Intuition The ONE core ideaWhen several quantities change together and each one's rate of change depends on all the others, we stack them into a single list called a vector and treat that list as one moving object. Then the same four-slope averaging trick (RK4) that works for one quantity works for the whole list, with nothing new invented.
This page assumes zero prior notation. We build every symbol the parent note 4.8.27 leans on, one brick at a time, so that when you read the recipe there, every mark on the page already means something to you.
t t time . A quantity like y different value at each moment — think of the height of a swing. We write y ( t ) y t
Picture a single dot sliding along a vertical line as the clock ticks. At t = 0
Why we need this: the whole topic is about predicting where the dot will be next given where it is now and how fast it is moving.
y ′ y ′ y rate of change . If y y ′
y ′ slope
Plot y t steepness (slope) of that tiny line is y ′ y ′
Why we need this: an ODE is a rule that tells us the slope. If we know the slope everywhere, we can march forward and trace the whole curve — that marching is the numerical method.
f ( t , y ) f machine : feed it the current time t y y ′ y ′ = f ( t , y ) "the rate of change equals whatever the machine says, given where and when we are."
Worked example Reading a slope-rule out loudy ′ = − y y ′ = y e t
Why we need this: an ODE (Ordinary Differential Equation) is exactly one such rule. "Solving" it means finding the moving value y ( t ) f
Definition Starting pointt 0 start time (often 0 y 0 known value there . Together y ( t 0 ) = y 0
Intuition Why one slope-rule is not enoughThe rule f shape of the motion but not where it begins . Many curves have the same slope pattern shifted up or down — like parallel hillsides. The initial value picks which one you are on. A slope-rule plus a starting point is called an IVP (Initial Value Problem) .
h h size of one time-hop we take. Instead of knowing y every instant, we settle for knowing it at t 0 , t 0 + h , t 0 + 2 h , … h
Small h n y n n y n ≈ y ( t 0 + nh )
Why we need this: computers cannot store infinitely many instants, so we walk the curve in discrete hops of length h h accuracy against work — the theme of Local vs Global Truncation Error .
y y 1 , y 2 , … , y n column list :
y = ( y 1 , y 2 , … , y n ) T . bold y T
list and not separate numbers
Predator and prey, or position and velocity, move together : each one's rate depends on the other. Keeping them in one list lets us say "advance the whole state by one hop" in a single stroke — and, crucially, forces us to advance them at the same time , never one while the other is stale.
The number n dimension ). n = 1 n = 2
Definition Vector slope-rulef t whole list y list of slopes — one for each quantity:
f ( t , y ) = ( f 1 , f 2 , … , f n ) , y i ′ = f i ( t , y 1 , … , y n ) . f i all the y j coupling .
Common mistake "The equations don't really touch, so solve them one at a time."Why it feels right: y 1 ′ y 2 ′ Why it's wrong: because f 1 y 2 y 2 y 1 Fix: always advance the entire list together , step by step. This is the single most important habit the parent note drills.
Definition Order of an ODEThe order is the highest prime-count. y ′ first-order . y ′′ own slope, i.e. acceleration) → second-order .
Intuition Why we only ever need first-orderA second-order rule like y ′′ = g ( t , y , y ′ ) repackaged as a first-order system by naming the derivative as a brand-new variable:
u 1 = y , u 2 = y ′ . u 1 ′ = u 2 u 2 ′ = g ( t , u 1 , u 2 ) two 1st-order ones — see Reducing higher-order ODEs to first-order systems .
Why we need this: it means "RK4 for systems" secretly solves every ODE, of any order, once you reduce it.
Definition Small notation, spelled out
≈ approximately equal to " — our hops give near-answers, not exact ones.∫ t t + h … d s definite integral : the exact accumulated change over one hop, the area under the slope curve between t t + h Simpson's Rule ).O ( h 4 ) big-O : "the error shrinks in step with h 4 h 2 4 = 16 Local vs Global Truncation Error .2 h 6 h
Recall WhyO ( h 4 ) h 2 4 = 16
k i k i trial slope-list — the vector slope f 6 1 ( k 1 + 2 k 2 + 2 k 3 + k 4 )
Intuition Why four peeks, weighted1 , 2 , 2 , 1 Euler's method for systems ) drifts because the slope changes during the hop. Peeking again in the middle and at the end, and trusting the middles more (weight 2 h 4 k i list , you must fill the whole list before moving to the next peek.
vector rule f bold: coupling
reduction of higher order
exact integral over one hop
four probe slopes k1 to k4
Everything on the left is a single number or idea; the two boldings (y f
Read what a symbol means it maps to a picture; if not, reread that section.
y ( t ) the value of a moving quantity at time t
y ′ the rate of change of y
f ( t , y ) the slope-rule machine: input time and value, output the slope.
y ( t 0 ) = y 0 the starting point that selects which curve you're on (an IVP).
h the size of one time-hop; smaller = more accurate, more work.
y a stacked list of several quantities treated as one moving object.
f ( t , y ) the vector slope-rule; each component may depend on all others (coupling).
Coupling each equation's rate depends on the other unknowns, so you must advance all together.
Order of an ODE the highest prime-count; second-order reduces to a first-order system.
Reduction trick set u 1 = y , u 2 = y ′ u 1 ′ = u 2 , u 2 ′ = g ( t , u 1 , u 2 )
k i a trial slope-list; RK4 averages four of them with weights 1 , 2 , 2 , 1 6
∫ t t + h f d s the exact accumulated change over one hop, estimated by sampling slopes.
O ( h 4 ) halving h 16 ×
Parent topic — Hinglish RK4 for a single ODE — the scalar method these symbols generalise.Euler's method for systems — one-peek cousin, good for building intuition.Reducing higher-order ODEs to first-order systems Local vs Global Truncation Error — meaning of O ( h 4 ) Simpson's Rule — the sampling idea behind the 1 , 2 , 2 , 1 Stiff systems and stability — where step size h 
