Q: How do you define the derivatives of the Heaviside, Sign, Absolute Value, and Delta functions? How do they relate to one another?

Physicist: These are four standard reference functions.  In the same way that there are named mathematical constants, like π or e, there are named mathematical functions.  These are among the more famous (after the spotlight hogging trig functions).

The Sign, Delta, Absolute Value, and Heaviside functions.

The Sign, Delta, Absolute Value, and Heaviside functions.  The graphs on top are the slope of the graphs on the bottom (and slope=derivative).

The absolute value function flips the sign of negative numbers, and leaves positive numbers alone.  The sign function is 1 for positive numbers and -1 for negative numbers.  The Heaviside function is very similar; 1 for positive numbers and 0 for negative numbers.  By the way, the Heaviside function, rather than being named after its shape, is named after Oliver Heaviside, who was awesome.

The delta function is a whole other thing.  The delta function is zero everywhere other than at x=0 and at x=0 it’s infinite but there’s “one unit of area” under that spike.  Technically the delta function isn’t a function because it can’t be defined at zero.  The “Dirac delta function” is used a lot in physics (Dirac was a physicist) to do things like describe the location of the charge of single particles.  An electron has one unit of charge, but it’s smaller than basically anything, so describing it as a unit of charge located in exactly one point usually works fine (and if it doesn’t, don’t use a delta function).  This turns out to be a lot easier than modeling a particle as… just about anything else.

The derivative of a function is the slope of that function.  So, the derivative of |x| is 1 for positive numbers (45° up), and -1 for negative numbers (45° down).  But that’s the sign function!  Notice that at x=0, |x| has a kink and the slope can’t be defined (hence the open circles in the graph of sgn(x)).

The derivative of the Heaviside function is clearly zero for x≠0 (it’s completely level), but weird stuff happens at x=0.  There, if you were to insist that somehow the slope exists, you would find that no finite number does the job (vertical lines are “infinitely steep”).  But that sounds a bit like the delta function; zero everywhere, except for an infinite spike at x=0.

It is possible (even useful!) to define the delta function as δ(x) = H’(x).  Using that, you find that sgn’(x) = 2δ(x), simply because the jump is twice the size.  However, how you define derivatives for discontinuous functions is a whole thing, so that’ll be left in the answer gravy.


Answer Gravy: The Dirac delta function really got under the skin of a lot of mathematicians.  Many of them flatly refuse to even call it a “function” (since technically it doesn’t meet all the requirements).  Math folk are a skittish bunch, and when a bunch of handsome/beautiful pioneers (physicists) are using a function that isn’t definable, mathematicians can’t help but be helpful.  When they bother, physicists usually define the delta function as the limit of a series of progressively thinner and taller functions (usually Gaussians).

One of the simplest ways to construct the delta function is the series of functions shown here (and infinitely many more).

One of the simplest ways to construct the delta function is the series of functions fn(x) = n, 0 ≤ x ≤ 1/n (the first four of which are shown).  The area under each of these is 1, and most/many of the important properties of delta functions can be derived by looking at the limit as n→∞.

Mathematicians take a different tack.  For those brave cognoscenti, the delta function isn’t a function at all; instead it’s a “distribution”, which is a member of the dual of function space, and it’s used to define a “bounded linear functional”.

So that’s one issue cleared up.

A “functional” takes an entire function as input, and spits out a single number as output.  When you require that the functional is linear (and why not?), you’ll find that the only real option is for the functional to take the form F(g) = \int f(x)g(x)\,dx.  This is because of the natural linearity of the integral:

\begin{array}{ll}F(g + h) \\= \int f(x)\left(g(x)+h(x)\right)\,dx \\= \int f(x)g(x)\,dx + \int f(x)h(x)\,dx \\= F(g) + F(h)\end{array}

In F(g) = \int f(x)g(x)\,dx, F is the functional, f(x) is the distribution corresponding to that functional, and g(x) is the function being acted upon.  The delta function is the distribution corresponding to the functional which simply returns the value at zero.  That is, \int \delta(x)g(x)\,dx = g(0).  So finally, what in the crap does “returning the value at zero” have to do with the derivative of the Heaviside function?  As it happens: buckets!

