SLUSCHI
I own the Arizona license plate of SLUSCHI, and I run a SLUSCHI scavenger hunt with my students.

SLUSCHI Job Status

Submit a SLUSCHI job

Submission Policy

To keep this public service available, each email address may submit only one SLUSCHI job every 24 hours. Additional submissions within 24 hours will not be queued.

If you would like to run more calculations or discuss additional configurations, please email qhong@alumni.caltech.edu to discuss a plan or collaboration.

This page refreshes every 30 seconds.

MeltBench

SLUSCHI-UP predictions are continuously validated against experimentally measured melting temperatures through MeltBench, our open benchmark platform for inorganic materials.

View MeltBench benchmarks →

Definitions

SLUSCHI-UP MT: SLUSCHI-UP/uMLIP prediction from the simulation workflow.
PBE Correction: a density-functional-theory (DFT) correction to the raw SLUSCHI-UP/uMLIP melting-temperature prediction. The correction is derived from PBE calculations and is intended to reduce systematic errors inherited from the underlying machine-learning potential. Additional theoretical details are available in Hong et al. (2026) .

EXPT Correction: MeltBench reveals that universal machine-learning interatomic potentials (uMLIPs) can systematically overestimate or underestimate melting temperatures for certain classes of materials. To account for these residual biases, we train a surrogate correction model on the difference between uMLIP/PBE predictions and experimentally measured melting temperatures. The reported value applies this learned correction to the baseline prediction (using the PBE-corrected value when available, otherwise the raw uMLIP melting temperature). GNN MT: melting temperature predicted by the GNN model.
Expt MT: experimental melting temperature from MeltBench.
Melting temperature here is T0 in the phase diagram community. In other words, SLUSCHI gives the temperature at which the solid and liquid phases of the same composition have equal free energy.

Days Online
43
since launch
Users
24
unique email addresses
Visits
42235
jobs page loads
Jobs
266
completed entries
GPU Hours
4143.8
total from MeltBench dataset
Error [K]
140
best mean absolute error
Choose All uMLIP+PBE+Expt
Best % Error
9.1
best mean percentage error
Choose All uMLIP+PBE+Expt

Showing jobs with status: queued

Click a Job ID (for example, ab123) to view detailed results, simulation logs, and output files.

Avoid duplicate calculations. Before submitting a new SLUSCHI-UP job, please search the MeltBench benchmark table or download the MeltBench CSV dataset to check whether the material has already been studied. Reusing existing results helps conserve community GPU resources and reduces unnecessary duplicate calculations.
Job ID Chemical Formula MLP SLUSCHI-UP MT PBE Correction EXPT Correction GNN MT Expt MT Email Status GPU Hours Created (MST) Updated (MST)
No jobs found.