If I have sets:
set F := G1 G2;
and
set PR[G1] := p1 p2;
set PR[G2] := p3 p4;
how to make the set PR in Julia. For each element of F a specific set of elements belong to PR.
Thank you!
Nikola
Use something like
F = [:G1, :G1]
PR = Dict(:G1 => [:p1, :p2], :G2 => [:p3, :p4])
To help more, you will need to provide more information.
Related
I'm curious of model parallelism, and I've read the code from Yaroslav Bulatov. And in that example, we should partition our model (or in tensorflow we called Graph) manually to different partition (left_network & right_network).
So, I was wondering if I have to make partitions manually, what's the simple_placer.cc and graph_partition.cc have done to the whole graph? And I'm still not clear as all.
In my thought(let me know if anythong wrong):
If the graph has 8 partitions(subgraph) which can be seen as 8 jobs, and 4 workers,How the partitions distributed to workers can be done through:
explicit annotations via tf.device(), or
distributed training, tf.train.replica_device_setter()
share the variables across parameter servers, and otherwise put all
ops on the worker device
But how's the graph make partitions?
I want to trace what's the subgraph (op-nodes set) looks like?
Can I dump the result or I need to trace/modified which code file?
Please let me know if any concepts is wrong or vague.
I'm a rookie of these, any opinion is appreciated.
In the code below, is matmul a op-node, would it be partition into
different jobs?
y_ = tf.placeholder(tf.float32, [None, 10])
x = tf.placeholder(tf.float32, [None, 784])
W = tf.Variable(tf.zeros([784, 10]))
b = tf.Variable(tf.zeros([10]))
y = tf.matmul(x, W) + b
You can get the result of the placement algorithm by passing additional options when you call tf.Session.run()
# ...
y = tf.matmul(x, W) + b
sess = tf.Session()
options = tf.RunOptions(output_partition_graphs=True)
metadata = tf.RunMetadata()
sess.run(y, options=options, run_metadata=metadata)
# `metadata` now contains information about what happened during the `run()` call.
for partition in metadata.partition_graphs:
# `partition` is a `tf.GraphDef` representing all the nodes that ran on a single
# device. All nodes in `partition` have the same `device` value.
device = partition.node[0].device
for node in partition.node:
# e.g. print each node or store it in a dictionary for further analysis.
# ...
I am new to SAS would like my searches to be a more granular.
In this example, I would like my output to show the variable SalesPrice and only those with CentralAir (another variable). I would like to show the statistics for another variable. In this case I want to view the OverallQal variable only if the variable is over 7 and the observations have '1FAM' as its BldgType (which is another variable).
I understand my syntax is incorrect. Any guidance would be appreciated. Thank you!
proc means data=MYDATA.AMES_HOUSING_DATA n nmiss p1 p10 q1 mean q3 p90 stddev median;
var SalePrice if (CentralAir = 'Y');
var OverallQual if (OverallQual GT 7 AND BldgType = '1FAM');
run;
Use a WHERE statement (or the WHERE= dataset option) to limit the records that the proc uses. You can only use one WHERE clause per procedure though, so you would need to run it twice to select the two different sets of records. You also might want to use PROC UNIVARIATE to get a summary of your variables' distribution.
proc univariate data=MYDATA.AMES_HOUSING_DATA ;
where CentralAir = 'Y';
var SalePrice;
run;
proc univariate data=MYDATA.AMES_HOUSING_DATA ;
where OverallQual GT 7 AND BldgType = '1FAM' ;
var OverallQual ;
run;
I'm currently using Stata 13.1 to examine a long list of float variables (e.g., A1 - A60). Each of these variables represents the frequency of a different medical symptom (e.g., "Insomnia", "Anxiety", "Nausea"). I'd to add labels to each variable to make data analysis a bit easier, but would prefer something more elegant than:
label var A1 "Insomnia"
label var A2 "Anxiety"
.
.
.
label var A60 "Nausea"
Any suggestions are very much appreciated!
Initially, you need to store the labels in some place. You can use a local macro for that. Below an example with variables that follow some naming pattern (like your example does).
clear
set more off
*----- example data -----
gen A1 = .
gen A2 = .
gen A3 = .
*----- what you want -----
local mylabels "Insomnia Anxiety Nausea"
local n: word count `mylabels'
forvalues i = 1/`n' {
label variable A`i' `:word `i' of `mylabels''
}
describe
The looping over parallel lists technique is from: http://www.stata.com/support/faqs/programming/looping-over-parallel-lists/.