Assume here that A<0<B,

\begin{array}{ll}    \int_A^B \delta(x)g(x)\,dx \\[2mm]    = H(x)g(x)\big|_A^B - \int_A^B H(x)g^\prime(x)\,dx & \textrm{(integration by parts)} \\[2mm]    = H(B)g(B) - H(A)g(A) - \int_A^B H(x)g^\prime(x)\,dx \\[2mm]    = H(B)g(B) - \int_0^B H(x)g^\prime(x)\,dx & (H(x)=0, x<0) \\[2mm]    = g(B) - \int_0^B g^\prime(x)\,dx & (H(x)=1, x>0) \\[2mm]    = g(B) - g(x)\big|_0^B & \textrm{(fundamental theorem of calculus)} \\[2mm]    = g(B) - \left[g(B) - g(0) \right] \\[2mm]    =g(0)    \end{array}

Running through the same process again, you’ll find that this is a halfway decent way of going a step further and defining the derivative of the delta function, δ’(x).

\begin{array}{ll}    \int_A^B \delta^\prime(x)g(x)\,dx \\[2mm]    = \delta(x)g(x)\big|_A^B - \int_A^B \delta(x)g^\prime(x)\,dx \\[2mm]    = - \int_A^B \delta(x)g^\prime(x)\,dx \\[2mm]    = -g^\prime(0)    \end{array}

δ’(x) also isn’t a function, but is instead another profoundly abstract distribution.  And yes: this can be done ad nauseam (or at least ad queasyam) to create distributions that grab higher and higher derivatives of input functions.

Posted in -- By the Physicist, Equations, Math | 2 Comments

Q: What does “E=mc2” mean?

Physicist: This famous equation is a little more subtle than it appears.  It does provide a relationship between energy and matter, but importantly it does not say that they’re equivalent.

First, it’s worth considering what energy actually is.  Rather than being an actual “thing” in the universe, energy is best thought of as an abstract (there’s no such thing as pure energy).  Energy takes a heck of a lot of forms: kinetic, chemical, electrical, heat, mechanical, light, sound, nuclear, etc.  Each different form has it’s own equation(s).  For example, the energy stored in a (not overly) stretched or compressed spring is E=\frac{1}{2}kx^2 and the energy of the heat in an object is E=CT.  Now, these equations are true insofar as they work (like all true equations in physics).  However, neither of them are saying what energy is.  Energy is a value that we can calculate by adding up the values for all of the various energy equations (for springs, or heat, or whatever).

The useful thing about energy, and the only reason anyone ever even bothered to name it, is that energy is conserved.  If you sum up all of the various kinds of energy one moment, then if you check back sometime later you’ll find that you’ll get the same sum.  The individual terms may get bigger and smaller, but the total stays the same.

For example, the equation used to describe the energy of a swinging the pendulum is E = \frac{1}{2}mv^2 + mgh where the variables are mass, velocity, gravitational acceleration, and height of the pendulum.  These two terms, the kinetic and gravitational-potential energies, are included because they change a lot (speed and height change throughout every swing) and because however much one changes, the other absorbs the difference and keeps E fixed.  There are more terms that can be included, like the heat of the pendulum or its chemical potential, but since those don’t change much and the whole point of energy is to be constant, those other terms can be ignored (as far as the swinging motion is concerned).

In fact, it isn’t obvious that all of these different forms of energy are related at all.  Joule had to do all kinds of goofy experiments to demonstrate that, for example, the sum of gravitational potential energy and thermal energy stays constant.  He had to build a machine that turned the energy of an elevated weight into heat, and then was careful to keep track of exactly how much of the first form of energy was lost and how much of the second was gained.

