Chapter 3 Convex functions 凸函数的相关内容
3.1 Basic properties and examples
Definition(定义1): d o m f \mathbf{dom}f domf is a convex set, 0 ≤ θ ≤ 1 0\leq\theta\leq 1 0≤θ≤1, f ( θ x + ( 1 − θ ) y ) ≤ θ f ( x ) + ( 1 − θ ) f ( y ) f(\theta x+(1-\theta)y)\leq\theta f(x)+(1-\theta)f(y) f(θx+(1−θ)y)≤θf(x)+(1−θ)f(y).
-
f f f is concave if − f -f −f is convex.
-
(定义2) f f f is convex ⇔ \Leftrightarrow ⇔ for all x ∈ d o m f x\in\mathbf{dom}f x∈domf and all v v v, g ( t ) = f ( x + t v ) g(t)=f(x+tv) g(t)=f(x+tv) is convex, t ∈ { t ∣ x + t v ∈ d o m f } t\in\{t|x+tv\in\mathbf{dom}f\} t∈{t∣x+tv∈domf}).
-
extended-value extension:
f ~ ( x ) = { f ( x ) x ∈ d o m f ∞ x ∉ d o m f . \tilde{f}(x)=\begin{cases} f(x) & x\in\mathbf{dom}f\\ \infty & x\notin\mathbf{dom}f. \end{cases} f~(x)={f(x)∞x∈domfx∈/domf.- indicator function:
I ~ C ( x ) = { 0 x ∈ C ∞ x ∉ C . \tilde{I}_C(x)=\begin{cases} 0 & x\in C\\ \infty & x\notin C. \end{cases} I~C(x)={0∞x∈Cx∈/C.
- indicator function:
-
(定义3) f f f is differentiable, f f f is convex ⇔ \Leftrightarrow ⇔ d o m f \mathbf{dom}f domf is convex, f ( y ) ≥ f ( x ) + ∇ f ( x ) T ( y − x ) f(y)\geq f(x)+\nabla f(x)^\mathrm{T}(y-x) f(y)≥f(x)+∇f(x)T(y−x) for all x , y ∈ d o m f x,y\in\mathbf{dom}f x,y∈domf.
-
(定义4) f f f is twice differentiable, f f f is convex ⇔ \Leftrightarrow ⇔ d o m f \mathbf{dom}f domf is convex, for all x ∈ d o m f x\in\mathbf{dom}f x∈domf, ∇ 2 f ( x ) ⪰ 0 \nabla^2f(x)\succeq 0 ∇2f(x)⪰0.
-
Examples
- Exponential. e a x e^{ax} eax is convex on R \mathbb{R} R, for any a ∈ R a\in\mathbb{R} a∈R.
- Powers. x a x^a xa is convex on R + + \mathbb{R}_{++} R++ when a ≥ 1 a\geq 1 a≥1 or a ≤ 0 a\leq 0 a≤0, and concave for 0 ≤ a ≤ 1 0\leq a\leq 1 0≤a≤1.
- Powers of absolute value. ∣ x ∣ p \vert x\vert^p ∣x∣p, for p ≥ 1 p\geq 1 p≥1, is convex on R \mathbb{R} R.
- Logarithm. log x \log x logx is concave on R + + \mathbb{R}_{++} R++.
- Negative entropy. x log x x\log x xlogx (either on R + + \mathbb{R}_{++} R++ or on R + \mathbb{R}_+ R+, defined as 0 for x = 0 x=0 x=0) is convex.
- Norms. Every norm on R n \mathbb{R}^n Rn is convex.
- Max function. f ( x ) = m a x { x 1 , … , x n } f(x)=max\{x_1,\ldots,x_n\} f(x)=max{x1,…,xn} is convex on R n \mathbb{R}^n Rn.
- Quadratic-over-linear function. The function f ( x , y ) = x 2 / y f(x,y)=x^2/y f(x,y)=x2/y, with d o m f = R × R + + = { ( x , y ) ∈ R 2 ∣ y > 0 } \mathbf{dom}f=\mathbf{R}\times\mathbf{R}_{++}=\{(x,y)\in\mathbb{R}^2\vert y>0\} domf=R×R++={(x,y)∈R2∣y>0}, is convex.
