By using the functional type cone expansion and compression fixed-point theorem in cones, some new and general results on the existence of positive solution for twin singular boundary value problems with damping term are obtained. An example is given to illustrate our results.
1. Introduction
The study of multipoint boundary value problem for linear second-order ordinary differential equation was initiated by I1'in and Moiseev [1] motivated by the work of Bitsadze and Samarskiĭ [2–4] on nonlocal linear elliptic boundary value problem, which is a new area of still fairly theoretical exploration in mathematics. Several authors have expounded on various aspects of this theory; see the survey paper by Gupta et al. [5–7] and the references cited therein. Thereamong, the study of singular boundary value problem for ordinary differential equation has led to several important applications in applied mathematics and physical science, such as the Thomas-Fermi problem
x′′(t)-t-1/2x3/2=0,0<t<1,x(0)=0=x(1),
which appears in determining the electrical potential in an atom. For other application results, we refer to [8–10]. With regards to this, increasing attention is paid to question of singular boundary value problem and has obtained many excellent results of the existence of the positive solution for two multiple points nonlinear singular boundary value problem [11–15]. The main techniques are the upper and lower solutions method [16], the Leray-Schauder continuation theory [17], and the fixed-point theory in cones [18] et al. We would like to mention some results of Ma and O'Regan [13] and Yao [15], which motivated us to consider the singular boundary value problems. In [13], the authors have studied a multipoint boundary value problem
x′′(t)=f(t,x(t),x′(t))+e(t),0<t<1,x′(0)=0,x(1)=∑i=1m-2aix(ξi),
where ξi∈]0,1[, ai∈ℝ, i=1,2,…,m-2, 0<ξ1<ξ2<⋯<ξm-2<1, f:[0,1]×ℝ2→ℝ is a function satisfying Carathéodory's conditions and (1-t)e(t)∈L1]0,1[; the Leray-Schauder continuation theorem leads to the existence of single C1[0,1[ solution. The literature [15] has discussed a second-order boundary value problem
x′′(t)+h(t)f(x(t))=0,0<t<1,ax(0)-bx′(0)=0,ax(1)+bx′(1)=0,
where h(t) is symmetric on ]0,1[ and may be singular at both end points t=0 and t=1. The author has proved the existence of n symmetric positive solutions and established a corresponding iterative scheme, the main tool being the monotone iterative technique.
A powerful tool for proving existence of solution to boundary value problem is the fixed-point theory. In many cases, it is possible to find single, double, or multiple solutions for boundary value problem, and for the same problem, using the various methods, one can obtain different results under some appropriate conditions. To my best knowledge, very little work has been done on the existence of positive solution for boundary value problem by using the functional type cone expansion and compression fixed-point theorem. The aim of this paper is to establish some new and general results on the existence of positive solution to singular boundary value problems with damping term
u′′(t)-λu′(t)+h(t)f(t,u(t))=0,a<t<b,u′(a)-λu(a)=0,γu(b)+δu′(b)=∑i=1naiu(ti),γu(a)+δu′(a)=∑i=1naiu(ti),u′(b)-λu(b)=0,
where λ, γ, δ, ai, ti, (i=1,2,…,n), h, and f satisfy
h:]a,b[→[0,∞[ is continuous function, h(t) may be singular at t=a and/or t=b, and 0<∫abh(s)ds<∞,
f:[a,b]×[0,∞[→[0,∞[ satisfies the Carathéodory condition, that is, for each x∈[0,∞[, the mapping t→f(t,x) is Lebesgue measurable on [a,b] and for a.e. t∈[a,b], the mapping x→f(t,x) is continuous on [0,∞[, and for each r>0, there exists ϕr∈L1[a,b] such that f(t,x)≤ϕr(t) for all u∈[0,r] and for a.e. t∈[a,b].
We will impose some advisable conditions on the nonlinearity f to ensure the existence of at least one positive solution for the above problems. In order to obtain our results, we construct special operator which is the base for further discussion and provide two crucial functionals on cones. Applying the functional type cone expansion and compression fixed-point theorem to the operator and functionals, we obtain some new and general results on the existence of at least one positive solution for the twin singular problems (1.4), (1.5) and (1.4), (1.6). Our results improve and generalize those in [15, 19].
