AROPS

AROPS: A Framework of Automated Reaction Optimization with Parallelized Scheduling AROPS

• The state-of-the-art Bayesian optimizer (PISC-BO) implemented in AROPS can accommodate multi-reactor/multi-analyzer optimization.
• AROPS provides three different scheduling schemes with optional experiment discarding mechanism, which can be selected according to users’ preference to time cost or reagent consumption.
• A multi-reactor/multi-analyzer automated synthesis screening platform simulator was constructed to evaluate the optimization algorithm under various scenarios.

Installation

Installation Requirements

• Python >= 3.7
• PyTorch >= 1.10
• gpytorch==1.8.1
• botorch==0.6.0
• scipy
• scikit-optimize

Manual install

You can do a manual install. For a basic install, run:

https://github.com/Ruan-Yixiang/AROPS.git
pip install -r AROPS_requirements.txt

Getting Started

Here's a run down of the optmization example of Case 1 of Benchmark A.

# The optmization example of Case 1 of Benchmark A.
from arops import AROPS

# Create AROPS instance
simulator = AROPS(reactor_number=2, analysis_instrument_number=1, schedule='ARIA-PI',
                  benchmark='Benchmark_A1', analysis_time=10, pi_min=1e-4)

# Run the optimization process
res = simulator.run()

# Results output
print('Optimum: {:.4f}, \nTime: {:.2f} min, \nNumber of Experiments: {:d}'.format(res.opt_obj, res.time, res.n_exps))
print('Optimal conditions:', res.opt_con)

Benmark B can only run in the Linux environment.

License

AROPS is distributed under an MIT License.