I have an issue when trying to create jobs for Decision Optimization when using size XL in Watson Machine Learning (WML). The first job for the day I have no issues what so ever to create. But the second job is failing.
If changing to smaller instance (S or M) there is no issue to start a new job. But for size XL I can not start a second job. I can not find out why. Any ideas?
I get the following fault codes:
Code:
error_in_instance_creation
Message:
Instance creation of t-shirt-size XL and type do12.10 failed.
We encountered the same issue; for us it had to to with the facts that we did not properly remove our old models / deployments / jobs. After cleaning up all left-over resources accordingly, we were able to startup an instance again.
Thanks everyone for your support. We have now found out that it seems like IBM has updated the service.
Without doing any changes it is now possible to deploy multiple models in a day and execute the one you are interested in.
So I would say it solved it self
I am a user of Vensim P.L.E. 8.0.4. x64. This is the Personal Learning Edition, that is freely available for 60 days.
I am trying to get familiar with it.
I tried to replicate some examples available online, such as these two:
first video
second video
and I was successful.
Now I would like to include in my simulations not only a scalar, but an array or a vector of numbers. This is because I would like to use a set of real world data I have for different municipalities and it would be of little informative to do it for each of them separately.
I followed the indications provided here:
Link for the use of the function tabbed array
which substantially provides what I would need. Yet, at the very same page I read "Availability: Professional and DSS only." I am thus inclined to think that this is not a function I can have access to.
I tried to specify an equation for one of my variables by making use of TABBED ARRAY, but I got the error message "Expecting an operator" in the relevant section of the window, as reported below.
I would be very grateful if anyone could help me to understand if the error is related to my sintax, or to the Versim version I am using or to other possible reasons.
I understood, but correct me if I am wrong, that there is a way to work with Vensim by coding but I am not sure this possibility is open to the version I am using.
I do thank you very much for your help.
Marco
I think that error message really comes from the version PLE you are using.
You also could try import the constants from Excel using the GET XLS DATA('file','tab','row or col','cell') function [1]. But this function don't work into PLE version too.
So I advise you use Insight Maker in the first steps of modelling, that is general-purpose simulation and modeling tool, open source and that run in your web browser [2]. On this platform you could import data manually using converters objects [3].
Below we see a print screen from the Insight Maker workspace with a converter object to data import.
I am trying to make a c++ program that is able to detect unused permissions in an android application. I was able to do reverse engineering process and get the java classes of the application. Then I used an xml parser to be able to get the list of permissions defined in Android manifest. Now I want to inspect the java classes to check the permissions that weren't used at all. I don't want to find the problem of checking permissions during run time or the permissions in the unreachable part of code. I just want to find the unused permissions that don't require dynamic analysis. If any one of you know a library that takes java classes and enables me to find the permission or an idea about the steps I should follow since googling isn't getting me anywhere. Thank you.
I'd like to use line_profiler to profile a single large method line by line in my GoogleAppEngine application.
Unfortunately GAE doesn't seem to let you import .so libraries, even on a local dev server.
How could I go about achieving my goal?
I'd be happy to use a python-only solution, if there's one out there, or take suggestions as to how to write my own.
Use gae_mini_profiler.
It can either keep track of all function calls and their timings (instrumented) or can periodically examine the call stack to figure out in which functions time is being spent during a request (sampling). You can see an example of it in action here - http://mini-profiler.appspot.com/
I have a number of C functions which implement mathematical formulae. To-date these have been tested for mathematical "soundness" by passing parameters through command line applications or compiling DLL's for applications like Excel. Is there an easy way of doing this testing over the web?
Ideally something along the lines of:
compile a library
spend five minutes defining a web form which calls this code
testers can view the webpage, input parameters and review the output
A simple example of a calculation could be to calculate the "accrued interest" of a bond:
Inputs: current date, maturity date, coupon payment frequency (integer), coupon amount (double)
Outputs: the accrued interest (double)
You should have a look into automated testing. Manual tests will all have to be repeated every time you change something in your code. Automated tests are the solution for your kind of tests. Let testers write test cases with the accompanying results, then make them into unit tests.
See also: unit testing
The quickest thing I can think of is to have these C programs compiled on the server. And create a PHP page that received command-line parameters and then execute compiled program on the server, parsing the output. Technologies other than PHP would also work just fine. What you need to figure out, for specific technology, are:
How to start a process
How to redirect standard input/output
I have also seen number of web site which let users submit their C code and then it get compiled on the server. After that the program will be given some input file and give output. The output of program is then verified with correct answer. For example visit this site, http://acm.timus.ru/
If you're going to do this, you should be sure that every web interaction is captured in a permanent database of tests. Then you can use this database to
Automatically re-run all tests if the software changes
Possibly find inconsistencies that result if a person gives you the wrong answer
In other words, the web form should be the front end to a persistent infrastructure for testing, not a means of running tests that disappear just after they are viewed.
Or similarly, create a Perl CGI that checks input values and then passes them through to the C program. BTW This should only be done for testing and not for final deployment.
You should really automate the testing to check your behaviour is as expected over a wide range of values.
Or shouldn't you be testing this in an environment that is as close as possible to the final deployment environment?
cheers,
Rob
This is what you're looking for:
http://codepad.org/
It will execute C, C++, D, Haskell, Lua, and many others online, and display the results.
If you've got a large library to compile it may get unwieldy, but testing a function briefly is simply a matter of pasting the code and hitting "Submit".
This sounds much like FIT. You could probably make a new fixture for it, or for one of the other language ports like the Python one, that calls a C library with your function. This would take advantage of the work that's gone into making FIT convenient, the kind of work Norman Ramsey recommends in his answer.