As the weight falls, it turns an agitator that heats the water.  Joule's device couples the gravitational potential of the weight with the thermal energy of the water in a tank.

As the weight falls, it turns an agitator that heats the water. Joule’s device couples the gravitational potential of the weight with the thermal energy of the water in a tank.  The sum of the two stayed constant.

Enter Einstein.  He did a few fancy things in 1905, including figuring out a better way of doing mechanics.  Newtonian mechanics had some subtle inconsistencies that modern (1900 modern) science was just beginning to notice.  Special relativity helped fixed the heck out of that.  Among his other predictions, Einstein suggested (with some solid, consistent-with-experiment, reasoning) that the kinetic energy of a moving object should be E = \frac{mc^2}{\sqrt{1-\left(\frac{v}{c}\right)^2}}, where the variables here are mass, velocity, and the speed of light (c).  This equation has since been tested to hell and back and it works.  What’s bizarre about this new equation for kinetic energy is that even when the velocity is zero, the energy is still positive.

Up to about 40% of light speed (mach 350,000), E = mc^2 + \frac{1}{2}mv^2 is a really good approximation of Einstein’s kinetic energy equation, E = \frac{mc^2}{\sqrt{1-\left(\frac{v}{c}\right)^2}}.  The approximation is good enough that ye natural philosophers of olde can be forgiven for not noticing the tiny error terms.  They can also be forgiven for not noticing the mc2 term.  Despite being huge compared to all of the other terms, mc2 never changed in those old experiments.  Like the chemical potential of the pendulum, the mc2 term wasn’t important for describing anything they were seeing.  It’s a little like being on a boat at sea; the tiny rises and falls of the surface are obvious, but the huge distance to the bottom is not.

So, that was Einstein’s contribution.  Before Einstein, the kinetic energy of a completely stationary rock and a missing rock was the same (zero).  After Einstein, the kinetic energy of a stationary rock and a missing rock were extremely different (by mc2 in fact).  What this means in terms of energy (which is just the sum of a bunch of different terms that always stays the same) is that “removing an object” now violates the conservation of energy.  E=mc2 is very non-specific and at the time it was written: not super helpful.  It merely implies that if matter were to disappear, you’d need a certain amount of some other kind of energy to take its place (wound springs, books higher on shelves, warmer tea, some other kind); and in order for new matter to appear, a prescribed amount of energy must also disappear.  Not in any profound way, but in a “when the pendulum swings up, it also slows down” sort of way.  Einstein also didn’t suggest any method for matter to appear or disappear (that came later).  So, energy is a sort of strict economy (total never changes) with many different currencies (types of energy).  Einstein showed that matter needed to be included in that “economy”, and that some things in physics are simpler if it is.

While it is true that the amount of mass in a nuclear weapon decreases during detonation, that’s also true of every explosive.  For that mater, it’s true of everything that releases energy in any form.  When you drain a battery it literally weighs a little less because of the loss of chemical energy.  The total difference for a good D-battery is about 0.015 picograms, which is tough to notice especially when the battery is ten billion times more massive.  About the only folks who regularly worry about the fact that energy and matter can be sometimes be exchanged are high-energy physicists.

Cloud chamber tracks like these provided some of the earliest evidence of particle creation, and have entertained aged nerds for decades.

Cloud chamber tracks like these provided some of the earliest evidence of particle creation and have entertained aged nerds for decades.

As far as a particle physicist is concerned, particles don’t have masses; they have equivalent energies.  If you happen to corner one at a party (it’s not hard, because they’re meek), ask them the mass of an electron.  They’ll probably say “0.5 mega-electronvolts” which is a unit of energy (the kinetic energy of a single unit of charge accelerated by 500,000 volts).  In particle physics, the amount of energy released/sequestered when a particle is annihilated/created is typically more important that the amount that a particle physically weighs (I mean, how hard is it to pick up a particle?).  So when particle physicists talk shop, they use energy rather than matter.  For those of us unbothered by creation and annihilation, the fact that rest-mass is a term included among many different energy terms is pretty unimportant.  Nothing we do or experience day-to-day is affected by the fact that rest-mass has energy.  Sure the energy is there, but changing it, getting access to it, or doing anything useful with it is difficult.