Let α and β be nonnegative continuous functionals on a cone 𝒫 in real Banach space ℬ. For positive numbers r and L, we define the sets
P(α,r)={x∈P:r<α(x)},P(β,L)={x∈P:β(x)<L},P(β,α,r,L)={x∈P:r<α(x),β(x)<L}.
We state the functional type cone expansion and compression fixed-point theorem [20].
Lemma 1.1.
Let 𝒫 be a cone in a real Banach space ℬ, and let α and β be nonnegative continuous functionals on 𝒫. Let 𝒫(β,α,r,L) be a nonempty bounded subset of 𝒫,
T:P(β,α,r,L)¯⟶P
is a completely continuous operator with
infx∈∂P(β,α,r,L)‖Tx‖>0,P(α,r)¯⊆P(β,L),
for 𝒫(β,L). If α(𝒯x)≥r for all x∈∂𝒫(α,r), β(𝒯x)≤L for each x∈∂𝒫(β,L), and for each y∈∂𝒫(α,r), z∈∂𝒫(β,L), θ∈]0,1], and μ∈[1,∞[, the functionals satisfy the properties
α(θy)≤θα(y),β(μz)≥μβ(z),β(0)=0,
then 𝒯 has at least one positive fixed-point x such that
r≤α(x),β(x)≤L.
2. Main Results
Let ℬ be the Banach space C1[a,b] with the norm ∥u∥=max{∥u∥0,∥u′∥1}, where ∥u∥0=supt∈[a,b]|u(t)| and ∥u′∥1=supt∈]a,b[|u′(t)|, and let a cone in ℬP1={v∈B:visnondecreasingon[a,b],v(a)=0}.
For v∈𝒫1, define the operator 𝒯1 by
T1v(t)=∫ath(s)f(s,exp(λs)(c1(v)+∫sbv(ω)exp(-λω)dω))ds,t∈[a,b],
where c1(v)=(1/d1)(δv(b)+∑i=1naiexp(λti)∫tibv(ω)exp(-λω)dω).
Lemma 2.1.
If v∈𝒫1 is a fixed-point 𝒯1, then
u(t):=exp(λt)(c1(v)+∫tbv(ω)exp(-λω)dω)
is one solution of the problem (1.4), (1.5).
Proof.
Suppose that v∈𝒫1 is a fixed-point 𝒯1 and u(t)=exp(λt)(c1(v)+∫tbv(ω)exp(-λω)dω), thus we have
u′(t)-λu(t)=-v(t)=-T1v(t)=-∫ath(s)f(s,exp(λs)(c1(v)+∫sbv(ω)exp(-λω)dω))ds.
Further,
u′′(t)-λu′(t)=-h(t)f(t,exp(λt)(c1(v)+∫ttnv(ω)exp(-λω)dω))=-h(t)f(t,u(t)).
The boundary condition (1.5) is satisfied due to the relation between u, v, and c1(v).
For v∈𝒫1, we define the nonnegative continuous functionals α and β on 𝒫1 by
α(v)=∫t1bv(ω)exp(-λω)dω,β(v)=v(t1).
Lemma 2.2.
Let r1>0. If v∈∂𝒫1(α,r1), then
v(t1)≤exp(λb)b-t1r1,∫abv(ω)exp(-λω)dω≥r1.
Proof.
For v∈∂𝒫1(α,r1), that is, α(v)=r1. Since v(t) nondecreasing on [a,t1], we have
r1=α(v)=∫t1bv(ω)exp(-λω)dω≤∫t1bv(b)exp(-λω)dω≤v(b)exp(-λt1)(b-t1),
so we get (2.7). Furthermore,
∫abv(ω)exp(-λω)dω=∫at1v(ω)exp(-λω)dω+∫t1bv(ω)exp(-λω)dω≥r1.
Lemma 2.3.
Let L1>0. If v∈∂𝒫1(β,L1), then
∫t1bv(ω)exp(-λω)dω≥L1exp(-λb)(b-t1).
Proof.
For v∈∂𝒫1(β,L1), that is to say, β(v)=L1. In view of v(t) is nondecreasing on [a,b], for ω∈[t1,b], we have
v(ω)≥v(t1)=β(v)=L1.
Hence,
∫t1bv(ω)exp(-λω)dω≥L1∫t1bexp(-λω)dω≥L1exp(-λb)(b-t1).
Lemma 2.4.
