How to get thread # from a C++ AMP tiled index? - c++-amp

Basically, the only reason I'm using a tiled index is to get access to its barrier functions. The input data is 1 dimensional so I'm doing it like this:
parallel_for_each((BitM.extent / 2).tile<1>(), [=](tiled_index<1> TID) restrict(amp) {...
I need 1 thread per 2 units of input. Anyway, I think I can only get the tile # out of the tiled_index which in this case is always 0. The threads need their unique thread ID for the algorithm to run. Where can I get it from?

Related

Add missing/extra values in data array in Matlab

I have recorded WiFi CSI sensor data 5000 packets in 5 seconds(5000 packets x 57 subcarriers). But due to dynamic hardware configuration sometimes I only receive 4998 x 57. I want to add and estimate 2 rows so that my original design has consistent 5000 rows x 57 columns.
As you can see some data are 5000x57, and some are 4998x57.
You can achieve your desired output using mean()-function combined with the concatenation operator [] and the repmat() like this:
A=randi(100,4998,57);
A=[A;repmat(mean(A),2,1)];
Most of the functions in Matlab that take arrays as an input will calculate for each column except if the input array hast just 1 row. So does the mean function and you can just append means output to your arrays.
If you show me the code that you used to import the data, I might be able to help you create a cleaner data structure and thus be able to automatically process all of your arrays. The way the data is currently designed it's only possible to do this with dynamic variable names which is considered bad programming practice.

(matlab matrix operation), Is it possible to get a group of value from matrix without loop?

I'm currently working on implementing a gradient check function in which it requires to get certain index values from the result matrix. Could someone tell me how to get a group of values from the matrix?
To be specific, for a result matrx res with size M x N, I'll need to get element res(3,1), res(4,2), res(1,3), res(2,4)...
In my case, M is dimension and N is batch size and there's a label array whose size is 1xbatch_size, [3 4 1 2...]. So the desired values are res(label(:),1:batch_size). Since I'm trying to practice vectorization programming and it's better not using loop. Could someone tell me how to get a group of value without a iteration?
Cheers.
--------------------------UPDATE----------------------------------------------
The only idea I found is firstly building a 'mask matrix' then use the original result matrix to do element wise multiplication (technically called 'Hadamard product', see in wiki). After that just get non-zero element out and do the sum operation, the code in matlab should look like:
temp=Mask.*res;
desired_res=temp(temp~=0); %Note: the temp(temp~=0) extract non-zero elements in a 'column' fashion: it searches temp matrix column by column then put the non-zero number into container 'desired_res'.
In my case, what I wanna do next is simply sum(desired_res) so I don't need to consider the order of those non-zero elements in 'desired_res'.
Based on this idea above, creating mask matrix is the key aim. There are two methods to do this job.
Codes are shown below. In my case, use accumarray function to add '1' in certain location (which are stored in matrix 'subs') and add '0' to other space. This will give you a mask matrix size [rwo column]. The usage of full(sparse()) is similar. I made some comparisons on those two methods (repeat around 10 times), turns out full(sparse) is faster and their time costs magnitude is 10^-4. So small difference but in a large scale experiments, this matters. One benefit of using accumarray is that it could define the matrix size while full(sparse()) cannot. The full(sparse(subs, 1)) would create matrix with size [max(subs(:,1)), max(subs(:,2))]. Since in my case, this is sufficient for my requirement and I only know few of their usage. If you find out more, please share with us. Thanks.
The detailed description of those two functions could be found on matlab's official website. accumarray and full, sparse.
% assume we have a label vector
test_labels=ones(10000,1);
% method one, accumarray(subs,1,[row column])
tic
subs=zeros(10000,2);
subs(:,1)=test_labels;
subs(:,2)=1:10000;
k1=accumarray(subs,1,[10, 10000]);
t1=toc % to compare with method two to check which one is faster
%method two: full(sparse(),1)
tic
k2=full(sparse(test_labels,1:10000,1));
t2=toc

Periodic to incremental (data reduction)

