IJulia notebooks on using JuMP for optimization

[Shuvomoy]

These are the notebooks I prepared for a tutorial on using JuMP, given at UTORG, University of Toronto.

[mlubin]

We need a little bot to post nbviewer links to PRs :)
http://nbviewer.ipython.org/github/Shuvomoy/juliaopt-notebooks/tree/master/notebooks/

[mlubin]

Thanks for contributing these! We've more than doubled the available content.
A few comments:

• We haven't really settled on naming conventions, but there will certainly be multiple notebooks on column generation, benders decomposition, and "getting started" in the future. I would suggest prefixing the notebook names with UTORG- or Shuvo- or Shuvomoy- for now. I'm hoping to figure out a reasonable way to make the notebooks searchable metadata and searchability #3.
• In the benders decomposition and column generation notebooks, could you replace the use of @defConstrRef with the three-argument version of @addConstraint? E.g.

@defConstrRef consRef[1:cardinalityM] # References for the constraints
for (i in M)
    consRef[i] = @addConstraint(cutstockMain, sum{A[i,j]*x[j], j in Jprime}==b[i])
end

becomes

@addConstraint(cutstockMain, consRef[i=1:cardinalityM], sum{A[i,j]*x[j], j in Jprime}==b[i])

We prefer this form and only recommend @defConstrRef for cases where the more compact syntax is difficult to apply.

• In the "getting started" notebook, you may want to add a note that to use the CPLEX, Gurobi, and Mosek packages, you first need to set up the software and appropriate licences.
• The use of \preceq instead of \leq for element-wise inequality seems a bit unusual to me, although I'm aware that some authors use it. Could you add a note of its meaning?

[Shuvomoy]

Hi Miles,

Thanks. I will make the changes that you suggested and upload them soon.

Best Regards,
Shuvo

On Fri, Jan 30, 2015 at 1:43 PM, Miles Lubin notifications@github.com
wrote:

Thanks for contributing these! We've more than doubled the available
content.
A few comments:

• We haven't really settled on naming conventions, but there will
  certainly be multiple notebooks on column generation, benders
  decomposition, and "getting started" in the future. I would suggest
  prefixing the notebook names with UTORG- or Shuvo- or Shuvomoy- for
  now. I'm hoping to figure out a reasonable way to make the notebooks
  searchable metadata and searchability #3 metadata and searchability #3
  .
• In the benders decomposition and column generation notebooks, could
  you replace the use of @defConstrRef with the three-argument version
  of @addConstraint? E.g.

@defConstrRef consRef[1:cardinalityM] # References for the constraints
for (i in M)
consRef[i] = @addConstraint(cutstockMain, sum{A[i,j]*x[j], j in Jprime}==b[i])
end

becomes

@addConstraint(cutstockMain, consRef[i=1:cardinalityM], sum{A[i,j]*x[j], j in Jprime}==b[i])

We prefer this form and only recommend @defConstrRef for cases where the
more compact syntax is difficult to apply.

• In the "getting started" notebook, you may want to add a note that
  to use the CPLEX, Gurobi, and Mosek packages, you first need to set up the
  software and appropriate licences.
• The use of \preceq instead of \leq for element-wise inequality seems
  a bit unusual to me, although I'm aware that some authors use it. Could you
  add a note of it's meaning?

—
Reply to this email directly or view it on GitHub
#4 (comment)
.

[Shuvomoy]

Hi Miles,

Made the changes suggested by you. Here is the nbviewer link:

http://nbviewer.ipython.org/github/Shuvomoy/juliaopt-notebooks/tree/master/notebooks/

Please let me know if I need to make any other changes.

Regards,
Shuvo

[mlubin]

LGTM. @IainNZ, @joehuchette, any comments?
I will also double check that the notebooks run locally before merging.

[IainNZ]

They're very impressive work, I like them a lot. LGTM.

[mlubin]

No issues with running the notebooks. Thanks for the contribution!

[Shuvomoy]

Thanks for merging the pull request!

On Fri, Jan 30, 2015 at 4:41 PM, Miles Lubin notifications@github.com
wrote:

No issues with running the notebooks. Thanks for the contribution!

—
Reply to this email directly or view it on GitHub
#4 (comment)
.