Runge-Kutta ODE Solver

JJ Ben-Joseph headshot Editorial review by: JJ Ben-Joseph

Introduction: why one RK4 step is useful for a first-order ODE

Runge-Kutta ODE Solver is meant for the common case where you know a first-order differential equation, an initial point, and a step size, and you want the next numerical value without hunting for a closed-form solution. The RK4 method samples the slope at the start, halfway through the step, and at the end, then combines those samples into a better local estimate than a single straight-line extrapolation.

Because the answer depends on the exact expression you type into f(x,y), small input slips can change the result more than you might expect. A sign error, a missing parenthesis, or an h that is too large for a curved solution can all move y1 in the wrong direction even though the math still runs cleanly.

The notes below explain what the solver is doing with your slope function, how the four RK4 samples fit together, and how to judge whether the output is believable for the ODE you entered.

What problem does this Runge-Kutta ODE Solver solve?

This Runge-Kutta ODE Solver is designed to answer a precise question: starting from x0 and y0, what does one RK4 step predict for y1 after advancing by h?

If your model can be written as dy/dx = f(x,y), the page gives you a structured way to test one step of the numerical method before you build a longer integration by hand or in code. That makes it easier to compare different step sizes, see whether the slope rule produces the expected rise or fall, and confirm that the algebra in your setup matches the equation you intended to solve.

How to use this Runge-Kutta ODE Solver

Enter f(x,y) as the right-hand side of dy/dx = f(x,y), using the exact expression you want RK4 to evaluate.
Enter x0 as the starting x value where the step begins.
Enter y0 as the starting y value at x0.
Enter step h as the size of the single RK4 advance from x0 to x0+h.
Submit the form to compute y1 and update the result box.
Check whether y1 has the expected scale, sign, and direction of change before comparing a second RK4 run.

If you are testing more than one step size, write down the equation, x0, y0, and h for each run so you can reproduce the same numerical estimate later.

Inputs: how to choose values for a Runge-Kutta ODE step

For this calculator, each input maps directly to the ODE itself, so the most useful habit is to match the derivative rule, the initial conditions, and the step length exactly.

Function f(x,y): the right-hand side of dy/dx = f(x,y), written in the same syntax the page evaluates.
x0: the x-coordinate where the RK4 step begins.
y0: the known y value at that starting point.
step h: the increment from x0 to x0+h for the single numerical advance.

When the ODE carries real-world units, keep the x-axis, y-axis, and h consistent before you click calculate. If the slope rule expects one unit system and the initial conditions are entered in another, the solver will still return a number, but the number will describe the wrong physical situation.

Any prefilled values are only sample inputs for the RK4 form; replace them with the differential equation and starting conditions you actually want to solve. If you are unsure about the size of h, start with a smaller step and compare it to a second run with a larger one, because the difference between those outputs is often the fastest way to see whether the interval is too coarse.

Formulas: the RK4 weights used by this ODE solver

An RK4 update is built from four slope samples: one at the beginning of the interval, two in the middle using the earlier midpoint estimate, and one at the far end of the step.

k_{1} = f (x_{0}, y_{0}), k_{2} = f (x_{0} + \frac{h}{2}, y_{0} + \frac{h}{2} k_{1}), k_{3} = f (x_{0} + \frac{h}{2}, y_{0} + \frac{h}{2} k_{2}), k_{4} = f (x_{0} + h, y_{0} + h k_{3})

y_{1} = y_{0} + \frac{h}{6} (k_{1} + 2 k_{2} + 2 k_{3} + k_{4})

The important detail is that the midpoint samples use the earlier k values, so each later slope sees a slightly refined estimate of y. That feedback is what makes RK4 more accurate than a basic Euler step for many smooth problems, and it is also why a strongly curved or stiff equation can become sensitive if h is too large.

When you inspect the formula, think of k1 through k4 as four questions about the same ODE rather than four unrelated calculations. If those questions all produce similar slope values, the interval is probably gentle; if one sample jumps far away from the others, the step may be spanning too much change for a single update.

Worked example: one RK4 step for y' = y

To show the Runge-Kutta ODE Solver in a concrete case, use the simple exponential-growth equation dy/dx = y with x0 = 0, y0 = 1, and h = 0.1.

Because the derivative depends only on y, the sample slopes are easy to follow: k1 = 1, k2 = 1.05, k3 = 1.0525, and k4 = 1.10525. Substituting those values into the RK4 update gives y1 = 1.1051708333, which is very close to the exact value e^0.1 for this test case.

That example is useful because it shows what the solver is trying to preserve: the method should move in the same growth direction as the ODE and should stay near the smooth curve when h is modest. If you swap in a different derivative rule, the same four-sample pattern still applies, but the intermediate slopes will reflect the shape of your own equation instead of the exponential example.

Sensitivity notes: how the Runge-Kutta ODE Solver reacts to h and f(x,y)

Instead of reading a generic scenario table, compare one RK4 run against a second run where only a single input changes, because that is the clearest way to see what drives the one-step answer.

When f(x,y) changes, the slope samples can swing immediately, especially if the derivative depends strongly on y. When h changes, the solver moves the sampling points farther apart or closer together, so the effect is often a change in how much curvature the step can capture.

Keep x0 and y0 fixed when you want to isolate the effect of the slope formula.
Keep f(x,y) fixed when you want to see how a smaller or larger h changes the estimate.
Watch the midpoint samples in your mental model: they are usually where a badly chosen step first shows up.
Use the same starting conditions when comparing two runs, otherwise you are testing a different ODE setup rather than a different RK4 choice.

If a revised run flips direction or changes by far more than you expected, the problem is often not RK4 itself but the scale of h or the way the derivative rule reacts to the current y value.

How to interpret the Runge-Kutta ODE Solver result

The result box reports y1, the next numerical approximation after one RK4 step, so treat it as a local estimate rather than a complete solution curve.

When you read that number, ask whether the sign makes sense, whether the magnitude matches the derivative you entered, and whether a smaller h would likely give a similar trend. For many smooth ODEs, agreement across those checks is a good sign that the input expression and the step size fit together properly.

If you need to save a run, copy the equation, x0, y0, h, and the displayed y1 into your notes or spreadsheet. This page does not include a file export, so the clearest record is the exact set of values you typed alongside the value the solver returned.