Frequently Asked Questions

1. Why does RAM usage increase when creating multiple BayesMBAR instances?

When creating multiple BayesMBAR instances in a loop, you may notice that RAM usage increases continuously. This is caused by JAX caching jitted functions.

Solution: To manage memory usage, periodically call jax.clear_caches() when creating many BayesMBAR instances:

import jax

for i in range(many_iterations):
    # Create and use BayesMBAR instance
    mbar = BayesMBAR(...)
    # ... your code ...

    # Periodically clear JAX caches
    if i % 10 == 0:  # adjust frequency as needed
        jax.clear_caches()

Note: Clearing caches will affect all JAX jitted functions in your program, not just those from BayesMBAR.