Contegrust and ReusableHyperOptimizer

[mertall]

Question: Why doesn't cotengrust work here? How can I find an exhaustive list of optimizers available to hyperoptimizers/rehearsal methods?

Other optimizers work, but saw on your docs to try cotengrust, just experimenting.

CODE:

# Setup the contraction optimizer using cotengra.
opt = ctg.ReusableHyperOptimizer(
    parallel="dask",
    optlib="cotengrust",
    max_time="rate:1e8",  # Limit optimization time.
    directory=True,
    progbar=True,
)

# Rehearse the sampling path (pre-optimizes contraction paths for each marginal).
print("Optimizing contraction path...")
rehs = circ.sample_gate_by_gate_rehearse(
    group_size=5,
    optimize=opt,
    simplify_sequence="ADCRS"  # Use a simpler sequence to reduce overhead.
)
print("Contraction path optimized.\n")

ERROR:

Optimizing contraction path...
INFO:distributed.http.proxy:To route to workers diagnostics web server please install jupyter-server-proxy: python -m pip install jupyter-server-proxy
INFO:distributed.scheduler:State start
INFO:distributed.scheduler:  Scheduler at:     tcp://127.0.0.1:40437
INFO:distributed.scheduler:  dashboard at:  http://127.0.0.1:8787/status
INFO:distributed.scheduler:Registering Worker plugin shuffle
INFO:distributed.nanny:        Start Nanny at: 'tcp://127.0.0.1:36971'
INFO:distributed.nanny:        Start Nanny at: 'tcp://127.0.0.1:36749'
INFO:distributed.scheduler:Register worker addr: tcp://127.0.0.1:44449 name: 1
INFO:distributed.scheduler:Starting worker compute stream, tcp://127.0.0.1:44449
INFO:distributed.core:Starting established connection to tcp://127.0.0.1:39288
INFO:distributed.scheduler:Register worker addr: tcp://127.0.0.1:40093 name: 0
INFO:distributed.scheduler:Starting worker compute stream, tcp://127.0.0.1:40093
INFO:distributed.core:Starting established connection to tcp://127.0.0.1:39292
INFO:distributed.scheduler:Receive client connection: Client-5432b3d3-1746-11f0-a458-0242ac1c000c
INFO:distributed.core:Starting established connection to tcp://127.0.0.1:39304
/usr/local/lib/python3.11/dist-packages/cotengra/parallel.py:338: UserWarning: Parallel specified but no existing global dask client found... created one (with 2 workers).
  warnings.warn(
---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
[<ipython-input-4-74f8d0f6c3f7>](https://zja11e4tzdj-496ff2e9c6d22116-0-colab.googleusercontent.com/outputframe.html?vrz=colab_20250410-060105_RC00_745970647#) in <cell line: 0>()
     10 # Rehearse the sampling path (pre-optimizes contraction paths for each marginal).
     11 print("Optimizing contraction path...")
---> 12 rehs = circ.sample_gate_by_gate_rehearse(
     13     group_size=5,
     14     optimize=opt,

11 frames
[/usr/local/lib/python3.11/dist-packages/cotengra/hyperoptimizers/hyper.py](https://zja11e4tzdj-496ff2e9c6d22116-0-colab.googleusercontent.com/outputframe.html?vrz=colab_20250410-060105_RC00_745970647#) in __init__(self, methods, minimize, max_repeats, max_time, parallel, simulated_annealing_opts, slicing_opts, slicing_reconf_opts, reconf_opts, optlib, space, score_compression, on_trial_error, max_training_steps, progbar, **optlib_opts)
    484 
    485         self._optimizer = dict(
--> 486             zip(["init", "get_setting", "report_result"], _OPTLIB_FNS[optlib])
    487         )
    488 

KeyError: 'cotengrust'

P.S. really cool work man, I am amazed by the amount of clean and stable code you have written, cotengra, quimb, cotengrust are so awesome to read through, and the best part EASY to read through. Really appreciative of this work you have done. Please keep it up 🎆

[mertall]

Nevermind, I forgot the hyper optimizer was imported from cotengra. closing this.