Replace values in a array

Hey to all, thanks in advance

Is there a way of: for each row of distance_matrix, for the values that are not zero,find the most aproximate values in the array range with those of that row in distance_matrix. Once we identified in which column the values are most similar to those in the row of distance_matrix, replace those values in range by the values present in each row of RMC2.

Thank you

[Rik]

It sounds like you have the steps clear. What did you try to implement them?

[Miguel Albuquerque]

distance_matrix=round(distance_matrix);
range=round(range);
RMC2=round(RMC2);
for i = 1:size(RMC2,1)
        thisrow=RMC2(i,:);
    for ii= 1:size(distance_matrix,1);
        thisrow2=distance_matrix(ii,:);
        isR2=find(thisrow2);
        thisrow2=distance_matrix(ii,isR2);
        [val,pos]=intersect(range,thisrow2)
        pos=pos.';
        first_element=pos(1,1);
        last_element=pos(1,end);
        value=(last_element-first_element)+1;
        thisrow=RMC2(i,(1:value))
        range(:,(first_element:last_element))=thisrow;

    end
end

I did this, and is working fine, just an error, that its replacing the first iteration, so in column of range(:,253:272) it should be 95, and in 2nd iteration from range(:,272:341) it should be 102. But my code is replacing the first iteration by the second, meaning range(:,253:272)=102 and range(:,272:341)=102, and I dont want that,Can you help me pls?

[dpb]

Still most confusing what you're actually trying to do here -- if I look at the arrays, I find the following --

>> whos distance_matrix range RMC
  Name                 Size              Bytes  Class     Attributes

  RMC                  2x400              6400  double              
  distance_matrix      2x400              6400  double              
  range                1x1600            12800  double              

>> 

but that range is just a fixed delta from 0 so it could be computed on the go.

Figuring out what is inside the other two (got tired typing so named distance, 'd')--

>> sum(d~=0,2)
ans =
    93
    93
>> sum(RMC~=0,2)
ans =
    93
    93
>> 

both have 93 elements that aren't zero in each row -- if you want the new RMC values there to go into distance, why not just assign them from the starting indices of each?

The Q? asked about nearest leads to a conundrum -- since 100 elements are too many to observe easily, I just picked first 10 out of the location of d as

ix=find(d(1,:));   % the first row non zero distances
ir=interp1(range,1:numel(range),d(1,ix(1:10)),'nearest');   % nearest element in range first 10
>> ir
ir =
   254   255   255   256   256   256   257   257   258   258
>> range(ir)                % what are those range values?
ans =
   94.8750   95.2500   95.2500   95.6250   95.6250   95.6250   96.0000   96.0000   96.3750   96.3750
>> d(1,ix(1:10))            % compare to actual distances...
ans =
   94.9718   95.1356   95.3010   95.4681   95.6369   95.8074   95.9797   96.1538   96.3297   96.5074
>>

So, now what? The nearest range to the distance isn't unique -- so which RMC to use -- the duplicates from the ir indices that are nearest? If so, I'm still at a loss as to why you don't just calculate the values for the given distances.

The previous answers re: using logical addressing still hold, only your RMC vector isn't matching because it is full of irrelevant information that I had presumed wouldn't be there is all.

[Miguel Albuquerque]

This is a result from a very complex processing of radar.

Basically I use distance_matrix as the estimated parameters.

RMC2 is a correction to the vector range, given by distance_matrix,

Range- first vector

RMC2- Correction to that vector.

Distance_matrix- Estimated values.

The code I made just gives me this, but im doing an error I cant figure out where it is, because its replacing the 2row of RMC2 by the first row.

So at the end I should have range corrected, and I m getting that but first iteration is being replaced by second one, how can I solve that?

[dpb]

Your Q? are as varying as the radar...they keep moving target.

If you want the nearest value of the RMC vector, then use

for i=size(d,2)
  ir=interp1(range,1:numel(range),d(i,:)),'nearest');
  range(i,range(i,:)~=0)=RMC(i,ir);
end

otherwise it's just

for i=size(d,1)
  range(i,range(i,:)~=0)=RMC(i,find(RMC(i,:));
end

Your above code uses : for row indexing that addresses both rows.

That could/would be ok if the result of ir were the same for both rows and you addressed both rows of RMC and they also overlap.

[Miguel Albuquerque]

Sorry I didnt understand how can I change that in my code, is not there a way I could correct what you said in my code,

[dpb]

That does what you asked for...I'm not trying to decipher/debug yours...

[Miguel Albuquerque]

done

distance_matrix=round(distance_matrix);
range=round(range);
RMC2=round(RMC2);
for i =1:size(RMC2,1)
    thisrow=RMC2(i,:)
    thisrow2=distance_matrix(i,:)
    isR2=find(thisrow2)
    thisrow2=distance_matrix(i,isR2)
    [val,pos]=intersect(range,thisrow2)
    pos=pos.';
    first_element=pos(1,1)
    last_element=pos(1,end)
    value=(last_element-first_element)+1
    thisrow=RMC2(i,(1:value))
    range(1,(first_element:last_element))=thisrow     
end