See also help macro and help help extended_fcn.
I wrote code that executes a function it receives from the (future) client, in a loop with some parameters. will call it func(name it).
Inside the function the client usually generate expression in the same variables(by GetUncertainty - each variable must be cleared before use). To do so , the simple idea is to use Block. Later , a code is executed that handles di and i outside the function.So, di and i must be globals(there could be more, it is flexible).
BTW, I know it is not efficient, but efficiency is not an issue.
func[v_, a_, r_] :=
(V = v; A = a; R = r;
Block[{V, A, R},i = V A + A 10 + R 100; di = GetUncertainty[i, {V, A, R}];] ;
Print[di])
The problem is that the client must reset the vars by hand. That means that the function parameters can't be V_,A_,R_ , otherwise The vars in the block will be replace by the values. I didn't manage to overcome this in any other way.
Another question in a similar issue. if I have
vars = {V,A,R,DR} , then
Block[vars , ..code.. ] , throws error that it is not a list.whereas Block[ {V,A,R,DR},..code..] works. How to overcome this?
Thanks.
its hard to unravel what you are trying to do, but the best approach may be to simply never assign values to symbols that need to be used as pure symbols in some context. Then you don't even need the Block[].
func[v_, a_, r_] := (
i = V A + A 10 + R 100;
di = GetUncertainty[i, {V, A, R}];
Print[di /. {V->v,A->a,R->r])
starting your own symbol names with Caps is frowned upon by the way as you risk conflict with built in symbols.
Note also there is a dedicated site mathematica.stackexchange.com
If I understand your application what you need are Formal Symbols. These are a set of Symbols with the Attribute Protected so that they cannot accidentally be assigned a value. They may be entered with e.g. Esc$AEsc for Formal Capital A. You can then use ReplaceAll (short form /.) as george showed to substitute your desired values.
Your code would be something like this:
func[v_, a_, r_] :=
Module[{i, di},
i = \[FormalCapitalV] \[FormalCapitalA] + \[FormalCapitalA] 10 + \[FormalCapitalR] 100;
di = GetUncertainty[i, {\[FormalCapitalV], \[FormalCapitalA], \[FormalCapitalR]}];
di /. {\[FormalCapitalV] -> v, \[FormalCapitalA] -> a, \[FormalCapitalR] -> r}
]
That looks horrible here but in a Notebook it looks like this:
I included Module to show how you should properly localize utility Symbols such as i and di but this particuarly simple function could also be written without them:
Your second question regarding "vars = {V,A,R,DR} then Block[vars , ..code.. ]" is answered here: How to set Block local variables by code?
Dedicated StackExchange site:
I posted this earlier and it got closed because I did not show my try at coding it, so here is the question:
SECTIONS
$160 = section 1
$220 = section 2
$280 = section 3
$350 = section 4
$425 = section 5
Develop pseudocode that accepts as input the name of an unspecified number of masqueraders who each have paid the full cost of their costume and the amount each has paid.
A masquerader may have paid for a costume in any of the five sections in the band. The algorithm should determine the section in which the masquerader plays, based on the amount he/she has paid for the costume. The algorithm should also determine the number of masqueraders who have paid for costumes in each section.
The names of persons and the section for which they have paid should be printed. A listing of the sections and the total number of persons registered to play in each section should also be printed, along with the total amount paid in each section.
Here is my try at it: *Note this is programmed in Pascal, and I need help fixing it up and finishing it. Please help and thanks again.
