∇×E = −∂B/∂t
V − E + F = 2
θ ← θ − η∇L(θ)
det(A − λI) = 0
0.1 + 0.2 != 0.3
Fᵢ = Fᵢ₋₁ + Fᵢ₋₂
sin²θ + cos²θ = 1
1 pc = 3.0857e16 m
sp · 180° · linear
ReLU(x) = max(0, x)
∫ u dv = uv − ∫ v du
z = λ_obs/λ_emit − 1
no surjection X → 2^X
A*: f(n) = g(n) + h(n)
C(n,k) = n!/(k!(n−k)!)
BFS queues · DFS stacks
consistent ⇒ incomplete
∮_γ f(z)dz = 2πi·ΣRes(f)
0x1F8B — gzip, every time
f(x) = Σ f⁽ⁿ⁾(a)(x−a)ⁿ/n!
2 ATP in, 4 ATP out, net 2
C₂H₅OH + 3O₂ → 2CO₂ + 3H₂O
there is no set of all sets
200 · 301 · 404 · 418 · 500
K_m is the [S] at half V_max
c = mᵉ mod n · m = c^d mod n
softmax(z)ᵢ = e^{zᵢ}/Σⱼe^{zⱼ}
Y = λf.(λx.f(x x))(λx.f(x x))
Making the impossible legible.
¬∃H ∀P,x : H(P,x) = halts?(P,x)
meiosis halves · mitosis copies
6CO₂ + 6H₂O + hν → C₆H₁₂O₆ + 6O₂
go func() { ch <- struct{}{} }()
ρ(∂v/∂t + v·∇v) = −∇p + μ∇²v + ρg
P(A ∪ B) = P(A) + P(B) − P(A ∩ B)
AES-256-GCM · nonce reuse is fatal
N₂ + 3H₂ ⇌ 2NH₃ ΔH = −92 kJ/mol
one ball, cut, reassembled into two
∀ε>0 ∃δ>0 : |x−a|<δ ⇒ |f(x)−f(a)|<ε
bias² + variance + irreducible noise
V_m = (RT/zF)·ln([ion]_out/[ion]_in)
x & (x − 1) clears the lowest set bit
C₆H₁₂O₆ + 6O₂ → 6CO₂ + 6H₂O + ~30 ATP
four colors suffice for any planar map
attention(Q,K,V) = softmax(QKᵀ/√d_k)·V
c = 299792458 m/s — exact by definition
newtype State s a = State (s -> (a, s))
Magic is a formula you haven’t read yet.
Na⁺/K⁺ ATPase: three out, two in, one ATP
CH₄ + 2O₂ → CO₂ + 2H₂O ΔH = −890 kJ/mol
the geodesic is straight; the space is not
O_APPEND is atomic; seek-then-write is not
Hb + 4O₂ ⇌ Hb(O₂)₄ — cooperative, sigmoidal
ΔH_rxn = ΣΔH_f(products) − ΣΔH_f(reactants)
p = malloc(n * sizeof *p); if (!p) abort();
the diagonal always escapes the enumeration
hash table: O(1) amortized, O(n) adversarial
glycolysis → pyruvate → acetyl-CoA → citrate
LRU evicts the coldest thing you'll need next
loop() -> receive X -> handle(X), loop() end.
