Target specs
Properties with units
Targets and limits written once and tracked through every round.
Product 2 of 5 · Design + active learning
Inventa turns a property target into candidate structures, then chooses which ones to simulate or make next by balancing predicted value against uncertainty. It spends GPU-hours and lab weeks where they change the answer.
Abstract
01 · Capabilities
Target specs
Targets and limits written once and tracked through every round.
Substitution search
Enumerates allowed swaps around structures that are known to form.
Active learning
Bayesian selection that mixes likely winners with candidates the model is unsure about.
Budget planner
Plans a round within a compute budget and the batch size your lab can take.
Synthesizability gate
Checks hull distance and known precursors before a candidate is proposed for synthesis.
Generative design
Proposes new structures from a property target. Planned
02 · Method
Property, limit, units and exclusions.
Enumerate substitutions; size the space and the budget.
Send to Equiva and Dynara; collect predictions and uncertainties.
Rank by expected improvement; propose for approval.
03 · Catalogue
10 of 10 candidates shown
| ID | Class | Status | σ | Ehull | u | Decision |
|---|---|---|---|---|---|---|
| cand-0118 | Electrolyte | Stable | 12.4 | 8 | 0.06 | Exploit |
| cand-0342 | Electrolyte | Stable | 4.1 | 14 | 0.21 | Explore |
| cand-0731 | Electrolyte | Stable | 0.8 | 11 | 0.09 | Hold |
| cand-1207 | Cathode | Stable | — | 6 | 0.04 | Exploit |
| cand-1288 | Cathode | Off hull | — | 61 | 0.12 | Drop |
| cand-2051 | Catalyst | Stable | — | 19 | 0.30 | Explore |
| cand-2093 | Catalyst | Off hull | — | 88 | 0.08 | Drop |
| cand-0409 | Electrolyte | Off hull | 6.3 | 42 | 0.17 | Drop |
| cand-1544 | Cathode | Stable | — | 3 | 0.05 | Exploit |
| cand-2210 | Catalyst | Stable | — | 24 | 0.11 | Hold |
No candidates match both filters. Choose All in one of the groups.
04 · Specification
| Item | Detail |
|---|---|
| Inputs | Target spec, structure prototypes, allowed substitutions, budgets |
| Outputs | Ranked candidates, proposed batches with reasons, round-by-round target tracking |
| Selection | Expected improvement and upper-confidence-bound strategies; batch diversity constraint |
| Gates | Hull threshold, element exclusions, precursor availability, human approval |
| Generative proposals | Diffusion or flow models over crystal structures Planned |
| Provenance | Every proposal linked to the model version and data that produced it |
05 · Place in the loop
Inventa is the decision step. It takes the question Mentra framed, asks Equiva and Dynara for predictions, and decides what Synthex should make. It is also where the loop learns fastest: one well-chosen uncertain candidate can correct a model more than ten safe ones.
Reads the literature and your notebooks, frames the question and drafts the campaign.
Turns the target into candidates and picks the next batch by expected information.
Predicts energies, forces and stresses at near-DFT accuracy, with an uncertainty for each.
Relaxes, scores against the hull and runs GPU molecular dynamics on the survivors.
Sends the short list to partner or in-house labs and brings measurements back.
06 · Questions
Not in the first release. We start with enumeration and substitution around known prototypes, which already covers large spaces, and add generative proposals once they beat that baseline on our benchmarks.
Because the model is unsure about it. Measuring it reduces uncertainty for many similar candidates at once, which makes the next round's ranking more reliable.
Always. You approve every batch and can add or remove candidates; Inventa records the change and its reason.
Next step
The 2027 pilot pairs one of your programs with our scientists for twelve weeks. You bring a target and reference data; we bring the GPUs and the models.