how to solve this error

I got this error Subscript indices must either be real positive integer or logical

I am tried to removed the range of desired signal from dataset matrix and find the power of that range. then the remaining will be interferance and also i want to find the power of remaining range

the error occured when i want to find the power of interferance

PI=mean(mean(SysSetup.Dataset(:,RI).^2))

this is my code

[Geoff Hayes]

Redflower - so the code is failing at the 7th or 8th line? What can you tell us about RI? i.e. what are the values of this array? Are they positive integers or (as the error message suggests) something else?

[Walter Roberson]

When you use linspace() you should assume that you are going to get back values that are not necesarily positive integers. If you need positive integers, it is safer to use the colon operator, probably in conjunction with floor() or ceil().

Perhaps in your case it would be sufficient to index at round(RI) instead of at RI .

[Redflower]

dear Goff

the error in line 10 and the matlab file which i use to find the result

a

[Walter Roberson]

You attach code but not sysSetup or the other variables needed for us to run the code to test.

[Redflower]

those are all codes

i hope to find the solution

[Walter Roberson]

function ModIdentdataset not defined.

[Redflower]

this is the code i forgot to update it

[Walter Roberson]

It looks like it is necessary to call

  ModDatasetGenpaper;      %to define SysSetup
  Train_cumulantpaper;       %to define net
  something unknown at this point   %to define ModIdentdataset
  AccuVsSIR2cum;              %to do computation

Unfortunately without the code to define ModIdentdataset it is difficult for us to test.

[Redflower]

what you are saying correct

i attached again the all codes. and let me to know if they work

[Walter Roberson]

Okay, finally I can reproduce your error.

You copy the test dataset into RI and you delete the columns given by Rd (a list of positive integers) from that. The result is a 2D numeric array of floating point data with fewer columns that the test dataset.

You then try to index sysSetup.Dataset() using that floating point dataset as indices.

I suspect you want to instead take

RI = setdiff(1:size(Dataset_test,2), Rd);

That is, I suspect you want RI to be the list of remaining indices.

[Redflower]

the code you give is work but again matlab give me an error on line 16

matrix dimensions must be agree

[Walter Roberson]

Ah, yes, if RI is not the same length as Rd then line 16 would trigger an error.

I cannot suggest any solution due to the lack of documentation about what Rd and RI are intended to represent.

[Redflower]

thank you for your response

I try to solve the problem but tell now I didn't get

[Walter Roberson]

I do not know. If I had documentation about what Rd and RI should mean, then I might be able to assist.

[Redflower]

I work on digital modulation classification. I have seven classes each class has 1000 signal. at training phase, I use dataset of 7000 signals to built classifier (these signals without any effect of noise). at testing phase i want to test the classifier when there is interference.Also, i use dataset of 7000 signals.

for more details:

class Range

class 1 ( 2-ASK 1- 1000)

class2 (4-ASK 1001-2000)

class3 (2-PSK 2001-3000)

class4 (4-PSK 3001-4000)

class5 (8-PSK 4001-5000)

class6 (16-QAM 5001-6000)

class7 (64-QAM 6001-7000)

my task is to find the accuracy of classifier ( when only one class is desired and other will be interference)

Rd present the desired class

RI present the interference classes

Each time I must find the accuracy of desired class vs interference classes at different SIR value ( signal to interference ratio)

[Walter Roberson]

On line 16, the reference to RI should likely be a reference to Rd

[Redflower]

yes, that is previous version of code in that cade i find the accuracy of classifier ( when only one class is desired and one class is interfered)

now i want to find the accuracy of classifer when (one class is desired and others will be interference) but I don't know how to modify the code

[Walter Roberson]

It is a little after 1 AM where I am. I could go back to examine your code and predict what bugs you might have introduced when you made unknown changes to your code, and then tell you how to correct each of the possibilities... or I could go to bed and check in the morning to see whether you posted your updated code.

[Redflower]

ok, thank you for your response and sorry for disturbance.

good nigth

[Redflower]

i tried this code and the remaining code still same

but it give me an error in line 16

[Redflower]

i tried this code and the remaining code still same

but it give me an error in line 16

[Walter Roberson]

What are you expecting the linspace() to do for you at that point?

[Redflower]

I tried to to takes more than one class as interference at computation code AccVsSIR2cum ONLY change in

Acc(n)= Test_cumulantpaper(net, SysSetup, 1, [2 6],0)
where 1 is desired class 
and 2 and 6 is interference class

[Walter Roberson]

What do you think the result from that linspace() shoud be?

Are you expecting a series of consecutive integers? Are you expecting a series of integers with a constant gap between them? Are you trying to "sample" a range of integers at as close as you can get to equal partitions? Are you trying to extract the index numbers of all samples belonging to a particular class when the SysSetup information knows where the class starts and how long it is?

[Redflower]

I tried to to extract all sample belonging to particular class when the SysSetup information knows where the class start

for example if i select 2 and 6 [ 2 6]

i expect to take all samples of class 2 from 1001- 2000 and 5001-6000 and put them in one matrix

same if i select 1, 2 ,4 [1 2 4]

i expect to take all sample of class 1 from 1-1000 , 1001-2000 for class 2 and 3001-4000 for class 4 and put all of them in one matrix