Nicholas McKibben

Just to make you aware, fork with Cython wrappers on top of HiGHS [here](https://github.com/mckib2/HiGHS) and scipy discussions of Cython wrappers [here](https://github.com/scipy/scipy/pull/11759) `scikit-highs` proof-of-concept implementation [here](https://pypi.org/project/scikit-highs/) _EDIT: This has been included...

This issue was resolved by solution in #282 , update to C API was all it took

I can replicate the issue: - solver: ipm - presolve: off MPS file: ``` NAME ROWS N COST L row0 E row1 COLUMNS col0 COST -1 col0 row0 1 col1...

I'm going to try to get this included as a unit test in scipy so we can make sure we catch any breakages in the future, thanks for noticing @jajhall...

Interesting that the MPS was generated incorrectly. I have a wrapper I wrote over `writeMPS` that I used to generate it, so the output is from `writeMPS`. I'll take a...

With the MPS file that you provided I can also replicate the issue with IPX without presolve

I believe the issue may be in `basis.cc` (see [here](https://github.com/ERGO-Code/HiGHS/blob/24f175a7a11385e99164d6e6d8dafa94cce929dc/src/ipm/ipx/src/basis.cc#L858)). The value of `delta_obj` is calculated as 2 in the example we are testing which seems to be unreasonably high....

Can replicate on current master still, here's basic console output: ``` Running HiGHS 1.1.1 [date: 2021-12-19, git hash: ab302f1c1-dirty] Copyright (c) 2021 ERGO-Code under MIT licence terms LP issue280 has...

> The title of the issue refers to simplex, but I can't see any problem. Even with presolve=off, the simplex solvers aren't used since the logical basis is optimal. Ah...

If you don't need the tests (it appears to be a runtime error with the unit tests), you might try building with the `-DBUILD_TESTING=OFF`