Let (H1)–(H4) hold, then 𝒯1:𝒫1→𝒫1 is completely continuous.
Proof.
By (H1)–(H4), we observe that 𝒯1v∈C1[a,b], (𝒯1v)(t) is nondecreasing on [a,b], and (𝒯1v)(a)=0, so 𝒯1:𝒫1→𝒫1. Since h(t) may be singular at t=a and/or t=b, we take the arguments to show that 𝒯1 is completely continuous.
Assume that vn,v0∈𝒫1. In view of f satisfying the Carathéodory condition, it is easy to see that
‖vn-v0‖0⟶0impliesthatsups∈Ω|f(s,exp(λs)(c1(vn)+∫sbvn(ω)exp(-λω)dω))-f(s,exp(λs)(c1(v0)+∫sbv0(ω)exp(-λω)dω))(s,exp(λs)(c1(vn)+∫sbvn(ω)exp(-λω)dω))|ds⟶0
as n→∞, where Ω=[a,b] or ]a,b[. Thus, we have
‖T1vn-T1v0‖0=supt∈[a,b]|(T1vn)(t)-(T1v0)(t)|≤∫abh(s)|f(s,exp(λs)(c1(vn)+∫sbvn(ω)exp(-λω)dω))-f(s,exp(λs)(c1(v0)+∫sbv0(ω)exp(-λω)dω))|ds,‖(T1vn)′-(T1v0)′‖1=supt∈]a,b[|(T1vn)′(t)-(T1v0)′(t)|≤supt∈]a,b[(h(t)|f(t,exp(λt)(c1(vn)+∫sbvn(ω)exp(-λω)dω))-f(t,exp(λt)(c1(v0)+∫sbv0(ω)exp(-λω)dω))|).
Therefore,
‖T1vn-T1v0‖⟶0(n⟶∞).
This means that the operator 𝒯1:𝒫1→𝒫1 is continuous.
Choose two sequences {φn}n=1∞, {ψn}n=1∞⊂]a,b[ satisfying φn≤ψn for any n≥1, such that φn→a and ψn→b as n→∞, respectively. Define
hn(t)={infa≤t≤φnh(t),h(t),φn<t<ψninfψn≤t≤bh(t),
and an operator sequence {𝒯1n}n=1∞ by
T1nv=∫athn(s)f(s,exp(λs)(c1(v)+∫sbv(ω)exp(-λω)dω))ds.
Clearly, hn:[a,b]→[0,∞) is a piecewise continuous function, and the operator 𝒯1n:𝒫1→𝒫1 is well defined. Further, we can see that 𝒯1n:𝒫1→𝒫1 is completely continuous.
Let R>0, BR:={v∈𝒫1:∥v∥0≤R}, and MR=sup{f(t,u):(t,u)∈[a,b]×[0,R¯]}, where R¯=(Rexp(λb)/d1)(δ+(1/λ)∑i=1nai(1-exp(-λ(b-ti))))+(R/λ)(expλ(b-a)-1)>0. We will prove that 𝒯1n approach 𝒯1 uniformly on BR. From the absolute continuity of integral, we obtain
limn→∞∫l(n)h(s)ds=0,
where l(n)=[a,φn]∪[ψn,b]. For each v∈BR, t∈[a,φn], we have
‖T1nv-T1v‖0=supt∈[a,φn]|∫at(hn(s)-h(s))f(s,exp(λs)(c1(v)+∫sbv(ω)exp(-λω)dω))ds≤MR∫aφn|hn(s)-h(s)|ds⟶0(n⟶∞).
For each v∈BR, t∈[ψn,b], we have
‖T1nv-T1v‖0=supt∈[ψn,b]|∫at(hn(s)-h(s))f(s,exp(λs)(c1(v)+∫sbv(ω)exp(-λω)dω))ds≤MR∫ab|hn(s)-h(s)|ds⟶0(n⟶∞).
It is easy to see that, for each v∈BR and φn<t<ψn, there is ∥𝒯1nv-𝒯1v∥0→0 as n→∞. Similarly, for any v∈BR, and t∈[a,φn], ]φn,ψn[, [ψn,b], respectively, we can obtain that
‖(T1nv)′-(T1v)′‖1=supt|(hn(t)-h(t))f(t,exp(λt)(c1(v)+∫tbv(ω)exp(-λω)dω))|≤MR|hn(t)-h(t)|⟶0(n⟶∞).