The cloud chamber picture is from here, and there’s a video of one in action here.

Posted in -- By the Physicist, Equations, Philosophical, Physics, Relativity | 8 Comments

Q: Is it possible to have a completely original thought?

Physicist: Nope!  At least, not for the last 27 years.

The last truly, verifiably original thought was had by Kjersti Skramstad of Oslo, in October of 1987.  She reported her insight immediately, as all original thinkers do, and since then there’s been nothing new under the Sun.  That stunning insight, by the way, was “curling ville være lettere med lettere steiner!“.

In 1995 there was a lot of buzz around the scientists at Bell labs; they briefly skirted originality before it was realized that their entire venture had been sketched out, beginning-to-end, by Claude Shannon in one of his notebooks almost 50 years earlier.  In fact, there has been a quiet but insistent push in some industries to remove the phrase “reinventing the wheel” from common parlance, under the assertion that it is now redundant and applies to all invention.

In scientific circles the concern is fairly minimal.  There are enough “loose pieces” around that scientists will still be making great strides for decades.  For example, by combining lots of boring animals to create awesome crimes against nature (hippogriffs, cockatrices, manticores, etc.).  Or by taking an ordinary thing (e.g., elevators) and adding the word “space” to them (e.g., space elevators).  The ideas may be unoriginal, but science still happens when you try them out for the first time.

Piano -> Space piano

Piano -> Space Piano.  Science marches on.

For we ordinary folk, original thoughts aren’t too important, but artists (for whom originality pays the bills) have been in a panic since the late 70′s when it first became clear that the well of new ideas was running dry.  In particular, 1978 saw the album “More Songs about Buildings and Food” produced, bringing the epoch of original composition to an unceremonious close.  There’s some hope that Laurie Anderson may have done something completely novel with her masterpiece “three minutes and forty-four seconds of white-noise while wearing an extraneous prosthesis” but some more pessimistic parties have already drawn parallels to John Cage’s 4’33″.  Time will tell.

Oddly enough, no politicians have noticed.  Like, at all.

Posted in -- By the Physicist, April Fools | 10 Comments

Q: How can the universe expand faster than the speed of light?

Physicist: You’ll often hear that “the universe is expanding faster than the speed of light”.  However, this statement is akin to statements like “green is bigger than happy”.  It’s not even wrong.

There are two big things to remember about the expansion of the universe.  First, the universe doesn’t expand at a particular speed, it expands at a speed per distance.  Right now it’s about 70 kilometers per second per megaparsec.  That means that galaxies that are about 1 megaparsec (1 parsec = 3 lightyears and change) away are presently getting farther away at the rate of 70 km every second, on average.  Galaxies that are 2 megaparsecs away are presently getting father away at the rate of 140 km every second, on average.

Notice the awkward phrasing there: distant galaxies are “getting farther away”, but oddly enough they are not “moving away”.

Initially, the distance between Red and Yellow is 1, and the distance between Red and Green is 2.  After double the size of the "universe" the distances are 2 and 4, respectively.  Yellow receded by 1, but Green receded by 2.  Green would seem to be moving faster than Yellow, but in fact all of the dots are sitting still while the space they inhabit expands.

Initially, the distance between Red and Yellow is 1, and the distance between Red and Green is 2. After doubling the size of the “universe” the distances are 2 and 4, respectively. Yellow receded by 1, but Green receded by 2. Green would seem to be “moving” faster than Yellow, but in fact all of the dots are sitting still while the space they inhabit expands.