- Log-sum-exp. The function f ( x ) = log ( e x 1 + ⋯ + e x n ) f(x)=\log(e^{x_1}+\cdots+e^{x_n}) f(x)=log(ex1+⋯+exn) is convex on R n \mathbb{R}^n Rn.
- Geometric mean. The geometric mean f ( x ) = ( ∏ i = 1 n x i ) 1 / n f(x)=(\prod_{i=1}^n x_i)^{1/n} f(x)=(∏i=1nxi)1/n is concave on d o m f = R + + n \mathbf{dom}f=\mathbb{R}_{++}^n domf=R++n.
- Log-determinant. The function f ( X ) = log det X f(X)=\log\det X f(X)=logdetX is concave on d o m f = S + + n \mathbf{dom}f=\mathbf{S}_{++}^n domf=S++n.
-
Sublevel sets: α \alpha α-sublevel set C α = { x ∈ d o m f ∣ f ( x ) ≤ α } C_\alpha=\{x\in\mathbf{dom}f\vert f(x)\leq\alpha\} Cα={x∈domf∣f(x)≤α}. Sublevel sets of a convex function are convex for any value of α \alpha α.
- If f f f is concave, then its α \alpha α-superlevel set { x ∈ d o m f ∣ f ( x ) ≥ α } \{x\in\mathbf{dom}f\vert f(x)\geq\alpha\} {x∈domf∣f(x)≥α} is a convex set.
-
Epigraph: graph { ( x , f ( x ) ) ∣ x ∈ d o m f } \{(x,f(x))\vert x\in\mathbf{dom}f\} {(x,f(x))∣x∈domf}, epigraph e p i f = { ( x , t ) ∣ x ∈ d o m f , f ( x ) ≤ t } \mathbf{epi}f=\{(x,t)\vert x\in\mathbf{dom}f,f(x)\leq t\} epif={(x,t)∣x∈domf,f(x)≤t}, hypograph h y p o f = { ( x , t ) ∣ t ≤ f ( x ) } \mathbf{hypo}f=\{(x,t)\vert t\leq f(x)\} hypof={(x,t)∣t≤f(x)}.
- A function is convex ⇔ \Leftrightarrow ⇔ its epigraph is a convex set.
- A function is concave ⇔ \Leftrightarrow ⇔ its hypograph is a convex set.
-
Jensen’s inequality: f ( E x ) ≤ E f ( x ) f(\mathbf{E}x)\leq\mathbf{E}f(x) f(Ex)≤Ef(x), x x x is a random variable such that x ∈ d o m f x\in\mathbf{dom}f x∈domf with probability one, and f f f is convex.
-
a
θ
b
1
−
θ
≤
θ
a
+
(
1
−
θ
)
b
a^\theta b^{1-\theta}\leq\theta a+(1-\theta)b
aθb1−θ≤θa+(1−θ)b for
a
,
b
≥
0
a,b\geq 0
a,b≥0 and
0
≤
θ
≤
1
0\leq\theta\leq 1
0≤θ≤1, by convexity of
−
log
x
-\log x
−logx.
- ∑ i = 1 n x i y i ≤ ( ∑ i = 1 n ∣ x i ∣ p ) 1 / p ( ∑ i = 1 n ∣ y i ∣ q ) 1 / q \displaystyle\sum_{i=1}^nx_iy_i\leq\left(\sum_{i=1}^n\vert x_i\vert^p\right)^{1/p}\left(\sum_{i=1}^n\vert y_i\vert^q\right)^{1/q} i=1∑nxiyi≤(i=1∑n∣xi∣p)1/p(i=1∑n∣yi∣q)1/q for p > 1 p>1 p>1, 1 / p + 1 / q = 1 1/p+1/q=1 1/p+1/q=1 and x , y ∈ R n x,y\in\mathbb{R}^n x,y∈Rn, by taking a = ∣ x i ∣ p ∑ j = 1 n ∣ x i ∣ p a=\frac{\vert x_i\vert^p}{\sum_{j=1}^n\vert x_i\vert^p} a=∑j=1n∣xi∣p∣xi∣p, b = ∣ y i ∣ q ∑ j = 1 n ∣ y i ∣ q b=\frac{\vert y_i\vert^q}{\sum_{j=1}^n\vert y_i\vert^q} b=∑j=1n∣yi∣q∣yi∣q, θ = 1 / p \theta=1/p θ=1/p and summing over i i i.
