Niladri,
In both cases you are getting the error of an infeasible solution. The difference in this case is that you likely upgraded to the latest version of GEKKO on your system B. You can upgrade your GEKKO on system A with:
pip install gekko --upgrade
We recently released an update that better handles solver errors. If you don't want your code to fail when the solver is unsuccessful then you can wrap it in a try...except block such as:
try:
m.solve()
except:
print('Didn't find a solution')
If you are using the remote server then any version of Python or operating system (Windows, Linux, MacOS, ARM / Raspberry Pi) should produce the same solution if you have the same model and data. One difference is if you are switching to a local solve with
m = GEKKO(remote=False). The local executables don't have some of the linear solvers because they require a license to use locally. The local IPOPT solver uses the free and open-source linear solver MUMPS (http://mumps.enseeiht.fr) while the
remote server uses different (more capable) linear solver options.
Here are a couple questions to look for application differences. This isn't applicable in your case but just in case you think a difference is caused by your model or data:
1. Are you using the same model and data file? You can look in m.path and use differencing software to detect changes in your .apm (model) and .csv (data) files.
2. Are you using the same options? Please confirm remote=False or remote=True for your applications. You can also see the options in the options.json file in your run directory (m.path).
3. Has anything else changed between the two applications?
Best regards,
John Hedengren