The easiest way to think about the expansion of the universe is to think about the expansion of something simpler, like a balloon.  If for some reason you have a balloon covered in ants, and you inflate it slowly, then the ants that are nose-to-nose (pardon, “antennae-to-antennae”) with each other will barely notice the expansion.  However, ants the farther two ants are apart, the more the expansion increases the distance between them.  If an ant on one side tries to run to one of her sisters on the far side of the balloon, she may find that the distance between the two of them is increasing faster than she can close that distance.

The distance at which this happens (where the rate at which the distance decreases because of the movement of the ant and the rate at which the distance increases due to the expansion of the balloon) is a kind of “ant horizon”.  Any pair of ants that are already farther apart than this distance can never meet, and any pair closer than this distance may (if they want).  In the picture above, if an ant can run a distance of 2 during the expansion time, then an ant starting at the yellow point could reach the red point, but an ant starting at the green point will always find itself maintaining the same distance from the red point.

The “ant horizon” is a decent enough analog for the edge of the visible universe.  The speed at which the ant runs is described with respect to the part of the balloon it’s presently standing on and the speed at which light travels is with respect to the space it travels through (technically with respect to objects that are “sitting still” in space).  The oldest photons we see are those that have come from just barely on the near side of the distance at which light can’t close the gap.  It’s not that things beyond that distance are moving away faster than light (almost all the galaxies and gas and whatnot are moving slowly with respect to “the balloon”), it’s that the light they emit just isn’t moving fast enough to overcome the expansion.  Light beyond that is still moving at light speed, and it may even be trying to move toward us, but the distance is simply expanding too fast.

Here the analogy breaks down and starts making our intuition incorrect.  When you inflate a balloon the sides are obviously moving apart.  You can use a rule (maybe a tape measure) and a stopwatch and you can say “dudes and dudettes of the physics world, the speed of expansion is ____”.  Even worse, when a balloon expands it expands into the space around it, which begs the question “what is the universe expanding into?“.  But keep in mind, all that physics really talks about is the relationship between things inside of the universe (on the surface of the balloon).  If you draw a picture on the surface of a balloon, then if the balloon is dented somewhere or even turned inside-out, the picture remains the same (all the distances, angles, densities, etc. remain the same).

Point of fact: it may be that the balloon is a completely false metaphor for the universe as a whole, since the best modern measurements indicate that the universe is flat.  That is, rather than being a closed sphere (hypersphere) it just goes forever in ever direction.  This means that there is genuinely no way to describe the expansion of the universe in terms of a speed (there’s no “far side of the balloon” to reference).

It’s a little subtle (“subtle” = “math heavy”), but there are a number of phenomena that allow astronomers to clearly tell the difference between things “moving away” and things “expanding away” from us.  For example, beyond a certain distance galaxies no longer get smaller (the way things that are moving away should), instead they get redder and stay about the same size independent of distance, due to a lensing effect of the expansion.  Isn’t that weird?

Posted in -- By the Physicist, Astronomy, Physics, Relativity | 13 Comments

Q: How fast are we moving through space? Has anyone calculated it?

The original question was: Considering the spin of the earth, it’s orbit around the sun, the sun’s orbit around the Milky Way and the Milky Way’s journey through interstellar space, has anyone calculated our speed though the universe?


Physicist: The short answer is “yes”, and the long answer is “well… yes”.

The problem with motion is that “true motion” doesn’t exist.  The best we can do is talk about “relative motion” and that requires something else to reference against.  What you consider to be stationary (what you chose to define your movement with respect to) is a matter of personal choice.  The universe isn’t bothered one way or the other.

Relative to your own sweet self: Zero.  This sounds silly, but it’s worth pointing out.

Relative to the Earth: The Earth turns on its axis (you may have heard), and that amounts to about 1,000 mph at the equator.  The farther you are from the equator the slower you’re moving.  This motion can’t be “ignored using relativity”, since relativity only applies to constant motion in a straight line, and movement in a circle is exactly not that.  This motion doesn’t have much of an effect on the small scale (people-sized), but on a planetary scale it’s responsible for shaping global air currents (including hurricanes!).