-
a
θ
b
1
−
θ
≤
θ
a
+
(
1
−
θ
)
b
a^\theta b^{1-\theta}\leq\theta a+(1-\theta)b
aθb1−θ≤θa+(1−θ)b for
a
,
b
≥
0
a,b\geq 0
a,b≥0 and
0
≤
θ
≤
1
0\leq\theta\leq 1
0≤θ≤1, by convexity of
−
log
x
-\log x
−logx.
3.2 Operations that preserve convexity
- nonnegative weighted sums: a nonnegative weighted sum of convex functions
f
=
w
1
f
1
+
⋯
+
w
m
f
m
f=w_1f_1+\cdots+w_mf_m
f=w1f1+⋯+wmfm is convex.
- a nonnegative weighted sum of concave functions is concave.
- composition with an affine mapping: f : R n → R f:\mathbb{R}^n\rightarrow\mathbb{R} f:Rn→R, A ∈ R A\in\mathbb{R} A∈R, b ∈ R n b\in\mathbb{R}^n b∈Rn, g ( x ) = f ( A x + b ) g(x)=f(Ax+b) g(x)=f(Ax+b), then if f f f is convex, so is g g g; if f f f is concave, so is g g g.
- pointwise maximum and supremum: if
f
1
,
…
,
f
m
f_1,\ldots,f_m
f1,…,fm are convex, then their pointwise maximum
f
(
x
)
=
max
{
f
1
(
x
)
,
…
,
f
m
(
x
)
}
f(x)=\max\{f_1(x),\ldots,f_m(x)\}
f(x)=max{f1(x),…,fm(x)} is also convex.
- If for each y ∈ A y\in\mathcal{A} y∈A, f ( x , y ) f(x,y) f(x,y) is convex in x x x, then g ( x ) = sup y ∈ A f ( x , y ) \displaystyle g(x)=\sup_{y\in\mathcal{A}}f(x,y) g(x)=y∈Asupf(x,y) is convex in x x x.
- The pointwise infimum of a set of concave functions is a concave function.
- support function: S C ( x ) = sup { x T y ∣ y ∈ C } S_C(x)=\sup\{x^\mathrm{T}y\vert y\in C\} SC(x)=sup{xTy∣y∈C} , d o m S C = { x ∣ sup y ∈ C x T y < ∞ } \mathbf{dom}S_C=\{x\vert \sup_{y\in C}x^\mathrm{T}y< \infty\} domSC={x∣supy∈CxTy<∞} is convex.
- almost every convex function can be expressed as the pointwise supremum of a family of affine functions.
- composition:
f
(
x
)
=
h
(
g
(
x
)
)
f(x)=h(g(x))
f(x)=h(g(x))
- scalar composition:
h
:
R
→
R
h:\mathbb{R}\rightarrow\mathbb{R}
h:R→R,
g
:
R
n
→
R
g:\mathbb{R}^n\rightarrow\mathbb{R}
g:Rn→R
- f f f is convex if h h h is convex, h ~ \tilde{h} h~ is nondecreasing, and g g g is convex
- f f f is convex if h h h is convex, h ~ \tilde{h} h~ is nonincreasing, and g g g is concave
- f f f is concave if h h h is concave, h ~ \tilde{h} h~ is nondecreasing, and g g g is concave
- f f f is concave if h h h is concave, h ~ \tilde{h} h~ is nonincreasing, and g g g is convex
- vector composition: h : R k → R h:\mathbb{R}^k\rightarrow\mathbb{R} h:Rk→R, g i : R n → R g_i:\mathbb{R}^n\rightarrow\mathbb{R} gi:Rn→R: similar derivations
- scalar composition:
h
:
R
→
R
h:\mathbb{R}\rightarrow\mathbb{R}
h:R→R,
g
:
R
n
→
R
g:\mathbb{R}^n\rightarrow\mathbb{R}
g:Rn→R
- minimization: f f f is convex in ( x , y ) (x,y) (x,y), C C C is a convex nonempty set ⇒ \Rightarrow ⇒ g ( x ) = inf y ∈ C f ( x , y ) g(x)=\displaystyle\inf_{y\in C}f(x,y) g(x)=y∈Cinff(x,y) is convex in x x x.
- perspective: g ( x , t ) = t f ( x / t ) g(x,t)=tf(x/t) g(x,t)=tf(x/t). f f f is a convex function ⇒ \Rightarrow ⇒ g g g is convex. Similarly, if f f f is concave, then so is g g g.