From the above argument, we obtain
‖T1nv-T1v‖=max{‖T1nv-T1v‖0,‖(T1nv)′-(T1v)′‖1}⟶0(n⟶∞).
That is to say, the sequence 𝒯1n is uniformly approximate 𝒯1 on any bounded subset of 𝒫1. Therefore, 𝒯1:𝒫1→𝒫1 is completely continuous.
For convenience, we set
m1=∫t1bexp(-λt)(∫ath(s)ds)dt,M1=∫abh(s)ds,u1=r1exp(λa)(δexp(λa)d1(b-a)+1),u2=L1exp(λt1)δ+Σi=1nai(b-ti)exp(λ(ti-b))d1.
We are now ready to apply a functional type cone expansion and compression fixed-point theorem to the operator 𝒯1 to give the sufficient conditions for the existence of at least one positive solution to the problem (1.4), (1.5).
Theorem 2.5.
Suppose that (H1)–(H4) hold. Assume that there exist positive numbers k1, r1, and L1 with (exp(λb))/(b-t1)r1<L1 such that
f(t,w)≥k1, (t,w)∈[a,b]×[r1,∞[,
f(t,w)≥r1/m1, (t,w)∈[a,t1]×[u1,∞[,
f(t,w)≤L1/M1,(t,w)∈[t1,b]×[u2,∞[,
then the operator 𝒯1 has at least one fixed-point v such that r1≤α(v) and β(v)≤L1, and the problem (1.4), (1.5) has at least one positive solution u such that
u(t)=exp(λt)(c1(v)+∫tbv(ω)exp(-λω)dω).
Proof.
The cone 𝒫1 and operator 𝒯1 are defined by (2.1) and (2.2), respectively. By the properties of operator 𝒯1, it suffices to show that the conditions of Lemma 1.1 hold with respect to 𝒯1. In view of Lemma 2.1, it is not difficult to prove that a fixed point of 𝒯1 is coincident with the solution of the boundary value problem (1.4), (1.5), so we concentrate on the existence of the fixed point of the operator 𝒯1. Set 𝒫1(β,α,r1,L1) is a nonempty bounded subset of 𝒫1. From Lemma 2.4, it can be shown that
T1:P1(β,α,r1,L1)¯⟶P1
is completely continuous by the Arzela-Ascoli lemma. For v∈∂𝒫1(β,α,r1,L1), the assumption (A1) implies that
‖T1v‖0=∫abh(s)f(s,exp(λs)(c1(v)+∫sbv(ω)exp(-λω)dω))ds≥k1∫abh(s)ds,‖(T1v)′‖1=supt∈]a,b[h(t)f(t,exp(λt)(c1(v)+∫tbv(ω)exp(-λω)dω))≥k1supt∈]a,b[h(t).
the hypotheses of (H3) lead to
infv∈∂P1(β,α,r,L)‖T1v‖≥k1min{supt∈]a,b[h(t),∫abh(s)ds}>0.
If v∈𝒫1(α,r1)¯, then β(v)=v(t1)≤(exp(λb))/(b-t1)r1<L1, and so v∈𝒫1(β,L1)¯, that is, 𝒫1(α,r1)¯⊆𝒫1(β,L1). It follows that the conditions of Lemma 1.1 hold with respect to 𝒯1. By the definition of functionals α and β, we can check that the functionals satisfy the properties α(θy)=∫t1bθy(ω)exp(-λω)dω=θ∫t1by(ω)exp(-λω)dω=θα(y) for y∈∂𝒫1(α,r1) and θ∈]0,1], β(μz)=μz(t1)=μβ(z) for z∈∂𝒫1(β,L1) and μ∈[1,∞[, β(0)=0.
We now prove that α(𝒯1v)≥r1, in Lemma 1.1, holds. In fact, if v∈∂𝒫1(α,r1), by the properties of c1(v) and Lemma 2.2, for each t∈[a,t1],
u(t)=exp(λt)(c1(v)+∫tbv(ω)exp(-λω)dω)≥exp(λa)(c1(v)+r1)≥r1exp(λa)(δexp(λa)d1(b-a)+1).
