Edit:Value Exceeeds while dividing

I have 16 bit values and 14 bit values of same size

A=16 bit values ,B=14 bit values

i have sorted suxh a way that maximum value of 14 bit comes under max valu of 16 bit and so on,finally minimum of 14 bit comes min of 16 bit

used command as

[~,idx]=sort(A)
[~,idx1]=sort(idx)
C=B(idx1);
Dw=C./Cw

Dw=Dw*100;

the value of Dw exceeds 5 bit ,plese tell why i get this

suppose if max of 14 bit 8192/max of 16 bit 32767 we get 25 ,which is a 5 bit ,then why get more than 5 bit,please help

[Walter Roberson]

What do you see as the point of doing the sort(idx) ? Are you trying to find indices of unique values? If the values are already unique then idx1 is just going to be (1 : length(A))

[Pat]

 A=[120 150 12 10 36 25 ];
 B=[2 9 1 6 8 10];

i need as out=[9 10 2 1 8 6]

i tried using

[ii jj k]=unique(Sw);
B=unique(Cw);
P=B(k)

it works ,but suppose if my 14 and 16 bit consists of unique and negative values, what command must be used

[Walter Roberson]

I do not understand how you get out=[9 10 2 1 8 6] instead of out=[8 10 9 2 6 1] ?

[Pat]

walter max of B is 10 ,so it comes undre max(A) which is 150,2nd position in out,min(B) is 1 is in 4th position in out because min(A) is in 4th position

[Walter Roberson]

Ah, okay I see that ordering now.

What problem do you see if your "14 and 16 bit consists of unique and negative values" ?

[Pat]

if i perform max(14bit)/max(16bit) i should get .25,but i get more value 12.12,dont have an idea of sorting negative and unique values

[Walter Roberson]

sort() and unique() work on negative values as well. Is the concern about the possibility that the max(16bit) might be negative and of small absolute value, but max(14bit) might be positive, leading to a large number divided by a number that is small absolute value?

Can you reproduce the large-result problem with a small demonstration vector that you could post?

[Pat]

A1 =

-443  -221  -266    89   -45  -266  -177  -133  -266  -266  -177   140

B1 =

   -66   -55    33    78   122   111    66    22    22   -33   -78   -78

Answer 1

[Andrei Bobrov]

[~,i1] = sort(A);
B1 = sort(B);
[~,i2] = sort(i1);
C = B1(i2);

[Pat]

Andre i have negative and unique values in it..my max(B) is 1540 and max(32768)

so 1540/32768 is 0.0470

but i get maximum 0.0995 after dividing C./A

Answer 2

[Walter Roberson]

Remember, negative divided by negative gives positive, so if -12 as a 14 bit number happened to get paired with -1 as a 16 bit number, the ratio would be -12/-1 which would be 12.

But you don't even need to take into account negatives for this. If 12 as a 14 bit number happened to get paired with 1 as a 16 bit number, the ratio would be 12 anyhow. Consider for example

14 bit: [12 30]
16 bit: [1  100]

then 12/1 = 12, 30/100 = 1/3, so the ratio of their maximums (30, 100) would be much smaller the internal ratio.

[Pat]

Thanks walter