Relative to the Sun: The Earth orbits the Sun at slightly different speeds during the year; fastest around new years and slowest in early July (because it’s farther from or closer to the Sun respectively).  But on average it’s around 66,500 mph.  By the way, the fact that this lines up with our calendar year (which could be argued to be based on the tilt of the Earth, which dictates the length of the day) to within days is a genuine, complete coincidence.  This changes slowly over time, and in several thousand years from now it will no longer be the case.  Fun fact.

Relative to the Milky Way: The Sun moves through the galaxy at somewhere around 52,000 mph.  This is surprisingly tricky to determine.  There’s a lot of noise in the the speed of neighboring stars (It’s not unusual to see stars with a relative speed of 200,000 mph) and those are the stars we can see the clearest.  Ideally we would measure our speed relative to the average speed of the stars in the galactic core (like we measure the speed at the equator with respect to the center of the Earth), however that movement is “sideways” and in astronomy it’s much much easier to measure “toward/away” speed using the Doppler effect.  Of the relative speeds mentioned in this post, the speed of our solar system around the galaxy is the only one that isn’t known very accurately.

Relative to the CMB: The Milky Way itself, along with the rest of our local group of galaxies, is whipping along at 550 km/s (1.2 million mph) with respect to the Cosmic Microwave Background.  Ultimately, the CMB may be the best way to define “stationary” in our corner of the universe.  Basically, if you move quickly then the light from in front of you becomes bluer (hotter), and the light from behind you gets redder (colder).  Being stationary with respect to the CMB means that the “color” of the CMB is the same in every direction or more accurately (since it’s well below the visual spectrum) the temperature of the CMB is the same in every direction (on average).

Posted in -- By the Physicist, Astronomy, Physics | 13 Comments

Q: If you flip a coin forever, are you guaranteed to eventually flip an equal number of heads and tails?

The original question was: Lets say we have a fair coin that is flipped a hundred times and at the end of the trial there have been 40 tails and 60 heads. At this time there have been 20 more heads than tails and it could be said that heads is “dominant”.

Is it inevitable that, given enough time to occur, eventually a “switch-over” in dominance will occur and tails will be dominant, or is it the case that dominance by heads or tails can carry on indefinitely?

If a “self correction” is inevitable, how long, on average, would a correction take (the equalizing of the number of heads and tails)? Is it something you will see by the end of the day or something you won’t see in your lifetime. (on average of course, I accept that both are theoretically possible)?

What is the likely number of flips between each “switch-over” and is a “switch-over” guaranteed?


Physicist: In the very specific case of this question: yes.  In fact (for this very specific case) any given tally (e.g., 7,823 more heads than tails) will happen eventually, regardless of what the imbalance presently is.  Since that includes #-of-heads = #-of-tails, then eventually (in less than infinite time) you’ll always flip the same number of heads as tails.  Even “better”, this will happen an infinite number of times.  If you really keep at it.

However, this is entirely do to the fact that this coin example is the same as the “1 dimensional discrete random walk”.  If you assign “+1″ to heads and “-1″ to tails, then you can keep a running tally with one number.  The fact that, starting with a tally of zero, you’ll eventually return to zero isn’t any kind of “self correcting” property of the universe.  Even slightly different situations (such as one side of the coin being even a tiny, tiny bit more probable) will create “non-recurrent” strings of coin flips.

The tally changes by about the square root of the number of flips (on average), and this can be either an increase or decrease (equal chance).  This is an important part of the “central limit theorem“.  Starting at zero, if you’re flipping 10 coins a minute, then at the end of the day you can expect the difference between the number of heads and tails to be in the neighborhood of \sqrt{10\times60\times24} = 120.

Three days spent filling and counting coins will produce running tallies like this.

