01 / MEANING
Keep scientific
context visible.
In the building energy example, observation identities, building groups and declared target units stay connected through prediction.
Follow the data’s contextA new language, taking shapeData science.
Data science.
On your terms.
A programming language for scientific workflows.
Aner is being developed to connect data preparation, scientific context, modeling and computation. Keep the meaning of your data close to the code that uses it.
A binary research preview is available.
Development exampleProgram excerpt
// Each image gets its own buffers.
let parameters = List(64, 64, 3)
let results = parallel.repeat(
iterations: 64,
task: process_image,
parameters: parameters,
execution: parallel.local(workers: 4),
seed: 2026)
print(results.rows())
print(results.mean(column: 0))Same image function. Four worker threads, scheduled by the OS. Private buffers and ordered results.
64 images → 4 workers → ordered results
3.3×historical measured speedupOctober 16, 2026 · 3 trials
KEEP THE QUESTION CONNECTED
ObservationsScientific contextExperimentUnderstanding
WHY A LANGUAGE FOR SCIENCE?Less glue.
Less glue.
More science.
A result depends on more than a model. The meaning of an observation, the way you prepare it and the computation you choose all matter. Aner explores what becomes possible when those concerns are designed together.
Keep scientific
02 / PREPARATION
Make preparation
an explicit choice.
Declare separate data populations. Fit standardization on training observations, then apply that fitted preparation to held out data.
See the complete workflow03 / COMPUTATION
Give computation
a clear boundary.
Preview a CSV within a scan budget, or choose local CPU workers for independent simulations. Inspect the documented limits of each operation.
Explore bounded data executionA SCIENTIFIC WORKFLOW, IN CONTEXTA prediction should
A prediction should
keep its context.
Consider a building energy study. Temperature and occupancy are inputs; energy is the target. But the building, the observation and the declared units still matter when you read the prediction.
This documented development example keeps whole buildings in declared populations and fits preparation on the training population. The prediction retains its observation identity and declared target unit.
MeasurementsGroup partitionsObservation identity
Explore the building energy exampleBuilding observations
energy · kWhIdentity, building group, measurements
Training8 observationsB0, B1, B2, B3
Validation4 observationsB4, B5
Test4 observationsB6, B7
Fit preparation on training data
Apply the fitted standardizer to each population.
PUBLISHED EXPECTED OUTPUT
A prediction you can identify.
- Validation row identity
- 401
- Declared target unit
- kWh
THE DECISION STAYS IN THE CODE
Fit on the training population.
Source excerpt. The complete example prepares observations before these lines.
let populations = data.partition_groups(observations,
List("B0", "B1", "B2", "B3"), List("B4", "B5"), List("B6", "B7"))
let preprocessing = ml.fit_standardizer(populations.train())MORE THAN A MODELStart with your
Explore all modulesStart with your
research question.
01Data and
Modeling and
Scientific
Data and
scientific context
Bounded previews, streaming summaries and explicit execution receipts for CSV workflows.
Inspect a data workflow02Modeling and
evaluation
Follow Iris from observations through training and evaluation on a separate test dataset.
Read a modeling example03Scientific
computation
Explore heat diffusion with uncertain parameters and independent simulations on local CPU workers.
Explore the simulationGOOD SCIENCE STARTS WITH A CONVERSATIONFollow Aner’s
Follow Aner’s
development.
Bring a workflow you wish were simpler, a research question or a fresh perspective. Get release updates and help shape a language around the work that matters to you.
IDEAS, MEET MOMENTUM.
MAIN MASCOTAnero
Your focused, hardworking partner for building great ideas.
COMMUNITY COMPANIONArenita
Meet Anero & ArenitaA positive companion who sparks curiosity and great ideas.