1

It is clear that it is preferable to put your source code in a version control system e.g. SVN. When a source is compiled, you get an executable.

Different versions of the executable and various different input parameters (which in my case ~1000) and other input data is proving to be hard to manage. On top of this I make hundreds of simulations on many different computers. All this complexity will add up in future.

Right now I use an excel file but it is proving to be an incomplete method. It is not clear is how do I keep track of my simulations and the results. What I mean is the need of an automated "diary" or a catalogue "Library like itunes" where I can catalog, manage and organize them across computers. How do you manage this?

Also, can someone suggest an appropriate tag for this question?

  • Are the parameters always the same, or do they vary greatly between different versions of the software? – Kasper van den Berg Aug 26 '16 at 19:58
  • Many change, others don't. They are in many different files. Each run generates gigabytes of data – wander95 Aug 26 '16 at 20:52
1

You might be able to database. Before running the simulation the simulator (or a different application) connects to your central database server and stores the simulation parameters and any data that allows you to retrieve it (such as time of running, tags, categories, a reference to the hardware platform, a reference to the binary or source code version the simulation is ran on, relations to other simulation runs). After running the simulation the simulator or a special logger application writes the results (or a summary thereof) to the same database.

If your simulation produces much data and the number of machines you run it on is high, you might look at grid computing (e.g. globus).

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.