Installation

vMPT is a local-only tool: it runs on your machine, files stay on your disk, computation uses your local Python.

With pip (recommended, v1.2.2+)

pip install jwst-vmpt
vmpt                                                             # opens the app

The console script vmpt accepts the same flags as the legacy run.sh:

vmpt --port 5010                                                 # different port
vmpt --fits img.fits --catalog a.csv --catalog b.csv             # stack catalogs
vmpt --jpg img.jpg --wcs wcs.fits --catalog targets.csv          # JPG + WCS pair
vmpt --jpg img.jpg --wcs wcs.fits --v3pa 209                      # start at a given roll
vmpt examples download                                           # grab example_a370 + example_r0600
vmpt --version                                                   # print version and exit

--v3pa DEG / --apa DEG set the initial pointing roll (V3 PA, or the NIRSpec aperture PA — APA = V3 PA + V3IdlYAngle; --v3pa wins if both are given). vmpt --version (also -version / -V) prints the version.

The wheel itself is ~20 MB (includes the MSA grid + per-shutter dispersion table with all 9 supported disperser/filter combos). The two example datasets (~70 MB combined) are fetched on demand via vmpt examples download and dropped into ~/.vmpt/examples/ (stable per-user cache, since v1.3.1). The in-app “Load Abell 370 example” and “Load RXCJ0600 example” buttons find them there automatically — no need to run vmpt from a particular working directory.

For sysadmins or shared machines, the cache location is overridable:

VMPT_EXAMPLES_DIR=/data/shared/vmpt_examples vmpt examples download
VMPT_EXAMPLES_DIR=/data/shared/vmpt_examples vmpt           # app reads the same env var

From source (developers)

git clone https://github.com/fengwusun/vMPT.git
cd vMPT
pip install -e .            # editable install, picks up local edits
./run.sh                    # same as `vmpt` after install
pytest tests/               # 238 passed, 5 skipped

STScI’s stenv

If you already use the STScI JWST/HST pipeline environment, the heavy deps (astropy, jwst, pysiaf, …) are already there:

conda activate stenv
pip install jwst-vmpt       # or `pip install -e .` from a checkout
vmpt

What’s installed

After pip install jwst-vmpt:

Thing	Where	Purpose
vmpt/ package	site-packages	Bokeh app + optimizer + IO
vmpt/data/nirspec_msa_v2v3.npz	site-packages	MSA shutter centres in V2/V3
vmpt/data/dispersion_cutoffs.npz	site-packages	per-shutter λ table
vmpt console script	$PATH	starts the Bokeh server

Verifying

vmpt --help                            # should print usage
python -c "import vmpt; print(vmpt.__file__)"

If vmpt is missing from $PATH, your venv’s bin/ directory isn’t being sourced — activate the venv or run with the absolute path printed by pip show -f jwst-vmpt | grep vmpt$.