Hence, by the assumption (A2) and (2.28), there is
α(T1v)=∫t1b(T1v)(t)exp(-λt)dt=∫t1bexp(-λt)(∫ath(s)f(s,exp(λs)(c1(v)+∫sbv(ω)exp(-λω)dω))ds)dt≥r1m1∫t1bexp(-λt)(∫ath(s)ds)dt=r1.
Finally, we assert that β(𝒯1v)≤L1, in Lemma 1.1, also holds. If v∈∂𝒫1(β,L1), by Lemma 2.3, fort∈[t1,b],
u(t)=exp(λt)(c1(v)+∫tbv(ω)exp(-λω)dω)≥v(t1)exp(λt1)δ+Σi=1nai(b-ti)exp(λ(ti-b))d1=L1exp(λt1)δ+Σi=1nai(b-ti)exp(λ(ti-b))d1.
The assumption (A3) and (2.30) imply that
β(T1v)=(T1v)(b)=∫abh(s)f(s,exp(λs)(c1(v)+∫sbv(ω)exp(-λω)dω))ds≤L1M1∫abh(s)ds=L1.
To sum up, the hypotheses of Lemma 1.1 are satisfied. Hence, the operator 𝒯1 has at least one fixed point, that is, the problem (1.4), (1.5) has at least one positive solution.
Let the cone
P2={v∈B:visnonincreasingon[a,b],v(b)=0}.
Evidently, 𝒫2⊂ℬ. For v∈𝒫2, define the operator 𝒯2 by
T2v(t)=∫tbh(s)f(s,exp(λs)(c2(v)+∫asv(ω)exp(-λω)dω))ds,t∈[a,b],
where c2(v)=(1/d2)(δv(a)-∑i=1naiexp(λti)∫ativ(ω)exp(-λω)dω).
We only give the preliminary lemmas and result of the problem (1.4), (1.6), the proofs are similar to the above argument.
Lemma 2.6.
If v∈𝒫2 is a fixed-point 𝒯2, then
u(t)=exp(λt)(c2(v)+∫atv(ω)exp(-λω)dω)
is one solution of the problem (1.4), (1.6).
For v∈𝒫2, the nonnegative continuous functionals α and β on 𝒫2 are defined by
α(v)=∫atnv(ω)exp(-λω)dω,β(v)=v(tn).
Lemma 2.7.
Let r2>0. If v∈∂𝒫2(α,r2), then
v(tn)≤exp(λtn)tn-ar2,∫abv(ω)exp(-λω)dω≥r2.
Lemma 2.8.
Let L2>0. If v∈∂𝒫2(β,L2), then
∫atnv(ω)exp(-λω)dω≥L2exp(-λtn)(tn-a).
Lemma 2.9.
Let (H1)–(H4) hold, then 𝒯2:𝒫2→𝒫2 is completely continuous.
For convenience, we set
m2=∫atnexp(-λt)(∫tbh(s)ds)dt>0,M2=∫abh(s)ds>0,u3=r2exp(λb),u4=L2exp(λa)δ-Σi=1nai(ti-a)exp(λ(ti-a))d2.
Theorem 2.10.
Suppose that (H1)–(H4) hold. Assume that δ-Σi=1naiexp(λ(ti-a))(ti-a)>0, then there exist positive numbers k2, r2, and L2 with ((exp(λtn))/(tn-a))r2<L2 such that
f(t,w)≥k2, (t,w)∈[a,b]×[r2,∞[,
f(t,w)≥(r2/m2), (t,w)∈[tn,b]×[u3,∞[,
f(t,w)≤(L2/M2), (t,w)∈[a,tn]×[u4,∞[.
Then the operator 𝒯2 has at least one fixed-point v such that r2≤α(v) and β(v)≤L2, and the problem (1.4), (1.6) has at least one positive solution u such that
u(t)=exp(λt)(c2(v)+∫atv(ω)exp(-λω)dω).
3. Examples
Consider the problems
u′′(t)-u′(t)+h(t)f(t,u(t))=0,0<t<1,u′(0)-u(0)=0,u(1)+2u′(1)=2u(12),u(0)+2u′(0)=2u(12),u′(1)-u(1)=0.
Let
h(t)=1t(1-t),f(t,w)={10-3t+2w+2,w<2,10-3t+2-2,w≥2.
