Ising

Ising is a local-first particle simulation workstation. Its current scientific core evaluates a shifted Lennard–Jones 12–6 pair potential in three dimensions, integrates motion with Velocity-Verlet, applies cubic periodic boundaries and the minimum-image convention, and can optionally regulate temperature through weak coupling. Particle positions and plotted observables come from simulation frames rather than a decorative animation.

The model uses reduced units. It is useful for exploring particle dynamics and physical behavior, but it is not a validated biomolecular package. It does not include bonded topology, electrostatics, protein force fields, constraints, particle-mesh Ewald, GPU kernels, or external HPC execution.

In Simulation Studio, choose a seeded particle arrangement and review the system size, density, periodic box, potential parameters, and initial velocity setup. The engine supports lattice construction, Maxwell-like velocity initialization, center-of-mass momentum removal, and target-temperature rescaling. Record the random seed and configuration so a run can be repeated.

Check for close contacts or an unstable starting geometry before integrating. Force-driven energy minimization with capped displacements and backtracking is available to prepare a system. Do not assume a visually plausible arrangement has low energy; inspect the force and energy diagnostics.

Start a run and use the available pause and single-step controls when you want to inspect a transition carefully. Trajectory frames are synchronized with observables and events. Move between the 3D theatre and Trajectory Explorer without losing which frame is selected. Add a bookmark and annotation when a configuration merits closer review.

Branch from a bookmarked or saved state when testing another configuration. The branch keeps its parentage, settings, provenance, and summary so it can be compared with the original. Export a portable `.ising.json` project to continue on another machine; validate the import rather than trusting a filename alone.

Observable Analysis provides potential, kinetic, and total energy; kinetic temperature; virial pressure; density; unwrapped mean-squared displacement; force and momentum diagnostics; and other statistics. Inspect energy drift before extending a run. For temperature-controlled runs, distinguish the thermostat behavior from NVE energy conservation expectations.

The analysis workspaces can calculate descriptive statistics, block uncertainty, autocorrelation, effective sample size, energy drift, diffusion regression, radial distribution functions, contact occupancy, energy-derived state populations, and transition counts. Review how each statistic was estimated and the time interval it covers. A number without its sampled frames and units is hard to interpret.

Use consistent parameters when comparing runs, or make the changed setting explicit. Different seeds can reveal variability; different time steps or ensembles alter the numerical experiment. Label any comparison with the parent run, integration settings, and observable definition. Keep raw trajectory data separate from plotted summaries when sharing results.

Project import and export perform validation and retain run summaries, bookmarks, branches, and provenance. Preserve the project file with the code version and analysis notes. Local browser storage is device-specific, so use exports as backups and transfer packages deliberately.

A simulation can help teach and explore model behavior, but its conclusions are limited by its potential and assumptions. Ising does not currently implement protein force fields, bonded topology, electrostatics, GPU kernels, or external high-performance computing. It should not be described as validated molecular dynamics for biomolecules.

If a run becomes unstable, inspect the time step, starting geometry, density, force diagnostics, and energy drift. Use a smaller step or revise the setup only after recording why. Do not hide numerical instability by selecting a later frame or changing display scaling.

Ising runs a real three-dimensional molecular-dynamics calculation for unbonded particles using a shifted Lennard–Jones 12–6 potential in reduced units. The model includes pair forces, a periodic cubic box, and a minimum-image convention. Read the model and units panel before comparing an observable with a value reported in physical units elsewhere.

This system is useful for studying particle dynamics and statistical behavior, but it is not a biological molecule simulator. It contains no bonded topology, protein force field, electrostatics, constraints, PME, GPU kernels, or external HPC backend. A particle visualization should be interpreted as the trajectory of this reduced model, not as an atomistic protein or material prediction.

Choose particle count or lattice configuration, density, temperature, and initialization seed. Velocity initialization is repeatable for a fixed configuration and seed; center-of-mass momentum is removed and velocities can be rescaled to the target temperature. Record the chosen box and state because they set the starting conditions for the whole trajectory.

Use force-driven minimization when a configuration needs relaxation before dynamics. Its capped displacements and backtracking reduce potential energy through the implemented algorithm. Review the minimized coordinates and energy before using them; minimization does not discover a global minimum or validate an experimental structure.

Velocity-Verlet integrates positions and velocities using the force at each step. Select NVE when the goal is to observe energy-conserving dynamics under the model and settings. The optional weak-coupling temperature control adjusts kinetic temperature toward a target; it changes the dynamical behavior and should be reported with the run.

Choose a time step and trajectory length appropriate to the engine's supported limits. Compare trajectories only when the input state, units, force model, and integration settings are known. A temperature controller is not a thermostat with a full physical bath model; describe the exact weak-coupling option in methods and avoid implying a canonical ensemble unless supported by the implementation.

The simulation runs in an isolated Web Worker so the interface can keep responding while the numerical engine advances. The trajectory viewer navigates recorded frames, and linked observable panels synchronize with the selected frame. Use bookmarks and annotations to mark a transition or interval that deserves later analysis; event records preserve notable run changes.

When a trajectory appears unstable, inspect the exact frame, force diagnostics, total energy, temperature, and momentum before deciding whether to continue. The application preserves simulation status separately from visualization. A paused display or a finished animation is not enough to establish that every numerical step succeeded; inspect the run event and frame history.

The observable panels include potential, kinetic, and total energy; kinetic temperature; virial pressure; density; unwrapped mean-squared displacement; force diagnostics; and momentum. Plot the full time series as well as selected summary values. Energy drift and total momentum can reveal numerical or configuration issues that are hidden by an attractive trajectory view.

Use block uncertainty, autocorrelation, and effective sample size to understand how many observations are effectively independent under the sampled trajectory. These diagnostics summarize the recorded series and their assumptions; a long trajectory does not automatically mean that all observables are equilibrated. State how burn-in, sampling interval, and blocks were chosen in any report based on the run.

The analysis workspace includes radial distribution functions, contact occupancy, energy-derived state populations, transition counts, and diffusion regression. Inspect the selected population, distance rule, lag or fitting interval, and unit basis. A radial distribution describes pair structure for the simulated model; it is not a biological contact map unless the underlying representation and method support that interpretation.

Diffusion regression depends on the chosen time interval and the relationship between unwrapped mean-squared displacement and time. State populations and transitions use the implemented energy-derived classification, so review thresholds and state definitions before comparing systems. Ising does not infer a free-energy landscape or a validated biochemical mechanism from a population plot.

Branch from a saved run when a new question changes the seed, density, temperature control, time step, or other configuration. The branch retains its parentage and method context; leave the original trajectory unchanged so the comparison remains inspectable. Use parallel comparison only when the selected runs have compatible observables and clearly recorded settings.

Projects persist device-locally and can be exported as validated `.ising.json` files. Check import validation and file identity when transferring a project. Preserve the full provenance and run summary, and keep screenshots or exported values linked to the exact project version from which they came.

Ising is appropriate for controlled exploration of a simple Lennard–Jones particle model, method demonstrations, and reproducible comparisons under explicit assumptions. It can show how a change in initialization or integration setup affects the computed trajectory and observables. Those are results of the specified reduced model only.

If a question requires molecular bonds, electrostatics, protein force fields, solvent models, long-range interactions, constraints, GPU acceleration, or a cluster queue, the current repository does not provide those features. The interface may show future backend contracts, but they are not active execution routes. Use a separately validated simulation package and preserve the method provenance when moving beyond this model.