for the user to press a Cancel button.

Now let us consider

function move_left_ver2(bricknum)
   bricknum = bricknum - 1;
   move_left_ver2(bricknum);
end

We call move_left_ver2(4) and it subtracts 1 making bricknum into 3, and calls move_left_ver2(3) -- which is a different input parameter, so there is hope so far. move_left_ver2(3) will call move_left_ver2(2) which will call move_left_ver2(1) . Then move_left_ver2(1) will decrement bricknum to 0, and will call move_left_ver2(0) which will call move_left_ver2(-1) then -2 then ... and it will keep going until MATLAB decides you've probably made a mistake.

If you have a program that calls itself and you've made sure to call with different parameter values... you need some way of detecting that you have finished.

... and your test for being finished has to be before the recursive call. The chain of calls has to end somewhere!