B-tree: fanout wide, depth shallow, disk happy
lock and key was wrong; induced fit was closer
95°C denature · 55°C anneal · 72°C extend · ×35
ΔΨ across the inner membrane turns the synthase
a coffee cup and a doughnut, up to homeomorphism
Dijkstra: relax every edge, settle each node once
static_assert(sizeof(void*) == 8, "64-bit only");
helicase unwinds, polymerase copies, ligase seals
CREATE INDEX idx ON t(col) WHERE col IS NOT NULL;
mergesort: O(n log n) always, at the cost of space
the observable radius is 46.5 Gly and still growing
O(1) < O(log n) < O(n) < O(n log n) < O(n²) < O(2ⁿ)
(defun fact (n) (if (= n 0) 1 (* n (fact (- n 1)))))
IEEE 754 binary64: 1 sign · 11 exponent · 52 fraction
Okazaki fragments run backwards on the lagging strand
quicksort: O(n log n) expected, O(n²) if you're unlucky
correlation constrains causation; it does not supply it
stress the equilibrium and it shifts to relieve the stress
let rec map f = function [] -> [] | h::t -> f h :: map f t
stack grows down · heap grows up · they meet at your peril
64 codons, 20 amino acids, and the redundancy is protective
CAP: pick two, and the network will pick the third for you
∃ infinitely many primes — assume finite, multiply, add one
almost every string is incompressible, and none can be shown to be
MPI_Allreduce(&local, &global, 1, MPI_DOUBLE, MPI_SUM, MPI_COMM_WORLD);
Dijkstra: relax every edge, settle each node once
static_assert(sizeof(void*) == 8, "64-bit only");
helicase unwinds, polymerase copies, ligase seals
(define (map f xs) (if (null? xs) '() (cons (f (car xs)) (map f (cdr xs)))))
mergesort: O(n log n) always, at the cost of space
the observable radius is 46.5 Gly and still growing
O(1) < O(log n) < O(n) < O(n log n) < O(n²) < O(2ⁿ)
(defun fact (n) (if (= n 0) 1 (* n (fact (- n 1)))))
IEEE 754 binary64: 1 sign · 11 exponent · 52 fraction
Okazaki fragments run backwards on the lagging strand
almost every string is incompressible, and none can be shown to be
quicksort: O(n log n) expected, O(n²) if you're unlucky
correlation constrains causation; it does not supply it
MPI_Allreduce(&local, &global, 1, MPI_DOUBLE, MPI_SUM, MPI_COMM_WORLD);
stress the equilibrium and it shifts to relieve the stress
let rec map f = function [] -> [] | h::t -> f h :: map f t
stack grows down · heap grows up · they meet at your peril
64 codons, 20 amino acids, and the redundancy is protective
CAP: pick two, and the network will pick the third for you
∃ infinitely many primes — assume finite, multiply, add one
(define (map f xs) (if (null? xs) '() (cons (f (car xs)) (map f (cdr xs)))))
almost every string is incompressible, and none can be shown to be
CREATE INDEX idx ON t(col) WHERE col IS NOT NULL;
Dijkstra: relax every edge, settle each node once
static_assert(sizeof(void*) == 8, "64-bit only");
mergesort: O(n log n) always, at the cost of space
MPI_Allreduce(&local, &global, 1, MPI_DOUBLE, MPI_SUM, MPI_COMM_WORLD);
the observable radius is 46.5 Gly and still growing
O(1) < O(log n) < O(n) < O(n log n) < O(n²) < O(2ⁿ)
(defun fact (n) (if (= n 0) 1 (* n (fact (- n 1)))))
IEEE 754 binary64: 1 sign · 11 exponent · 52 fraction
Okazaki fragments run backwards on the lagging strand
(define (map f xs) (if (null? xs) '() (cons (f (car xs)) (map f (cdr xs)))))
almost every string is incompressible, and none can be shown to be
quicksort: O(n log n) expected, O(n²) if you're unlucky
correlation constrains causation; it does not supply it
stress the equilibrium and it shifts to relieve the stress
let rec map f = function [] -> [] | h::t -> f h :: map f t
stack grows down · heap grows up · they meet at your peril
MPI_Allreduce(&local, &global, 1, MPI_DOUBLE, MPI_SUM, MPI_COMM_WORLD);
CAP: pick two, and the network will pick the third for you
64 codons, 20 amino acids, and the redundancy is protective
∃ infinitely many primes — assume finite, multiply, add one
(define (map f xs) (if (null? xs) '() (cons (f (car xs)) (map f (cdr xs)))))
almost every string is incompressible, and none can be shown to be
for (i = 0; i < n; i++) for (j = 0; j < n; j++) for (k = 0; k < n; k++) c[i][j] += a[i][k] * b[k][j];
SELECT s.name, count(*) FROM student s JOIN enroll e ON e.sid = s.id GROUP BY s.name HAVING count(*) > 3;
n! ~ √(2πn)(n/e)ⁿ
T_CMB = 2.72548 K
rate = k[A]^m[B]^n
free(p); p = NULL;
∫_ℝ e^{−x²} dx = √π
popcount(0b1011) = 3
A(4,2) = 2^65536 − 3
δ(x) = 0 ∀x≠0, ∫δ = 1
E = E° − (RT/nF) ln Q
d/dx ∫ₐˣ f(t)dt = f(x)
m − M = 5log₁₀(d/10pc)
lim_{n→∞}(1 + 1/n)ⁿ = e
H₀ ≈ 67–73 km s⁻¹ Mpc⁻¹
|G| = |Z(G)| + Σ[G:C(x)]
F(ω) = ∫ f(t)e^{−iωt} dt
1 + 2 + … + n = n(n+1)/2
pH = pKa + log([A⁻]/[HA])
?- append(X, Y, [a,b,c]).