3.3 The conjugate function
- definition: f ∗ ( y ) = sup x ∈ d o m f ( y T x − f ( x ) ) f^*(y)=\displaystyle\sup_{x\in\mathbf{dom}f}(y^\mathrm{T}x-f(x)) f∗(y)=x∈domfsup(yTx−f(x)), a convex function.
- basic properties:
- Fenchel’s inequality: f ( x ) + f ∗ ( y ) ≥ x T y f(x)+f^*(y)\geq x^\mathrm{T}y f(x)+f∗(y)≥xTy.
- if f f f is convex, and f f f is closed (i.e., e p i f \mathbf{epi}f epif is a closed set), then f ∗ ∗ = f f^{**}=f f∗∗=f.
- f f f is convex and differentiable with d o m f = R n \mathbf{dom}f=\mathbb{R}^n domf=Rn, let z ∈ R n z\in\mathbb{R}^n z∈Rn be arbitrary and define y = ∇ f ( z ) y=\nabla f(z) y=∇f(z), then f ∗ ( y ) = z T ∇ f ( z ) − f ( z ) f^*(y)=z^\mathrm{T}\nabla f(z)-f(z) f∗(y)=zT∇f(z)−f(z).
- g ( x ) = a f ( x ) + b g(x)=af(x)+b g(x)=af(x)+b, a > 0 a>0 a>0, b ∈ R b\in\mathbb{R} b∈R, then g ∗ ( y ) = a f ∗ ( y / a ) − b g^*(y)=af^*(y/a)-b g∗(y)=af∗(y/a)−b; g ( x ) = f ( A x + b ) g(x)=f(Ax+b) g(x)=f(Ax+b), A ∈ R n × n A\in\mathbb{R}^{n\times n} A∈Rn×n nonsingular, b ∈ R n b\in\mathbb{R}^n b∈Rn, then g ∗ ( y ) = f ∗ ( A − T y ) − b T A − T y g^*(y)=f^*(A^{-\mathrm{T}}y)-b^\mathrm{T}A^{-\mathrm{T}}y g∗(y)=f∗(A−Ty)−bTA−Ty.
- If f ( u , v ) = f 1 ( u ) + f 2 ( v ) f(u,v)=f_1(u)+f_2(v) f(u,v)=f1(u)+f2(v), f 1 f_1 f1 and f 2 f_2 f2 are convex functions with conjugates f 1 ∗ f_1^* f1∗ and f 2 ∗ f_2^* f2∗, respectively, then f ∗ ( w , z ) = f 1 ∗ ( w ) + f 2 ∗ ( z ) f^*(w,z)=f_1^*(w)+f_2^*(z) f∗(w,z)=f1∗(w)+f2∗(z).
3.4 Quasiconvex functions
- definition(定义1):
d
o
m
f
\mathbf{dom}f
domf and all sublevel sets
S
α
=
{
x
∈
d
o
m
f
∣
f
(
x
)
≤
α
}
S_\alpha=\{x\in\mathbf{dom}f\vert f(x)\leq\alpha\}
Sα={x∈domf∣f(x)≤α} for
α
∈
R
\alpha\in\mathbb{R}
α∈R are convex.
- a function is quasiconcave if − f -f −f is quasiconvex, i.e., every superlevel set { x ∣ f ( x ) ≥ α } \{x\vert f(x)\geq\alpha\} {x∣f(x)≥α} is convex.
- a function that is both quasiconvex and quasiconcave is called quasilinear.
- basic properties:
- (定义2) f f f is quasiconvex ⇔ \Leftrightarrow ⇔ d o m f \mathbf{dom}f domf is convex, x , y ∈ d o m f x,y\in\mathbf{dom}f x,y∈domf, 0 ≤ θ ≤ 1 0\leq\theta\leq 1 0≤θ≤1, f ( θ x + ( 1 − θ ) y ) ≤ max { f ( x ) , f ( y ) } f(\theta x+(1-\theta)y)\leq\max\{f(x),f(y)\} f(θx+(1−θ)y)≤max{f(x),f(y)}.
- f f f is quasiconvex ⇔ \Leftrightarrow ⇔ its restriction to any line intersecting its domain is quasiconvex.
