Table size limit?

Here is what I am seeing:

>> mnl = readmps('mnl.mps');
??? Error using ==> addtotable at 12
Too many entries in table.

Error in ==> readmps at 150
  rowtable=addtotable(rowtable,f{2},rowcount);

I am trying to use readmps, a file off the matlab exchange for creating linear programming problems in matlab. I have used the program before with no problems. I suspect that the mps file I am trying to read in is just to big. Is this likely the cause? Is there some size limit on tables that I am unaware of?

Answer 1

[Edward Umpfenbach]

Still hoping for an answer. Bump.

Answer 2

[Walter Roberson]

I do not find any readmps routine in the MATLAB File Exchange.

I find lpmexext which appears to have a read_mps command in it. That code appears to have a limit of 100 lps. You might be able to increase that by increasing the value of the LPMAX #define near the beginning of the C source.

[Walter Roberson]

In newtable.m change the 1000 to a larger value.

Answer 3

[Edward Umpfenbach]

I apologize. It must not have been on the exchange. I found it through google:

http://infohost.nmt.edu/~borchers/readmps.html

I will consider lpmexext, but only if I really need to. I could probably deal with C source code and Mex files, but my comfort zone is just m files.

Answer 4

[Jan]

There is no real chance, that somebody can guess, what's going on. We do neither know the function nor the file.

I suggest to use the debugger to stop Matlab, when the problem occurs:

dbstop if error

Then check the value of the local variables. Perhaps rowcount is Inf by accident or a damaged file? Perhaps it is negative and the error message is misleading?

Did you ask the author?

Answer 5

[Edward Umpfenbach]

The reason I didn't do any of what you have suggested is that I have successfully used this file before, no errors. The only thing that changed is that I have a larger mps file. In addition, it ran for approximately a day before I saw the error above. Thus, I assumed that I was hitting some Matlab limit on a data structure size or something.

I will try with the debugger. Thanks for the help.

[Walter Roberson]

The error message you indicate is hard-coded in to addtotable.m and is based upon 90% occupancy of the declared table size, where the declared table size is initialized to 1000 in newtable.m

The error message is _not_ produced as a result of a try/catch around having run out of memory: it is produce by the code detecting that the table it pre-created is full.

[Edward Umpfenbach]

Sounds good. So, just to be clear, open newtable.m, change the 1000 to a larger number, and I should be good?

[Walter Roberson]

Looks like it to me. Or at least _that_ won't be the limit you run in to ;-)