Simplex Algorithm

[Tommy Vee]

On 9/2/2013 5:55 AM, Robert Kern wrote:
> On 2013-09-02 02:26, Tommy Vee wrote:
>> Anyone know where I can get an easy to use Python class or algorithm
>> for the
>> Simplex optimization algorithm?  I've tried the one in the link below,
>> but I
>> can't figure out if a) I'm using it properly, or b) where to get the
>> solution.
>> BTW, I tried some test scenarios using MS Excel's "Solver" and just
>> can't get
>> this algorithm to match Excel's results (which is spot on).
>>
>> http://taw9.hubpages.com/hub/Simplex-Algorithm-in-Python
>>
>> BTW, if I can't something to work, I'm going to be forced to use the VBA
>> programmatic Excel interface. That wouldn't be too bad, but the data
>> comes from
>> a DB and getting it properly positioned to use Excel's "Solver" is very
>> painful.  A Python approach would be much cleaner.
>
> Can you show some of the test scenarios that you tried? There are
> different conventions in how to represent a linear programming problem,
> and different solvers may choose different conventions. You may have to
> convert between representations.
>
> You may have better luck with the PuLP interface:
>
>    https://pypi.python.org/pypi/PuLP
>
> PuLP itself is just a modelling language rather than a solver, but the
> sources do contain compiled binaries for the CoinMP solver so it will
> work out-of-box on popular platforms, like Windows.
>
>    https://projects.coin-or.org/CoinMP
>

Thank you, I will definitely look at these and other options.  BTW, try 
the test scenario in the link I sent.  Very simple, only 3 variables.

Maximize:  2x+3y+2z

Constraints: 2x+y+z <=4, x+2y+z <=7, z <= 5

The algorithm displays the Tableau after each pivot, but where is the 
answer for x, y and z?

When I run this in Excel's Solver, I get x=0, y=3, z=1. which is indeed 
the maximized solution (11).