Just so you don’t have to do it: Three days spent flipping and counting coins will produce running tallies like this.  The curves are the square root of the number of flips, and most of the time (~64%) the tally will stay between these lines.

Anything between, say, +200 heads or +200 tails is pretty normal, which is a little surprising considering that 200 is pretty small for a full day of coin flipping.  During the course of that day there’s a good chance you’ll “pass zero” several times.

The question of how long it will take for you to get back to the same tally is a little subtle.  If you’re close to zero, then there’s a good chance you’ll hit zero several more times in rapid succession.  But one of those zeros will be the last for a while since inevitably (and completely at random) you’ll get a string of tails or heads that carries you away from zero.  Every possible tally is achieved eventually, and since there are arbitrarily large numbers, there are arbitrarily long return-to-zero times.  That isn’t a rigorous proof, but it turns out that the average return time is infinite.

So basically, if you’re flipping a coin and the total number of heads is equal to the total number of tails, then the same is likely to be true soon.  But if it doesn’t happen soon, then it probably won’t happen again a long time.  Really, the best you can say is that “square root thing”, which is that the tally will usually be within \pm\sqrt{N}, where N is the number of flips.


Answer gravy: For those of you who wanted the details behind that last bit, it turns out to be not too difficult to figure out how long it will take you to get back to zero.

However you get back to zero, you’ll have to flip the coin 2N times (you can’t have an equal number of heads and tails if you flip an odd number of times).  Now say you have a series of tallies that starts at zero, then stays greater than or equal to zero until the 2N flip.  The number of possible ways for this to happen is the Nth Catalan number, CN, which can be described as “the number of ways of arranging N pairs of parenthesis so that they make sense”.  Turns out that C_N = \frac{1}{N+1}{2N \choose N} (this is a choose function).

The probability that the 2Nth coin flip is the first that brings the tally back to zero, P(2N), is the number of paths that didn’t come back to zero before, divided by the total number of paths.

CN is the number of ways for the tally to be greater than or equal to zero.  To be strictly greater than zero, we need a little trick.  Say the first coin flip takes the tally to +1.  If the 2Nth coin flip brings the tally to zero (for the first time), then on the 2N-1th coin flip the tally was +1 as well.  So, the total number of paths starting with “heads” that makes 2N the first return to zero is CN-1 (the number of paths greater than or equal to one, between the first flip and the second to last flip).  Same thing works if the first flip is tails, which means we need to multiply the number of paths by two: 2CN-1.

The total number of possible paths is easy: 22N.  So,

P(2N) = \frac{2C_{N-1}}{2^{2N}} = \frac{1}{N 2^{2N-1}}{2(N-1) \choose N-1} = \frac{1}{2^{2N-1}}\frac{(2N-2)!}{(N-1)!(N-1)!}

Just to double check: notice that P(2) = 1/2, which is exactly what you’d expect (“TH” and “HT”, but not “HH” and “TT”).  If you’re close to, or at zero, then there’s a good chance you’ll be there again in short order, and you’re guaranteed to come back eventually.

The average number of flips to return to zero is:

\sum_{N=1}^\infty 2N \,P(2N) = \sum_{N=1}^\infty \frac{2N(2N-2)!}{2^{2N}(N-1)!(N-1)!}

However!  This is what folk in the math biz call a “divergent series” because it “adds up to infinity”.  Each term in the summation are about equal to \frac{2N(2N-2)!}{2^{2N}(N-1)!(N-1)!} \approx \frac{2}{\sqrt{\pi}}\sqrt{\frac{1}{N}} (using Stirling’s approximation).  While these do add up slower and slower (because you’re more likely to return to zero sooner than later), they still add up to infinity.

So, in a very technical/mathematical sense, if you flip a coin forever, then eventually you’ll get the same number of heads as tails.  Technically the average return time is infinite.  In practice 90% of the time you’ll get an equal number within 64 flips (fact!), and if you don’t then stop anyway.

Posted in -- By the Physicist, Math, Probability | 13 Comments