sp³ · 109.5° · tetrahedral
map f xs = [f x | x <- xs]
K_w = [H⁺][OH⁻] = 1.0×10⁻¹⁴
G = 6.67430×10⁻¹¹ m³kg⁻¹s⁻²
AgNO₃ + NaCl → AgCl↓ + NaNO₃
int main(void) { return 0; }
sp² · 120° · trigonal planar
AUG starts · UAA UAG UGA stop
ψ(x) = A e^{ikx} + B e^{−ikx}
σ = 5.670374419×10⁻⁸ W m⁻² K⁻⁴
ds² = −c²dt² + dx² + dy² + dz²
Γ, x:σ ⊢ e:τ ⟹ Γ ⊢ λx.e : σ→τ
⁴He from 4¹H, and the sun burns
data Maybe a = Nothing | Just a
1s² 2s² 2p⁶ 3s² 3p⁶ 4s² 3d¹⁰ 4p⁵
h = 6.62607015×10⁻³⁴ J·s — exact
μ_s N ≥ |F_applied| until it isn't
−70 mV resting · +40 mV peak · 1 ms
fn main() { println!("{}", 0x2A); }
cat /dev/urandom | head -c 32 | xxd
2²⁵⁶ keys — brute force is not a plan
mutex.lock(); defer mutex.unlock();
E. coli doubles every 20 min at 37°C
∀α. α → α has exactly one inhabitant
SELECT count(*) FROM t WHERE id = ?;
TTL 300 · A · AAAA · CNAME · MX · TXT
mutation is random; the filter is not
6 = 1 + 2 + 3, and perfection is rare
telomerase appends TTAGGG and buys time
ζ(s) = Σ n^{−s} = Π_p (1 − p^{−s})^{−1}
∫|ψ|² dx = 1 or the state means nothing
^[a-z0-9._%+-]+@[a-z0-9.-]+\.[a-z]{2,}$
four equations, and light falls out of them
GC content raises Tm and lowers flexibility
T ≈ e^{−2κL}, and the barrier is not a wall
mRNA → tRNA → polypeptide → fold → function
no cloning: ∄U : U|ψ⟩|0⟩ = |ψ⟩|ψ⟩ for all ψ
5′ cap · poly(A) tail · introns spliced out
EXPLAIN ANALYZE — the plan is not the query
turbulence: still unsolved, still everywhere
2²³ gamete combinations before crossing over
Γᵘ_νρ = ½gᵘˢ(∂_ν g_sρ + ∂_ρ g_sν − ∂_s g_νρ)
selection acts on variance it did not create
sha256("") = e3b0c44298fc1c149afbf4c8996fb924…
every integer > 1 factors uniquely into primes
entropy never decreases in an isolated system
xⁿ + yⁿ = zⁿ has no positive solution for n > 2
mov rax, 60 ; xor rdi, rdi ; syscall
p < 0.05 was always an arbitrary line
u(ν,T) = (8πhν³/c³)/(e^{hν/k_BT} − 1)
C₆H₁₂O₆ + 6O₂ → 6CO₂ + 6H₂O + ~30 ATP
x & (x − 1) clears the lowest set bit
four colors suffice for any planar map
attention(Q,K,V) = softmax(QKᵀ/√d_k)·V
c = 299792458 m/s — exact by definition
newtype State s a = State (s -> (a, s))
telomerase appends TTAGGG and buys time
ζ(s) = Σ n^{−s} = Π_p (1 − p^{−s})^{−1}
∫|ψ|² dx = 1 or the state means nothing
backprop is the chain rule, run backwards
Na⁺/K⁺ ATPase: three out, two in, one ATP
CH₄ + 2O₂ → CO₂ + 2H₂O ΔH = −890 kJ/mol
the geodesic is straight; the space is not
O_APPEND is atomic; seek-then-write is not
Hb + 4O₂ ⇌ Hb(O₂)₄ — cooperative, sigmoidal
ΔH_rxn = ΣΔH_f(products) − ΣΔH_f(reactants)
p = malloc(n * sizeof *p); if (!p) abort();
the diagonal always escapes the enumeration
four equations, and light falls out of them
GC content raises Tm and lowers flexibility
T ≈ e^{−2κL}, and the barrier is not a wall
hash table: O(1) amortized, O(n) adversarial
glycolysis → pyruvate → acetyl-CoA → citrate
turbulence: still unsolved, still everywhere
2²³ gamete combinations before crossing over
LRU evicts the coldest thing you'll need next
loop() -> receive X -> handle(X), loop() end.
