0.1.6

Features

• Added job_name to speqtrum's submit method to allow the user to give a custom name to their job. (PR #86)

• SpeQtrum jobs now retain their concrete result types from submission through retrieval, so every handle produced by submit returns the right FunctionalResult whether you are sampling, evolving in time, or orchestrating a variational loop. Waiting for a fresh job still takes a single call, but the typed detail that comes back no longer needs manual casting:

  job_handle = speqtrum.submit(Sampling(circuit), device="cuda_state_vector")
  job = speqtrum.wait_for_job(job_handle)
  results = job.get_results()  # Your IDE will now know this is a SamplingResult

  The same strong typing applies when inspecting historical runs. Recreate a handle with the appropriate class method, pass it to get_job, and the returned TypedJobDetail exposes the precise payload you expect:

  job_handle = JobHandle.sampling(1200853)
  job = speqtrum.get_job(job_handle)
  results = job.get_results()  # Your IDE will now know this is a SamplingResult

  Variational programs are the only scenario that needs an extra hint: because they wrap another functional, recreate the handle with the inner functional’s result type so the execution results can be validated against your expectations:

  job_handle = JobHandle.variational_program(1200853, result_type=SamplingResult)
  job = speqtrum.get_job(job_handle)
  results = job.get_results()  # Your IDE will now know this is a VariationalProgramResult
  optimal = results.optimal_execution_results  # Your IDE will now know this is a SamplingResult

  Passing a bare integer job identifier to wait_for_job or get_job remains valid for quick checks and backwards-compatibility, but doing so skips the handle metadata and yields an untyped JobDetail, so you will need to inspect or cast the result manually. (PR #87)

Bugfixes

• Fixed the HardwareEfficientAnsatz so the first layer of single-qubit gates is not duplicated, restoring the expected gate and parameter counts. (PR #90)
• Fixed the scheduler so numpy floats are accepted without errors. Fixed the QuTiP backend by aligning the schedule-to-backend mapping structure. (PR #91)

Improved Documentation

• Updating the documentation and README to correspond to the latest version of QiliSDK. (PR #80)
• Added extra doc-strings, and fixed outdated doc-strings. (PR #84)

Misc

• PR #88, PR #92