-
f
:
R
→
R
f:\mathbb{R}\rightarrow\mathbb{R}
f:R→R is quasiconvex
⇔
\Leftrightarrow
⇔ (at least one condition holds)
- f f f is nondecreasing
- f f f is nonincreasing
- ∃ c ∈ d o m f \exists c\in\mathbf{dom}f ∃c∈domf, for t ≤ c t\leq c t≤c and t ∈ d o m f t\in\mathbf{dom}f t∈domf, f f f is nonincreasing; for t ≥ c t\geq c t≥c and t ∈ d o m f t\in\mathbf{dom}f t∈domf, f f f is nondecreasing
- (定义3) f f f is differentiable, f f f is quasiconvex ⇔ \Leftrightarrow ⇔ d o m f \mathbf{dom}f domf is convex, x , y ∈ d o m f x,y\in\mathbf{dom}f x,y∈domf, f ( y ) ≤ f ( x ) ⇒ ∇ f T ( x ) ( y − x ) ≤ 0 f(y)\leq f(x)\Rightarrow\nabla f^\mathrm{T}(x)(y-x)\leq 0 f(y)≤f(x)⇒∇fT(x)(y−x)≤0.
- (定义4)
f
f
f is twice differentiable:
- f f f is quasiconvex ⇒ \Rightarrow ⇒ x ∈ d o m f x\in\mathbf{dom}f x∈domf, y ∈ R n y\in\mathbb{R}^n y∈Rn, y T ∇ f ( x ) = 0 ⇒ y T ∇ 2 f ( x ) y ≥ 0 y^\mathrm{T}\nabla f(x)=0\Rightarrow y^\mathrm{T}\nabla^2 f(x) y\geq 0 yT∇f(x)=0⇒yT∇2f(x)y≥0.
- y T ∇ f ( x ) = 0 ⇒ y T ∇ 2 f ( x ) y > 0 y^\mathrm{T}\nabla f(x)=0\Rightarrow y^\mathrm{T}\nabla^2 f(x) y> 0 yT∇f(x)=0⇒yT∇2f(x)y>0 for all x ∈ d o m f x\in\mathbf{dom}f x∈domf and all y ∈ R n y\in\mathbb {R}^n y∈Rn, y ≠ 0 y\neq 0 y=0, then f f f is quasiconvex.
- operations that preserve quasiconvexity
- nonnegative weighted maximum: f = max { w 1 f 1 , … , w m f m } f=\max\{w_1f_1,\ldots,w_mf_m\} f=max{w1f1,…,wmfm} with w i ≥ 0 w_i\geq 0 wi≥0 and f i f_i fi quasiconvex, is quasiconvex.
- composition:
- g : R n → R g:\mathbb{R}^n\rightarrow\mathbb{R} g:Rn→R is quasiconvex, h : R → R h:\mathbb{R}\rightarrow\mathbb{R} h:R→R is nondecreasing, f = h ∘ g f=h\circ g f=h∘g is quasiconvex.
- f f f is quasiconvex, g ( x ) = f ( A x + b ) g(x)=f(Ax+b) g(x)=f(Ax+b) is quasiconvex.
- f f f is quasiconvex, g ~ ( x ) = f ( ( A x + b ) / ( c T x + d ) ) \tilde{g}(x)=f((Ax+b)/(c^\mathrm{T}x+d)) g~(x)=f((Ax+b)/(cTx+d)) is quasiconvex on { x ∣ c T x + d > 0 , ( A x + b ) / ( c T x + d ) ∈ d o m f } \{x\vert c^\mathrm{T}x+d>0,(Ax+b)/(c^\mathrm{T}x+d)\in\mathbf{dom}f\} {x∣cTx+d>0,(Ax+b)/(cTx+d)∈domf}.
- minimization: f ( x , y ) f(x,y) f(x,y) is quasiconvex jointly in x x x and y y y, C C C is a convex set, then g ( x ) = inf y ∈ C f ( x , y ) g(x)=\displaystyle\inf_{y\in C}f(x,y) g(x)=y∈Cinff(x,y) is quasiconvex.
- representation via family of convex functions: f ( x ) ≤ t ⇔ ϕ t ( x ) ≤ 0 f(x)\leq t\Leftrightarrow\phi_t(x)\leq 0 f(x)≤t⇔ϕt(x)≤0.
3.5 Log-concave and log-convex functions
- definition(定义1):
- log-concave: f ( x ) > 0 f(x)>0 f(x)>0 for all x ∈ d o m f x\in\mathbf{dom}f x∈domf and log f \log f logf is concave.