B-tree: fanout wide, depth shallow, disk happy
lock and key was wrong; induced fit was closer
sha256("") = e3b0c44298fc1c149afbf4c8996fb924…
every integer > 1 factors uniquely into primes
ΔΨ across the inner membrane turns the synthase
95°C denature · 55°C anneal · 72°C extend · ×35
xⁿ + yⁿ = zⁿ has no positive solution for n > 2
a coffee cup and a doughnut, up to homeomorphism
^[a-z0-9._%+-]+@[a-z0-9.-]+\.[a-z]{2,}$
mRNA → tRNA → polypeptide → fold → function
no cloning: ∄U : U|ψ⟩|0⟩ = |ψ⟩|ψ⟩ for all ψ
5′ cap · poly(A) tail · introns spliced out
EXPLAIN ANALYZE — the plan is not the query
selection acts on variance it did not create
Γᵘ_νρ = ½gᵘˢ(∂_ν g_sρ + ∂_ρ g_sν − ∂_s g_νρ)
entropy never decreases in an isolated system
backprop is the chain rule, run backwards
Na⁺/K⁺ ATPase: three out, two in, one ATP
CH₄ + 2O₂ → CO₂ + 2H₂O ΔH = −890 kJ/mol
the geodesic is straight; the space is not
O_APPEND is atomic; seek-then-write is not
Hb + 4O₂ ⇌ Hb(O₂)₄ — cooperative, sigmoidal
ΔH_rxn = ΣΔH_f(products) − ΣΔH_f(reactants)
p = malloc(n * sizeof *p); if (!p) abort();
the diagonal always escapes the enumeration
four equations, and light falls out of them
hash table: O(1) amortized, O(n) adversarial
glycolysis → pyruvate → acetyl-CoA → citrate
turbulence: still unsolved, still everywhere
2²³ gamete combinations before crossing over
LRU evicts the coldest thing you'll need next
loop() -> receive X -> handle(X), loop() end.
B-tree: fanout wide, depth shallow, disk happy
lock and key was wrong; induced fit was closer
sha256("") = e3b0c44298fc1c149afbf4c8996fb924…
every integer > 1 factors uniquely into primes
entropy never decreases in an isolated system
ΔΨ across the inner membrane turns the synthase
95°C denature · 55°C anneal · 72°C extend · ×35
xⁿ + yⁿ = zⁿ has no positive solution for n > 2
a coffee cup and a doughnut, up to homeomorphism
helicase unwinds, polymerase copies, ligase seals
c = 299792458 m/s — exact by definition
newtype State s a = State (s -> (a, s))
telomerase appends TTAGGG and buys time
ζ(s) = Σ n^{−s} = Π_p (1 − p^{−s})^{−1}
GC content raises Tm and lowers flexibility
T ≈ e^{−2κL}, and the barrier is not a wall
mRNA → tRNA → polypeptide → fold → function
no cloning: ∄U : U|ψ⟩|0⟩ = |ψ⟩|ψ⟩ for all ψ
5′ cap · poly(A) tail · introns spliced out
EXPLAIN ANALYZE — the plan is not the query
selection acts on variance it did not create
Γᵘ_νρ = ½gᵘˢ(∂_ν g_sρ + ∂_ρ g_sν − ∂_s g_νρ)
backprop is the chain rule, run backwards
CH₄ + 2O₂ → CO₂ + 2H₂O ΔH = −890 kJ/mol
the geodesic is straight; the space is not
O_APPEND is atomic; seek-then-write is not
Hb + 4O₂ ⇌ Hb(O₂)₄ — cooperative, sigmoidal
ΔH_rxn = ΣΔH_f(products) − ΣΔH_f(reactants)
p = malloc(n * sizeof *p); if (!p) abort();
the diagonal always escapes the enumeration
four equations, and light falls out of them
hash table: O(1) amortized, O(n) adversarial
glycolysis → pyruvate → acetyl-CoA → citrate
CREATE INDEX idx ON t(col) WHERE col IS NOT NULL;
turbulence: still unsolved, still everywhere
2²³ gamete combinations before crossing over
LRU evicts the coldest thing you'll need next
loop() -> receive X -> handle(X), loop() end.
