Numpy version pinning required with astropy.convolution in FITS display notebook

@jonesmg

@jonesmg noticed following issue in displaying-images.ipynb in #76:

I had to revert my numpy version to 2.3 because the latest version (2.4) seems to be incompatible with astropy.convolution. This might trip up a less experienced python user.

I'm not sure exactly why I'm having this issue and you are not (I'm running on an M2 Macbook). I deleted my environment and started from scratch, doing the following:

conda create -n firefly python=3 astropy jupyterlab
conda activate firefly
pip install firefly_client
export FIREFLY_URL="https://irsa.ipac.caltech.edu/irsaviewer"
jupyter lab
When I tried to import the modules from astropy.convolution I hit the error: AttributeError: module 'numpy' has no attribute 'in1d'
Doing some Googling, I think this has been deprecated for a while, and removed in numpy v2.4, but astropy (v7.1) is still using it I guess. I rebuilt the environment using:

conda create -n firefly python=3 numpy=2.3 astropy jupyterlab
Then all the astropy stuff imports.

@jonesmg

@jonesmg I tried conda create -n firefly python=3 astropy jupyterlab to reproduce your issue but I still don't see any errors with numpy when importing astropy.convolution (I'm on M1 though). This is how my astropy, numpy versions look:

astropy                   7.2.0              pyhd8ed1ab_0    conda-forge
astropy-base              7.2.0           py314hdcf55e8_0    conda-forge
astropy-iers-data         0.2026.1.26.0.43.56    pyhd8ed1ab_0    conda-forge
numpy                     2.4.1           py314hae46ccb_0    conda-forge

I suspect that the astropy version you're pulling in is not up-to-date which is requiring a numpy downgrade.

Can you please create a fresh environment again with above command, activate it, and then post the output of conda list | grep -e numpy -e astropy?

@jaladh-singhal

@jaladh-singhal You're correct, my conda is installing a slightly older version of astropy when I run the same environment command.

astropy                      7.1.1                py314hdbb4bc7_0
astropy-iers-data            0.2026.1.19.0.42.31  py314hca03da5_0
numpy                        2.4.1                py314hde9b8b7_0
numpy-base                   2.4.1                py314hb4b2851_0

Is conda-forge in your default channels perhaps.

Somewhat surprisingly, when I add an explicit conda-forge channel,

conda create -n firefly -c conda-forge python=3 astropy jupyterlab

It installed much older packages:

astropy                          5.1.1            py311h4add359_3       conda-forge
numpy                            1.23.4           py311ha92fb03_1       conda-forge

I just remembered that I'm using the mamba solver rather than the classic conda solver. Perhaps that's the issue.

@jonesmg

Is conda-forge in your default channels perhaps.

@jonesmg Yes, that's correct and I was about to ask you to try conda-forge channel. Thanks for already realizing and doing that.

Somewhat surprisingly, when I add an explicit conda-forge channel. It installed much older packages.

It is indeed surprising.

I just remembered that I'm using the mamba solver rather than the classic conda solver. Perhaps that's the issue.

I just updated my conda which uses mamba-solver now by default. Doing conda create -n firefly -c conda-forge python=3 astropy jupyterlab still installs astropy v7.2.0 and numpy v2.4.1 for me.

Can you post output of conda config --show, I wonder if that can give us better idea for the version discrepancy?

My conda version is 25.11.1, and here is the conda config output:

add_anaconda_token: True
add_pip_as_python_dependency: True
aggressive_update_packages:

ca-certificates

certifi

openssl
allow_conda_downgrades: False
allow_cycles: True
allow_non_channel_urls: False
allow_softlinks: False
allowlist_channels: []
always_copy: False
always_softlink: False
always_yes: None
anaconda_anon_usage: True
anaconda_heartbeat: False
anaconda_upload: None
auto_activate: True
auto_stack: 0
auto_update_conda: True
bld_path:
changeps1: True
channel_alias: https://conda.anaconda.org
channel_priority: flexible
channel_settings:

channel: https://repo.anaconda.cloud/*
auth: anaconda-auth
channels:

defaults
client_ssl_cert: None
client_ssl_cert_key: None
clobber: False
conda_build: {}
console: classic
create_default_packages: []
croot: /opt/anaconda3/conda-bld
custom_channels:
pkgs/main: https://repo.anaconda.com
pkgs/r: https://repo.anaconda.com
pkgs/pro: https://repo.anaconda.com
custom_multichannels:
defaults:

https://repo.anaconda.com/pkgs/main

https://repo.anaconda.com/pkgs/r
local:
debug: False
default_activation_env: base
default_channels:

https://repo.anaconda.com/pkgs/main

https://repo.anaconda.com/pkgs/r
default_python: 3.13
default_threads: None
denylist_channels: []
deps_modifier: not_set
dev: False
disallowed_packages: []
download_only: False
dry_run: False
enable_private_envs: False
env_prompt: ({default_env})
environment_specifier: None
envs_dirs:

