Have a look at NLOPT optimiser package

[MichaelClerx]

https://nlopt.readthedocs.io/en/latest/NLopt_Algorithms/

[MichaelClerx]

Global derivative-free optimisers:

• Direct/Direct-L:
  • NLOPT_GN_DIRECT_L
  • NLOPT_GN_DIRECT
  • NLOPT_GN_DIRECT_L_RAND
  • NLOPT_GNL_DIRECT_NOSCAL
  • NLOPT_GN_DIRECT_L_NOSCAL
  • NLOPT_GN_DIRECT_L_RAND_NOSCAL
• CRS
  • NLOPT_GN_CRS2_LM
• ISRES
  • NLOPT_GN_ISRES
• ESCH
  • NLOPT_GN_ESCH

Local derivative-free optimisers:

• COBYLA (Also in SciPy, doesn't really work)
  • NLOPT_LN_COBYLA
• BOBYQA
  • NLOPT_LN_BOBYQA
  • NLOPT_LN_NEWUOA_BOUND
• Principal axis
  • NLOPT_LN_PRAXIS
• Simplex / Nelder-Mead (also in SciPy)
  • NLOPT_LN_NELDERMEAD
• Subplex
  • NLOPT_LN_SBPLX

So seems worth a try