program Masqueraders;
uses
WinCrt; { Allows Writeln, Readln, cursor movement, etc. }
const
MAX = 5; {this determine the amount of masquarader entered}
Type
listname = Array[1..MAX] of string;
listsect = Array[1..MAX] of string;
var
names : listname;
sections : listsect;
i, amount, TotalMas, TotalAmt, c1, c2, c3, c4, c5, amt1, amt2, amt3, amt4, amt5 : integer;
begin
amount := 1;
while amount <> 0 do
begin
i := i + 1;
readln(names[i]);
readln(amount);
if(amount = 160) then
begin
c1 := c1 + 1; {Count the number of persons for section 1}
amt1 := amt1 + amount; {accumulate the amount for section 1}
sections[i] := 'Section 1';
end;
if(amount = 220) then
begin
c2 := c2 + 1; {Count the number of persons for section 1}
amt2 := amt2 + amount; {accumulate the amount for section 1}
sections[i] := 'Section 2';
end; {end the IF for section 2}
if(amount = 280) then
begin
c3 := c3 + 1; {Count the number of persons for section 1}
amt3 := amt3 + amount; {accumulate the amount for section 1}
sections[i] := 'Section 3';
end; {end the IF for section 3}
if(amount = 350) then
begin
c4 := c4 + 1;
amt4 := amt4 + amount;
sections[i] := 'Section4';
end; {end If for section 4}
if (amount = 425) then
begin
c5 := c5 + 1;
amt5 := amt5 + amount;
sections[i] := 'Section5';
end;{end the while loop}
TotalMas := c1 + c2 + c3;
TotalAmt := amt1 + amt2 + amt3;
writeln('Name Section'); {Heading for the output}
for i := 1 to MAX do
begin
write(names[i]);
writeln(' ',sections[i]);
end;
writeln('Section 1: ');
write('Masquader: ', c1);
write('Amount: ', amt1);
writeln('Total Number of Masquarader: ', TotalMas);
writeln('Total Amount Paid by masquarader: ', TotalAmt);
end;
end.
In short, it should accept an undefined number of people and assign them to their respective sections based on the amount of money they entered, then calculate the number of people in each section. This is my current output:
Name John Money=160 Section 1
Name Keith Money=220 Section John
This is what I want:
Name John Money=160 Section1
Name Keith Money=220 Section2
I'm not really familiar with Pascal so I can't tell you how to fix this but one issue I notice is that your code seems to violate the "DRY" rule: Don't Repeat Yourself. The code inside each if amt = xxx block looks almost exactly the same, so is there a way you can write that code once, and then call your code with different parameters each time? Something so that you're not rewriting the same code five times.
Here's are a few hints to improve your code:
Do you think there might be a way to print "Section {i}" in your final loop without looking it up using sections[i], thereby negating the need for the array completely?
How could you build this so that adding a section 6 wouldn't require modifications to your code?
listname and listsect are awfully similar, what modifications to your code could you do to eliminate the need to have two identical definitions?
What happens if the user enters 0 for the amount the third time they are prompted?
Note: one of these hints should point you directly to the source of your problem.
Here are the problems I see:
The requirements say an "unspecified number of masqueraders" but you have set the max to 5. You can't store the list of names in a fixed size array if there are possibly going to be more than 5 masqueraders. As is, the app may either crash or corrupt memory (depending on the version of Pascal you're using) when user enters the 6th masquerader. If you are allowed to print the masquerader's Name and Section immediately after it is input, then you should print that right after the user inputs the Name and Amount (not after the input while loop). However, if it is required that you print the list of Names and Sections after all input, then you need to store the Name+Section in a variable length data structure like a linked list or even simpler a string.
The variables c1 to c5 and amt1 to amt5 would be better declared as two arrays (c and amt) of 1..MAX instead of 10 variables. When user inputs the amount, use it determine the section number which then can be used as the index into the c and amt arrays.
Not sure what implementation of Pascal you're using but it would be safer to initialize all variables before use otherwise they could contain unpredictable values.
The sections array is unnecessary. All it ends up containing is the title of the section which doesn't need to be calculated or stored.
If you used a repeat/until loop instead of a while loop, it would avoid the slightly kludgy "amount := 1" initialization at the top to enter the while loop. The repeat/until would be until Amount = 0.
TotalMas is only adding c1 to c3 (what about c4 and c5)?
TotalAmt is only adding amt1 to amt3 (what about amt4 and amt5)?
The for-loop to print the names and section is completely wrong, see points 1 and 4.
Hope this helps and let me know if you need clarification on any points.
My gripes with this code:
1) The variable names stink. Virtually all of them are way too short, they do not communicate what they do.
2) The wrong loop control structure is used here.
3) I see virtually identical code repeated 5 times. This is simply crying out for arrays.
4) The comments explain what you are doing, not why you are doing it. Half of them are virtually completely useless, if you're going to comment an end you should simply comment it with the identity of whatever it's an end to. end; //if I see only one comment that even tries to explain why and you've got it wrong as you're mixing up people with sections.
5) Your sections array accomplishes nothing useful. The assignments to the values are always fixed and thus there is no reason to store them at all--if you need the labels you can create them at print time.
6) You are assuming there are only going to be 5 people. That's not given in the problem--there's 5 sections.