B-tree: fanout wide, depth shallow, disk happy
lock and key was wrong; induced fit was closer
sha256("") = e3b0c44298fc1c149afbf4c8996fb924…
every integer > 1 factors uniquely into primes
entropy never decreases in an isolated system
ΔΨ across the inner membrane turns the synthase
95°C denature · 55°C anneal · 72°C extend · ×35
xⁿ + yⁿ = zⁿ has no positive solution for n > 2
a coffee cup and a doughnut, up to homeomorphism
Dijkstra: relax every edge, settle each node once
static_assert(sizeof(void*) == 8, "64-bit only");
helicase unwinds, polymerase copies, ligase seals
mergesort: O(n log n) always, at the cost of space
CREATE INDEX idx ON t(col) WHERE col IS NOT NULL;
the observable radius is 46.5 Gly and still growing
O(1) < O(log n) < O(n) < O(n log n) < O(n²) < O(2ⁿ)
stress the equilibrium and it shifts to relieve the stress
SU(3) × SU(2) × U(1)
NP ⊆ PSPACE ⊆ EXPTIME
if (p != NULL) *p = v;
∇×B = μ₀J + μ₀ε₀ ∂E/∂t
x(t) = x₀ + v₀t + ½at²
Var(X) = E[X²] − E[X]²
ADP + Pᵢ + energy → ATP
HCl + NaOH → NaCl + H₂O
e^{iθ} = cos θ + i sin θ
v = V_max[S]/(K_m + [S])
two's complement: ~x + 1
ℏ = 1.054571817×10⁻³⁴ J·s
Z₀ = √(μ₀/ε₀) ≈ 376.730 Ω
K(x) = min{|p| : U(p) = x}
1 eV = 1.602176634×10⁻¹⁹ J
□Aᵘ = μ₀Jᵘ in Lorenz gauge
N_A = 6.02214076×10²³ mol⁻¹
R = 8.314462618 J mol⁻¹ K⁻¹
5'-ATGGCGAATTCCGGATCCTAA-3'
for i in 0..n { s += a[i]; }
1 u = 1.66053906660×10⁻²⁷ kg
awk '{s += $1} END {print s}'
f(E) = 1/(e^{(E−μ)/k_BT} + 1)
git log --oneline --graph --all
Rμν − ½Rgμν + Λgμν = (8πG/c⁴)Tμν
segmentation fault (core dumped)
T(n) = 2T(n/2) + Θ(n)
Σ(Xᵢ − μ)/σ√n → N(0,1)
def f(x): return x ** 2
e = 1.602176634×10⁻¹⁹ C
fact n = product [1..n]
Tm ≈ 4(G+C) + 2(A+T) °C
m_p/m_e ≈ 1836.15267343
20 nt guide + NGG motif
Fe₂O₃ + 3CO → 2Fe + 3CO₂
while (true) { tick(); }
H(X) = −Σ p(x)·log₂ p(x)
k_B = 1.380649×10⁻²³ J/K
P + ½ρv² + ρgh = constant
A–T two bonds · G≡C three
φ(n) = n·Π_{p|n}(1 − 1/p)
P(A|B) = P(B|A)·P(A)/P(B)
ẍ + 2γẋ + ω₀²x = F₀cos(ωt)
ribosome = 30S + 50S = 70S
gcd(a,b) = gcd(b, a mod b)
3'-TACCGCTTAAGGCCTAGGATT-5'
ls -la | sort -k5 -n | tail
uud = proton · udd = neutron
ψ(r,θ,φ) = R_nl(r) Y_lm(θ,φ)
²³⁵U + n → ¹⁴¹Ba + ⁹²Kr + 3n
happens-before, or undefined
λ_max·T = 2.897771955e−3 m·K
Ω_m + Ω_Λ + Ω_k = 1
SYN → SYN-ACK → ACK
sed -e 's/foo/bar/g'
d/dt(∂L/∂q̇) − ∂L/∂q = 0
git rebase -i HEAD~3
n(E) = 1/(e^{E/k_BT} − 1)