/opt/anaconda3/envs

/Users/mgjones/.conda/envs
envvars_force_uppercase: True
error_upload_url: https://conda.io/conda-post/unexpected-error
execute_threads: 1
experimental: []
export_platforms:

osx-arm64
extra_safety_checks: False
fetch_threads: 5
force: False
force_32bit: False
force_reinstall: False
force_remove: False
ignore_pinned: False
json: False
list_fields:

name

version

build

channel_name
local_repodata_ttl: 1
migrated_channel_aliases: []
migrated_custom_channels: {}
no_lock: False
no_plugins: False
non_admin_enabled: True
notify_outdated_conda: True
number_channel_notices: 5
offline: False
override_channels_enabled: True
override_virtual_packages: {}
path_conflict: clobber
pinned_packages: []
pkgs_dirs:

/opt/anaconda3/pkgs

/Users/mgjones/.conda/pkgs
plugins:
anaconda_telemetry: True
auto_accept_tos: False
use_sharded_repodata: False
prefix_data_interoperability: False
protect_frozen_envs: True
proxy_servers: {}
quiet: False
register_envs: True
remote_backoff_factor: 1
remote_connect_timeout_secs: 9.15
remote_max_retries: 3
remote_read_timeout_secs: 60.0
repodata_fns:

current_repodata.json

repodata.json
repodata_threads: None
repodata_use_zst: True
report_errors: None
rollback_enabled: True
root_prefix: /opt/anaconda3
safety_checks: warn
sat_solver: pycosat
separate_format_cache: False
shortcuts: True
shortcuts_only: []
show_channel_urls: None
signing_metadata_url_base: None
solver: libmamba
solver_ignore_timestamps: False
ssl_verify: True
subdir: osx-arm64
subdirs:

osx-arm64

noarch
target_prefix_override:
trace: False
track_features: []
unsatisfiable_hints: True
unsatisfiable_hints_check_depth: 2
update_modifier: update_specs
use_index_cache: False
use_local: False
use_only_tar_bz2: None
verbosity: 0
verify_threads: 1

Just to double check, I removed the environment and re-ran conda create -n firefly -c conda-forge python=3 astropy jupyterlab, and it now seems to work (see below). I tried updating conda and changing the solver between classic and mamba within the past few days. I'm not sure what exactly changed it, but it seems that with a fully up to date conda and conda-forge in channels, then it will install the right versions.

conda list | grep -e numpy -e astropy
astropy 7.2.0 pyhd8ed1ab_0 conda-forge
astropy-base 7.2.0 py314hdcf55e8_0 conda-forge
astropy-iers-data 0.2026.1.26.0.43.56 pyhd8ed1ab_0 conda-forge
numpy 2.4.1 py314hae46ccb_0 conda-forge

Just to double check, I removed the environment and re-ran conda create -n firefly -c conda-forge python=3 astropy jupyterlab, and it now seems to work (see below).

Interesting 😄 - I'm glad it worked out.

I tried updating conda and changing the solver between classic and mamba within the past few days. I'm not sure what exactly changed it, but it seems that with a fully up to date conda and conda-forge in channels, then it will install the right versions.

Yes, it's usually a good idea to update conda (or any other package manger for that matter). Looking at your conda config, I don't see conda-forge in default channels. It's a good idea to always use conda-forge over defaults channel, you can update your config as explained here: https://conda-forge.org/docs/user/introduction/#how-can-i-install-packages-from-conda-forge. Then you won't need to remember to supply -c conda-forge each time.

Another thing I do when creating environments is to use conda only for env creation and installing python (and non-python pacakges) and then use pip for installing python packages. So your example becomes:

conda create -n firefly python=3.13
conda activate firefly
pip install astropy jupyterlab

I'm closing it as it got resolved by upgrading the machinery. Feel free to comment if this problem reoccurs.

Navigation Menu

Repository navigation

Numpy version pinning required with astropy.convolution in FITS display notebook #77

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Numpy version pinning required with astropy.convolution in FITS display notebook #77

Description

Activity

jaladh-singhal commented on Jan 27, 2026

jonesmg commented on Jan 27, 2026

jonesmg commented on Jan 27, 2026

jaladh-singhal commented on Jan 28, 2026

jonesmg commented on Jan 29, 2026

jaladh-singhal commented on Jan 30, 2026

jaladh-singhal commented on Jan 30, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions