I want to generate random numbers within an activation function such that every time the activation function is called a random number is generated. I tried with random.uniform and with tf.random_uniform but it only generates a single random value when it's compiled and it doesn't change anymore. How can I make it update every time?
Funny fact:
When I create a variable using tf.Variable(random.uniform(1,2)) every time the function it's called the value is slightly larger, for instance:
1.22069513798
1.22072458267
1.22075247765
1.22077202797
Edit:
The function is very simple
Function:
def activation(tensor):
alpha = tf.Variable(random.uniform(1,2))
return alpha*tensor,alpha
I will omit all the lines in the neural network, but I simply call it as:
act,alpha = activation(dense_layer+bias)
I later get the value by simply:
[ts,c,alph]=sess.run([train_step,cost,alpha], feed_dict={xi: x_raw, yi: y_raw})
Thanks
Hard to tell without source code, but maybe you are initializing your variable with that random value and reusing same value?
Another possibility:
Related
I'm trying to move an object called "car" via the dat.gui. If the user changes the x value using the dat.gui slider, the car should move along the x-axis of its local coordinate system.
here I have copied the part of the code that is causing me problems:
var m = new THREE.Vector3;
m.copy(car.position);
if (changed.key=='X') car.translateX(changed.value-car.worldToLocal(m).x);
My problem is that the expression in car.translateX always evaluates to the value that is in changed.value. The part after the minus has no effect at all or maybe is permanently 0. I have printed the values with console.log and the values of car.position.x and m change in each step, but the subtraction still delivers in every step only the result that is already in changed.value anyway. Can someone help me and tell me why this happens?
Unfortunately, I am absolutely stuck.
car.worldToLocal(m)
I'm afraid this piece of code makes no sense since car.position (and thus m) already represents the car's position in local space.
Instead of using translateX() you achieve the same result by modifying car.position directly:
car.position.x = changed.value;
I am using jmeter to test ldap.
As part of my test, I want to search for a random uid on each iteration. I did not find a straight forward answer to this. So my idea was to first select a random number 1-200 and saving that number as a variable named uid, that number would correspond to a UDV name.
For example
uid = 2 and 2 = A123456
in my udv list. However when trying to reference this variable in my ldap search filter.I am trying to use
(uid=${${uid}})
in hopes to get the value of the value of uid. However the search results just show this as a string.
<searchfilter>(uid=${${uid}})</searchfilter>
Is there another way to achieve what I am looking for?
Use __V function
${__V(${uid})}
The V (variable) function returns the result of evaluating a variable name expression. This can be used to evaluate nested variable references
${__V(A${N})} - works OK. A${N} becomes A1, and the __V function returns the value of A1
Perhaps the easiest would be going for __RandomFromMultipleVars() function?
Another option is using __V() function, it can combine and evaluate JMeter Variables
I am trying to reduce the size of my data and I cannot make it work. I have data points taken every minute over 1 month. I want to reduce this data to have one sample for every hour. The problem is: Some of my runs have "NA" value, so I delete these rows. There is not exactly 60 points for every hour - it varies.
I have a 'Timestamp' column. I have used this to make a 'datehour' column which has the same value if the data set has the same date and hour. I want to average all the values with the same 'datehour' value.
How can I do this? I have tried using the if and for loop below, but it takes so long to run.
Thanks for all your help! I am new to Julia and come from a Matlab background.
======= CODE ==========
uniquedatehour=unique(datehour,1)
index=[]
avedata=reshape([],0,length(alldata[1,:]))
for j in uniquedatehour
for i in 1:length(datehour)
if datehour[i]==j
index=vcat(index,i)
else
rows=alldata[index,:]
rows=convert(Array{Float64,2},rows)
avehour=mean(rows,1)
avedata=vcat(avedata,avehour)
index=[]
continue
end
end
end
There are several layers to optimizing this code. I am assuming that your data is sorted on datehour (your code assumes this).
Layer one: general recommendation
Wrap your code in a function. Executing code in global scope in Julia is much slower than within a function. By wrapping it make sure to either pass data to your function as arguments or if data is in global scope it should be qualified with const;
Layer two: recommendations to your algorithm
Statement like [] creates an array of type Any which is slow, you should use type qualifier like index=Int[] to make it fast;
Using vcat like index=vcat(index,i) is inefficient, it is better to do push!(index, i) in place;
It is better to preallocate avedata with e.g. fill(NA, length(uniquedatehour), size(alldata, 2)) and assign values to an existing matrix than to do vcat on it;
Your code will produce incorrect results if I am not mistaken as it will not catch the last entry of uniquedatehour vector (assume it has only one element and check what happens - avedata will have zero rows)
Line rows=convert(Array{Float64,2},rows) is probably not needed at all. If alldata is not Matrix{Float64} it is better to convert it at the beginning with Matrix{Float64}(alldata);
You can change line rows=alldata[index,:] to a view like view(alldata, index, :) to avoid allocation;
In general you can avoid creation of index vector as it is enough that you remember start s and end e position of the range of the same values and then use range s:e to select rows you want.
If you correct those things please post your updated code and maybe I can help further as there is still room for improvement but requires a bit different algorithmic approach (but maybe you will prefer option below for simplicity).
Layer three: how I would do it
I would use DataFrames package to handle this problem like this:
using DataFrames
df = DataFrame(alldata) # assuming alldata is Matrix{Float64}, otherwise convert it here
df[:grouping] = datehour
agg = aggregate(df, :grouping, mean) # maybe this is all what you need if DataFrame is OK for you
Matrix(agg[2:end]) # here is how you can convert DataFrame back to a matrix
This is not the fastest solution (as it converts to a DataFrame and back but it is much simpler for me).
How can I have different parameters value defined in .ini file for each repeat in omnet using cmdenv? I have repeat value as 4 and trying to have different value of accidentStart and accidentDuration.
You can't. And shouldn't. The whole point of repetition is that all parameters have the same value, just the RNGs are seeded differently. So you get a different sample of the same distribution for every result value.
What you're looking for are iteration variables.
Something like this:
**.accidentStart = ${100, 200, 350}s
This will generate 3 runs without repetition, and 12 runs with repeat=4.
and if you add
**.accidentDuration = ${duration=300, 450, 600..1800 step 600}s
this will multiply the number of runs by another factor of 5.
By default, iteration variables produce a Cartesian product of their respective assigned sets of values. But there are ways to change this, consult the manual for how.
I am trying to assign multiple codes to existing variables. I am using the syntax below, but it will only assign the first code entered for that hosp.id.number.
Syntax example:
Do if (hosp.id.number=9037) or (hosp.id.number=1058) or (hosp.id.number=11256).
Compute role_EM_communication=10.
Else if (hosp.id.number=9037.
Compute role_EM_communication=11.
End if.
Execute.
hosp.id.number needs to be coded 10 and 11, but it will only code it at 10. Anyway to rephrase so that SPSS will accept 2 or more codes for a variable such as hosp.id.number?
Your role_EM_communication variable is a single variable, but from what you are saying, I think you need it to be a set (for the same record, it could take on more than just one code). So you need to create n variables named role_EM_communication_1 to role_EM_communication_n, where n is the maximum number of codes you estimate will be possible for one record.
For your example, it would translate like this:
create the 2 variables:
vector role_EM_communication_ (2, f2.0).
do the first recode:
if any(hosp.id.number,9037,1058,11256) role_EM_communication_1=10.
very important - execute the recode
exe.
check if the first variable has data, and populate the second variable if true:
if miss(role_EM_communication_1) and any(hosp.id.number,9037) role_EM_communication_1=11.
if ~miss(role_EM_communication_1) and any(hosp.id.number,9037) role_EM_communication_2=11.
exe.