- log-convex: f ( x ) > 0 f(x)>0 f(x)>0 for all x ∈ d o m f x\in\mathbf{dom}f x∈domf and log f \log f logf is convex.
- (定义2) f : R n → R f:\mathbb{R}^n\rightarrow\mathbb{R} f:Rn→R with convex domain and f ( x ) > 0 f(x)>0 f(x)>0 for all x ∈ d o m f x\in\mathbf{dom}f x∈domf is log-concave ⇔ \Leftrightarrow ⇔ for all x , y ∈ d o m f x,y\in\mathbf{dom}f x,y∈domf and 0 ≤ θ ≤ 1 0\leq\theta\leq 1 0≤θ≤1, f ( θ x + ( 1 − θ ) y ) ≥ f θ ( x ) f 1 − θ ( y ) f(\theta x+(1-\theta)y)\geq f^\theta(x)f^{1-\theta}(y) f(θx+(1−θ)y)≥fθ(x)f1−θ(y).
- properties:
- (定义3) f f f is twice differentiable with d o m f \mathrm{dom}f domf convex. f f f is log-convex ⇔ \Leftrightarrow ⇔ for all x ∈ d o m f x\in\mathbf{dom}f x∈domf, f ( x ) ∇ 2 f ( x ) ⪰ ∇ f ( x ) ∇ T f ( x ) f(x)\nabla^2f(x)\succeq \nabla f(x)\nabla^\mathrm{T} f(x) f(x)∇2f(x)⪰∇f(x)∇Tf(x); f f f is log-concave ⇔ \Leftrightarrow ⇔ for all x ∈ d o m f x\in\mathbf{dom}f x∈domf, f ( x ) ∇ 2 f ( x ) ⪯ ∇ f ( x ) ∇ T f ( x ) f(x)\nabla^2f(x)\preceq \nabla f(x)\nabla^\mathrm{T} f(x) f(x)∇2f(x)⪯∇f(x)∇Tf(x).
- Log-convexity and log-concavity are closed under multiplication and positive scaling.
- Log-convexity is preserved under sums. If f ( x , y ) f(x,y) f(x,y) is log-convex in x x x for each y ∈ C y\in C y∈C, g ( x ) = ∫ C f ( x , y ) d y g(x)=\int_Cf(x,y)\mathrm{d}y g(x)=∫Cf(x,y)dy is log-convex.
- If f : R n × R m → R f:\mathbb{R}^n\times\mathbb{R}^m\rightarrow\mathbb{R} f:Rn×Rm→R is log-concave, then g ( x ) = ∫ f ( x , y ) d y g(x)=\int f(x,y)\mathrm{d}y g(x)=∫f(x,y)dy is a log-concave function of x x x.
3.6 Convexity with respect to generalized inequalities
-
K-nondecreasing: f : R n → R f:\mathbb{R}^n\rightarrow\mathbb{R} f:Rn→R, x ⪯ K y ⟹ x\preceq_Ky\Longrightarrow x⪯Ky⟹ f ( x ) ≤ f ( y ) f(x)\leq f(y) f(x)≤f(y).
-
K-increasing: f : R n → R f:\mathbb{R}^n\rightarrow\mathbb{R} f:Rn→R, x ⪯ K y , x ≠ y ⟹ x\preceq_Ky,x\neq y\Longrightarrow x⪯Ky,x=y⟹ f ( x ) < f ( y ) f(x)< f(y) f(x)<f(y).
-
f f f is differentiable with convex domain is K-nondecreasing ⇔ \Leftrightarrow ⇔ ∇ f ( x ) ⪰ K ∗ 0 \nabla f(x)\succeq_K^* 0 ∇f(x)⪰K∗0.
-
f : R n → R m f:\mathbb{R}^n\rightarrow\mathbb{R}^m f:Rn→Rm is K-convex: for all x x x, y y y, and 0 ≤ θ ≤ 1 0\leq\theta\leq 1 0≤θ≤1, f ( θ x + ( 1 − θ ) y ) ⪯ K θ f ( x ) + ( 1 − θ ) f ( y ) f(\theta x+(1-\theta)y)\preceq_K\theta f(x)+(1-\theta)f(y) f(θx+(1−θ)y)⪯Kθf(x)+(1−θ)f(y).