It's some time I was thinking about solving this problem. I have a registration of angular data (Angle(~20000,1)) variating between 0 and 355 (a potentiometer attached to a rotary testing machine), and I wanted to convert it in an incremental form, since I want the final total angular displacement. The main issue is that between 355 and the next 0 there are no jumps but a fast decrement (with strongly negative slope in time vs angle space). I've tried 2 ways up to now:
Calculate the Angslope=diff(Angle), extract with find the indexes j1=find(Angslope>0.2 & Angslope<0.2) to avoid the negative slopes due to the inversion of angular signal, then try to apply those indexes to the original Angle(n,1), as Angle2=Angle(j1). The trouble is the n-1 length of Angslope and the fact that somehow there is not a simple shift of my indexes of one position.
For cycles and logical, wanting to exclude data if the previous one is < the current value,etc
Angle2=zeros(size(Angle,1),1);
for i=2:size(Angle,1)
if Angle(i,1)<Angle(i-1,1)
Angle2(i,1)=NaN;
else Angle2(i,1)=Angle(i,1);
end
end
Which works good, but I don't know how to "match up" the single increment steps I obtain!
Any help or simple comment would be of great help!!
You are probably looking for the unwrap function. For this you have to convert your angles into radians, but that's not a big deal.
You can get the increments in one line:
Inc = diff(unwrap(Angle*pi/180))*180/pi;
and your total angular displacement:
Tot = sum(Inc);
Best,

postgresql compute min value of columns conditiong on a value of other columns

can I do this with the standard SQL or I need to create a function for the following problem?
I have 14 columns, which represent 2 properties of 7 consecutive objects (the order from 1 to 7 is important), so
table.object1prop1, ...,table.object1prop7,table.objects2prop2, ..., table.objects2prop7.
I need compute the minimum value of the property 2 of the 7 objects that have smaller values than a specific threshold for property 1.
The values of the property 1 of the 7 objects take values on a ascending arithmetic scale. So property 1 of the object 1 will ever be smaller than property 2 of the objects 1.
Thanks in advance for any clue!
This would be easier if the data were normalized. (Hint, any time you find a column name with a number in it, you are looking at a big red flag that the schema is not in 3rd normal form.) With the table as you describe, it will take a fair amount of code, but the greatest() and least() functions might be your best friends.
http://www.postgresql.org/docs/current/interactive/functions-conditional.html#FUNCTIONS-GREATEST-LEAST
If I had to write code for this, I would probably feed the values into a CTE and work from there.

Finding 5 consecutive successes using Matlab?

I have a function which for 10 cycles finds the difference between individual sensor values and the average sensor value. The test will be done 100 times using this function. So every time cycle>10 I am forcing it to be zero so that in the 11th repetition it will restart counting from zero. Here is the code:
cycle=cycle +1;
if cycle>10
cycle=0;
end
for i=1: TotalnoOfGrids
for j=1: noOfNodes
if abs(char(Allquants{i}(j))-char(mostCommonLetters {i}))>0
if cycle>0
wrong{i}(j)=wrong{i}(j)+1;
else
wrong{i}(j)=0;
end
end
end
end
Now I need to know if the sensor performed 5 consecutive successes in the period of 10 cycles. How can I do that?
I thought of a loop but I read that it takes too much time.
Doing a search on the net I have found this SO question.
The problem is that the above function will be repeated for 100 cycles.I want for every 10 cycles see if there is consecutive successes so it is beeing done dynamically and I am not saving the success or failure status of the sensor for the cycles. So i do not have a vector containing 1 or 0 to use the function used in the above reference or as Jonas suggested
If a loop is the easiest thing, give it a try! Just because you've read it "takes too much time" doesn't mean it really makes a difference for your case! It is true that in Matlab it often makes sense to avoid loops; but in your case, 100*20*9 (if I understand you correctly) loop iterations doesn't seem so bad yet (depending on your speed requirement).
Edit (corrected answer)
I now understand from your comments that the code you show us is surrounded by a while or for loop which is being run ~100 times, and that Allquants and mostCommonLetters probably change inside that loop. In this case my previous answer didn't work for you, since it counted successes on different sensors; this should be better now.
If I read your code correctly, the condition abs(char(Allquants{i}(j))-char(mostCommonLetters {i}))>0 tells you that a result was "wrong"; consequently,
for i=1:TotalnoOfGrids
this_cycle_successes(i,:)=char(Allquants{i})==char(mostCommonLetters{i});
end
consecutive_successes=(consecutive_successes+1).*this_cycle_successes;
would calculate how many successes you had in a row. Note you need to initialize consecutive_successes before starting your cycle loop, e.g.
consecutive_successes = zeros(9,20);
After the 10 cycles, you can check which sensors had 5 successes like this:
has5successes = consecutive_successes>=5;
Note that this is a matrix operation, so now you will get 9*20 values, as you requested in your comment. This solution wouldn't require a loop over j.