It is easy to check that hypotheses (H1)–(H4) hold. For the problem (3.1), (3.2), by some calculations, we have d1≈4.858, m1≈0.348, and M1≈3.142. Taking k1=0.001, r1=0.2, and L1=2, satisfying the following conditions: f(t,w)≥0.001, (t,w)∈[0,1]×[0.2,∞[, f(t,w)≥r1/m1≈0.575, (t,w)∈[0,1/2]×[0.141,∞[, f(t,w)≤L1/M1≈0.637, and (t,w)∈[1/2,1]×[2.242,∞[. Thus, the hypotheses of Lemma 1.1 are fulfilled, and so the operator 𝒯1 has at least one fixed point, that is to say, the problem (3.1), (3.2) has at least one positive solution. For the problem (3.1), (3.3), by some calculations, we obtain d2≈0.297, m2≈0.348, and M2≈3.142. Taking k2=0.001, r2=0.1, and L2=2, combining with the following conditions: f(t,w)≥0.001, (t,w)∈[0,1]×[0.1,∞[, f(t,w)≥r2/m2≈0.287, (t,w)∈[1/2,1]×[0.272,∞[, f(t,w)≤L2/M2≈0.637, and (t,w)∈[0,1/2]×[2.364,∞[. So the problem (3.1), (3.3) has at least one positive solution.
I1'inV. A.MoiseevE. I.Nonlocal boundary value problem of the second kind for a Sturm-Liouville operator198723979987BitsadzeA. V.On the theory of nonlocal boundary value problems198430810ZBL0586.30036BitsadzeA. V.On a class of conditionally solvable nonlocal boundary value problems for
harmonic functions1985319194ZBL0607.30039BitsadzeA. V.SamarskiĭA. A.On some simple generalizations of linear elliptic boundary value problems196910398400GuptaC. P.Solvability of a three-point nonlinear boundary value problem for a second order ordinary differential equation1992168254055110.1016/0022-247X(92)90179-H1176010ZBL0763.34009GuptaC. P.NtouyasS. K.TsamatosP. Ch.On an m-point boundary-value problem for second-order ordinary differential equations199423111427143610.1016/0362-546X(94)90137-61306681GuptaC. P.TrofimchukS. I.Existence of a solution of a three-point boundary value problem and the spectral radius of a related linear operator199834448950710.1016/S0362-546X(97)00584-11635743ZBL0944.34009AndersonN.ArthursA. M.Analytical bounding functions for diffusion problems with Michaelis-Menten kinetics198547114515310.1016/S0092-8240(85)90009-6783121ZBL0554.92005AronsonD.CrandallM. G.PeletierL. A.Stabilization of solutions of a degenerate nonlinear diffusion problem19826101001102210.1016/0362-546X(82)90072-4678053ZBL0518.35050NachmanA.CallegariA.A nonlinear singular boundary value problem in the theory of pseudoplastic fluids198038227528110.1137/0138024564014ZBL0453.76002AgarwalR. P.O'ReganD.2003Kluwer Academic PublishersAsakawaH.Nonresonant singular two-point boundary value problems20014479180910.1016/S0362-546X(99)00308-91825782ZBL1042.34526MaR.O'ReganD.Solvability of singular second order m-point boundary value problems2005301112413410.1016/j.jmaa.2004.07.0092105925O'ReganD.1994River Edge, NJ, USAWorld Scientificxii+1541286741YaoQ.Existence and iteration of n symmetric positive solutions for a singular two-point boundary value problem2004478-91195120010.1016/S0898-1221(04)90113-72070974ZBL1062.34024De CosterC.HabetsP.Upper and lower solutions in the theory of ODE boundary value problems: classical and recent results199652Institut de Mathématique Pure et Appliquée, Université Catholique de Louvain1781465239ZBL0889.34018MawhinJ.197940Providence, RI, USAAmerican Mathematical Societyv+122NSFCCBMS Regional Conference Series in Mathematics525202LeggettR. W.WilliamsL. R.Multiple positive fixed points of nonlinear operators on ordered Banach spaces197928467368810.1512/iumj.1979.28.28046542951ZBL0421.47033ZhangY.Positive solutions of singular sublinear Emden-Fowler boundary value problems1994185121522210.1006/jmaa.1994.12431283052ZBL0823.34030AveryR. I.AndersonD. R.Fixed point theorem of cone expansion and compression of functional type20028111073108310.1080/102361902900153441942442ZBL1013.47019