Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let say I have this diagram,

enter image description here

How to find the direction vector passing through the intersection point of two straight lines?

Update: new vector is the bisector of two lines and vector may be arbitrary.

share|cite|improve this question
Do you mean for that vector to be the angle bisector of the other two? – Gerry Myerson Apr 3 '12 at 10:12
yes I mean this – user960567 Apr 3 '12 at 10:14
Then why don't you edit the question so it says what you actually mean? – Gerry Myerson Apr 3 '12 at 13:06
@GerryMyerson, updated now. Thanks. Can you please answer this if you can? – user960567 Apr 3 '12 at 13:11
What's wrong with the two answers you already have? Anyway, I'm afraid it's bedtime now. It will have to wait until tomorrow. – Gerry Myerson Apr 3 '12 at 13:20
up vote 1 down vote accepted

Normalize the two given vectors, i.e. compute $$e_a:={a\over |a|}={1\over\sqrt{13}}(2,3)\ ,\quad e_b:={b\over|b|}={1\over\sqrt{5}}(2,1)\ .$$ Then the angle bisector will pass through the point $$p:=e_a+e_b=\Bigl({2\over\sqrt{13}}+{2\over\sqrt{5}},{3\over\sqrt{13}}+{1\over\sqrt{5}}\Bigr)\doteq(1.44913,1.27926)$$ whose argument is $\arctan(0.882782)=0.723221=41.44^\circ$.

The following figure shows the geometric intuition behind the above computation:

enter image description here

share|cite|improve this answer
I think is not correct see the above calculations – user960567 Apr 3 '12 at 17:48
@user960567: My values coincide with the results of Dennis Gulko and pedja. – Christian Blatter Apr 3 '12 at 19:59
Yes got that. This seems to me more straight forward. Is this work for a vector between any 2 points? Can you please show me how? – user960567 Apr 4 '12 at 3:55
I have extensively tested all solutions. But your solution is always gives me correct values in my application. Dennis solution is also fine but sometimes I cannot get desired result. Your solution is also very very simple. Can you explain a little how you found this formula? – user960567 Apr 4 '12 at 6:20
@user960567: I did what you would do to solve this problem with ruler and compass: Draw a circle with center at the intersection of the two lines; then construct the median of the two points where the circle intersects the two lines. – Christian Blatter Apr 4 '12 at 7:56

Let's denote direction vector as $v$ ,then :


share|cite|improve this answer
I am not getting 1.1328 result. Can you please check it? – user960567 Apr 3 '12 at 10:54
@user960567 fixed – pedja Apr 3 '12 at 11:03
Not sure but you have different value than @DennisGulko. – user960567 Apr 3 '12 at 11:11
@user960567 Our solutions are basically the same... – pedja Apr 3 '12 at 11:15
PLease see the comment above if you can answer this. – user960567 Apr 3 '12 at 12:15

It seems that what OP really wants is, given direction vectors $(1,a)$ and $(1,b)$, find the direction vector of their angle bisector.

I'm going to assume $0\le a\le b\lt\infty$, and let OP figure out the necessary modifications if any for other parameter ranges.

Let $\alpha$ be the angle the 1st direction vector makes with the $x$-axis, and let $\beta$ be the angle the 2nd direction vector makes with the $x$-axis. Then we have $\tan\alpha=a$, and $\tan\beta=b$. The bisector makes an angle $(\alpha+\beta)/2$ with the $x$-axis, so its direction vector is $(1,\tan((\alpha+\beta)/2))$. By the half-angle formula for the tangent (see, e.g., we have $$\tan\left({\alpha+\beta\over2}\right)={\tan(\alpha+\beta)\over1+\sqrt{1+\tan^2(\alpha+\beta)}}$$ Combine this with the addition formula for the tangent, $$\tan(\alpha+\beta)={\tan\alpha+\tan\beta\over1-\tan\alpha\tan\beta}$$ and the values $\tan\alpha=a$ and $\tan\beta=b$ already mentioned, and you have a formula for the direction vector $(1,\tan((\alpha+\beta)/2))$ of the bisector. I think that second component simplifies to $${a+b\over1-ab+\sqrt{(a^2+1)(b^2+1)}}$$

share|cite|improve this answer
Thanks. But I found Christian Blatter solution is very simple and always gives correct result for any two vectors. – user960567 Apr 4 '12 at 6:21

You can do this by calculating the angle between the two and then finding the appropriate direction vector. Using that the angle between line passing through the points $(0,0)$ and $(x,y)$ and between the $x$-axis is given by $\arctan\frac{y}{x}$.
Then the angle between the given two lines will be: $\arctan\frac{3}{2}-\arctan\frac{4}{8}$ (do you see why?)
So, the angle between the desired line and the $x$-axis will be: $$\frac{1}{2}\left(\arctan\frac{3}{2}-\arctan\frac{1}{2}\right)+\arctan\frac{1}{2}=\frac{1}{2}\left(\arctan\frac{3}{2}+\arctan\frac{1}{2}\right)$$ Once again - do you see it?
Now, the direction vector will be $(1,y)$, where $y=\tan\left[\frac{1}{2}\left(\arctan\frac{3}{2}+\arctan\frac{1}{2}\right)\right]$.
You can calculate it with a calculator or using the identities of the $\tan$ and $\arctan$ functions.
Edit: The $\arctan$ function returns values between $-\frac{\pi}{2}$ and $\frac{\pi}{2}$ rad ($-90$ and $90$ degrees). So, if a point $(x,y)$ is in the first or fourth quadrant, then the angle between the line and the $x$-axis will be in that interval and will be exactly $\arctan\frac yx$. But if it's not, then you have to "repair" the angle: If $(x,y)$ is in the second quadrant (i.e. $x<0, y>0$) then the angle should be between $\frac{\pi}{2}$ and $\pi$ rad. The correct way to get the angle will be $\pi+\arctan\frac yx$.
If $(x,y)$ is in the third quadrant (i.e. $x<0, y<0$) then the angle should be between $-\pi$ and $-\frac{\pi}{2}$ rad. The correct way to get the angle will be $-\pi+\arctan\frac yx$.
(If you're used to calculating in degrees, replace $\pi$ by $180$ degrees everywhere).
Edit 2: If you want to calculate the angle of a line passing through two given points $(x_1,y_1)$, $(x_2,y_2)$ with the $x$-axis, You should replace each $\arctan\frac yx$ with $\arctan\frac {y_2-y_1}{x_2-x_1}$

share|cite|improve this answer
Thanks. I will check it with my application and let you know – user960567 Apr 3 '12 at 10:54
You should get $\frac18(\sqrt{65}-1)\approx 0.8827$ – Dennis Gulko Apr 3 '12 at 11:00
Not sure but you have different value than @pedja. – user960567 Apr 3 '12 at 11:07
@user960567: Yes, it will, as long as both $(x_1,y_1)$ and $(x_2,y_2)$ are in the first quadrant. If they are not, then, in some cases, the values of $\arctan$ need to be corrected by $\pi$, to have the correct angle. – Dennis Gulko Apr 3 '12 at 11:24
@user960567: I edited my post and added the correct way to calculate angles with the $x$-axis. You should be able to continue from here. – Dennis Gulko Apr 3 '12 at 19:46

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.