awk '{s += $1} END {print s}'
f′ = f(v ± v_o)/(v ∓ v_s)
git log --oneline --graph --all
σ = 5.670374419×10⁻⁸ W m⁻² K⁻⁴
[Ar] 3d⁵ 4s¹ — chromium cheats
¬∃H ∀P,x : H(P,x) = halts?(P,x)
Γ, x:σ ⊢ e:τ ⟹ Γ ⊢ λx.e : σ→τ
meiosis halves · mitosis copies
data Maybe a = Nothing | Just a
6CO₂ + 6H₂O + hν → C₆H₁₂O₆ + 6O₂
go func() { ch <- struct{}{} }()
1s² 2s² 2p⁶ 3s² 3p⁶ 4s² 3d¹⁰ 4p⁵
h = 6.62607015×10⁻³⁴ J·s — exact
ρ(∂v/∂t + v·∇v) = −∇p + μ∇²v + ρg
P(A ∪ B) = P(A) + P(B) − P(A ∩ B)
AES-256-GCM · nonce reuse is fatal
N₂ + 3H₂ ⇌ 2NH₃ ΔH = −92 kJ/mol
μ_s N ≥ |F_applied| until it isn't
one ball, cut, reassembled into two
∀ε>0 ∃δ>0 : |x−a|<δ ⇒ |f(x)−f(a)|<ε
−70 mV resting · +40 mV peak · 1 ms
fn main() { println!("{}", 0x2A); }
cat /dev/urandom | head -c 32 | xxd
bias² + variance + irreducible noise
V_m = (RT/zF)·ln([ion]_out/[ion]_in)
2²⁵⁶ keys — brute force is not a plan
E. coli doubles every 20 min at 37°C
∀α. α → α has exactly one inhabitant
TTL 300 · A · AAAA · CNAME · MX · TXT
mutation is random; the filter is not
6 = 1 + 2 + 3, and perfection is rare
K_m is the [S] at half V_max
softmax(z)ᵢ = e^{zᵢ}/Σⱼe^{zⱼ}
Y = λf.(λx.f(x x))(λx.f(x x))
AUG starts · UAA UAG UGA stop
ψ(x) = A e^{ikx} + B e^{−ikx}
f(E) = 1/(e^{(E−μ)/k_BT} + 1)
ds² = −c²dt² + dx² + dy² + dz²
⁴He from 4¹H, and the sun burns
Rμν − ½Rgμν + Λgμν = (8πG/c⁴)Tμν
segmentation fault (core dumped)
[Ar] 3d⁵ 4s¹ — chromium cheats
mutex.lock(); defer mutex.unlock();
¬∃H ∀P,x : H(P,x) = halts?(P,x)
SELECT count(*) FROM t WHERE id = ?;
meiosis halves · mitosis copies
6CO₂ + 6H₂O + hν → C₆H₁₂O₆ + 6O₂
go func() { ch <- struct{}{} }()
1s² 2s² 2p⁶ 3s² 3p⁶ 4s² 3d¹⁰ 4p⁵
h = 6.62607015×10⁻³⁴ J·s — exact
ρ(∂v/∂t + v·∇v) = −∇p + μ∇²v + ρg
P(A ∪ B) = P(A) + P(B) − P(A ∩ B)
AES-256-GCM · nonce reuse is fatal
N₂ + 3H₂ ⇌ 2NH₃ ΔH = −92 kJ/mol
μ_s N ≥ |F_applied| until it isn't
one ball, cut, reassembled into two
mov rax, 60 ; xor rdi, rdi ; syscall
p < 0.05 was always an arbitrary line
u(ν,T) = (8πhν³/c³)/(e^{hν/k_BT} − 1)
∀ε>0 ∃δ>0 : |x−a|<δ ⇒ |f(x)−f(a)|<ε
−70 mV resting · +40 mV peak · 1 ms
fn main() { println!("{}", 0x2A); }
cat /dev/urandom | head -c 32 | xxd
bias² + variance + irreducible noise
V_m = (RT/zF)·ln([ion]_out/[ion]_in)
2²⁵⁶ keys — brute force is not a plan
E. coli doubles every 20 min at 37°C
∀α. α → α has exactly one inhabitant
C₆H₁₂O₆ + 6O₂ → 6CO₂ + 6H₂O + ~30 ATP
x & (x − 1) clears the lowest set bit
mutation is random; the filter is not
TTL 300 · A · AAAA · CNAME · MX · TXT
6 = 1 + 2 + 3, and perfection is rare
four colors suffice for any planar map
attention(Q,K,V) = softmax(QKᵀ/√d_k)·V
∫|ψ|² dx = 1 or the state means nothing
softmax(z)ᵢ = e^{zᵢ}/Σⱼe^{zⱼ}
Y = λf.(λx.f(x x))(λx.f(x x))
AUG starts · UAA UAG UGA stop
ψ(x) = A e^{ikx} + B e^{−ikx}
σ = 5.670374419×10⁻⁸ W m⁻² K⁻⁴
ds² = −c²dt² + dx² + dy² + dz²
Γ, x:σ ⊢ e:τ ⟹ Γ ⊢ λx.e : σ→τ
⁴He from 4¹H, and the sun burns
data Maybe a = Nothing | Just a
git log --oneline --graph --all
Rμν − ½Rgμν + Λgμν = (8πG/c⁴)Tμν
segmentation fault (core dumped)
p < 0.05 was always an arbitrary line
mutex.lock(); defer mutex.unlock();
u(ν,T) = (8πhν³/c³)/(e^{hν/k_BT} − 1)
SELECT count(*) FROM t WHERE id = ?;
6CO₂ + 6H₂O + hν → C₆H₁₂O₆ + 6O₂
mov rax, 60 ; xor rdi, rdi ; syscall
1s² 2s² 2p⁶ 3s² 3p⁶ 4s² 3d¹⁰ 4p⁵
go func() { ch <- struct{}{} }()
h = 6.62607015×10⁻³⁴ J·s — exact
ρ(∂v/∂t + v·∇v) = −∇p + μ∇²v + ρg
P(A ∪ B) = P(A) + P(B) − P(A ∩ B)
AES-256-GCM · nonce reuse is fatal
N₂ + 3H₂ ⇌ 2NH₃ ΔH = −92 kJ/mol
μ_s N ≥ |F_applied| until it isn't
one ball, cut, reassembled into two
∀ε>0 ∃δ>0 : |x−a|<δ ⇒ |f(x)−f(a)|<ε
−70 mV resting · +40 mV peak · 1 ms
fn main() { println!("{}", 0x2A); }
cat /dev/urandom | head -c 32 | xxd
bias² + variance + irreducible noise
V_m = (RT/zF)·ln([ion]_out/[ion]_in)
2²⁵⁶ keys — brute force is not a plan
^[a-z0-9._%+-]+@[a-z0-9.-]+\.[a-z]{2,}$
E. coli doubles every 20 min at 37°C
∀α. α → α has exactly one inhabitant
C₆H₁₂O₆ + 6O₂ → 6CO₂ + 6H₂O + ~30 ATP
x & (x − 1) clears the lowest set bit
mutation is random; the filter is not
TTL 300 · A · AAAA · CNAME · MX · TXT
6 = 1 + 2 + 3, and perfection is rare
SELECT count(*) FROM t WHERE id = ?;
p < 0.05 was always an arbitrary line
u(ν,T) = (8πhν³/c³)/(e^{hν/k_BT} − 1)
four colors suffice for any planar map
attention(Q,K,V) = softmax(QKᵀ/√d_k)·V
∫|ψ|² dx = 1 or the state means nothing
Na⁺/K⁺ ATPase: three out, two in, one ATP
awk '{s += $1} END {print s}'
f(E) = 1/(e^{(E−μ)/k_BT} + 1)
softmax(z)ᵢ = e^{zᵢ}/Σⱼe^{zⱼ}
Y = λf.(λx.f(x x))(λx.f(x x))
AUG starts · UAA UAG UGA stop
ψ(x) = A e^{ikx} + B e^{−ikx}
[Ar] 3d⁵ 4s¹ — chromium cheats
σ = 5.670374419×10⁻⁸ W m⁻² K⁻⁴
ds² = −c²dt² + dx² + dy² + dz²
¬∃H ∀P,x : H(P,x) = halts?(P,x)
meiosis halves · mitosis copies
Γ, x:σ ⊢ e:τ ⟹ Γ ⊢ λx.e : σ→τ
⁴He from 4¹H, and the sun burns
^[a-z0-9._%+-]+@[a-z0-9.-]+\.[a-z]{2,}$
data Maybe a = Nothing | Just a
git log --oneline --graph --all
Rμν − ½Rgμν + Λgμν = (8πG/c⁴)Tμν
segmentation fault (core dumped)
6CO₂ + 6H₂O + hν → C₆H₁₂O₆ + 6O₂
mutex.lock(); defer mutex.unlock();
1s² 2s² 2p⁶ 3s² 3p⁶ 4s² 3d¹⁰ 4p⁵
mov rax, 60 ; xor rdi, rdi ; syscall
go func() { ch <- struct{}{} }()
h = 6.62607015×10⁻³⁴ J·s — exact
ρ(∂v/∂t + v·∇v) = −∇p + μ∇²v + ρg
P(A ∪ B) = P(A) + P(B) − P(A ∩ B)
AES-256-GCM · nonce reuse is fatal
N₂ + 3H₂ ⇌ 2NH₃ ΔH = −92 kJ/mol
μ_s N ≥ |F_applied| until it isn't
one ball, cut, reassembled into two
∀ε>0 ∃δ>0 : |x−a|<δ ⇒ |f(x)−f(a)|<ε
−70 mV resting · +40 mV peak · 1 ms
fn main() { println!("{}", 0x2A); }
cat /dev/urandom | head -c 32 | xxd
bias² + variance + irreducible noise
V_m = (RT/zF)·ln([ion]_out/[ion]_in)
2²⁵⁶ keys — brute force is not a plan
c = 299792458 m/s — exact by definition
newtype State s a = State (s -> (a, s))
telomerase appends TTAGGG and buys time
ζ(s) = Σ n^{−s} = Π_p (1 − p^{−s})^{−1}
GC content raises Tm and lowers flexibility
E. coli doubles every 20 min at 37°C
T ≈ e^{−2κL}, and the barrier is not a wall
mRNA → tRNA → polypeptide → fold → function
no cloning: ∄U : U|ψ⟩|0⟩ = |ψ⟩|ψ⟩ for all ψ
5′ cap · poly(A) tail · introns spliced out
∀α. α → α has exactly one inhabitant
EXPLAIN ANALYZE — the plan is not the query
selection acts on variance it did not create
Γᵘ_νρ = ½gᵘˢ(∂_ν g_sρ + ∂_ρ g_sν − ∂_s g_νρ)
C₆H₁₂O₆ + 6O₂ → 6CO₂ + 6H₂O + ~30 ATP
x & (x − 1) clears the lowest set bit
mutation is random; the filter is not
TTL 300 · A · AAAA · CNAME · MX · TXT
(defun fact (n) (if (= n 0) 1 (* n (fact (- n 1)))))
IEEE 754 binary64: 1 sign · 11 exponent · 52 fraction
four colors suffice for any planar map
Okazaki fragments run backwards on the lagging strand
attention(Q,K,V) = softmax(QKᵀ/√d_k)·V
∫|ψ|² dx = 1 or the state means nothing
c = 299792458 m/s — exact by definition
newtype State s a = State (s -> (a, s))
telomerase appends TTAGGG and buys time
ζ(s) = Σ n^{−s} = Π_p (1 − p^{−s})^{−1}
quicksort: O(n log n) expected, O(n²) if you're unlucky
correlation constrains causation; it does not supply it
^[a-z0-9._%+-]+@[a-z0-9.-]+\.[a-z]{2,}$
backprop is the chain rule, run backwards
Na⁺/K⁺ ATPase: three out, two in, one ATP
Zafetra
Magic is a formula you haven’t read yet.
COMING SOON
Coming soon