VISCOSITY METHODS FOR PIECEWISE SMOOTH SOLUTIONS TO SCALAR CONSERVATION LAWS. 0. Introduction Consider the single hyperbolic conservation laws

Size: px
Start display at page:

Download "VISCOSITY METHODS FOR PIECEWISE SMOOTH SOLUTIONS TO SCALAR CONSERVATION LAWS. 0. Introduction Consider the single hyperbolic conservation laws"

Transcription

1 MATHEMATICS OF COMPUTATION Volue 66, Nuber 28, April 997, Pages S VISCOSITY METHODS FOR PIECEWISE SMOOTH SOLUTIONS TO SCALAR CONSERVATION LAWS TAO TANG AND ZHEN HUAN TENG Abstract. It is proved that for scalar conservation laws, if the flu function is strictly conve, and if the entropy solution is piecewise sooth with finitely any discontinuities which includes initial central rarefaction waves, initial shocks, possible spontaneous foration of shocks in a future tie and interactions of all these patterns, then the error of viscosity solution to the inviscid solution is bounded by Oɛ log ɛ + ɛ inthel -nor, which is an iproveent of the O ɛ upper bound. If neither central rarefaction waves nor spontaneous shocks occur, the error bound is iproved to Oɛ.. Introduction Consider the single hyperbolic conservation laws. u t + fu =, <<, t >, subject to the initial condition.2 u, = u, <<. The viscosity ethod approiating the conservation laws. and.2 is to solve the parabolic equation.3 v ɛ t + fv ɛ = ɛv ɛ, subject to the sae initial condition.4 v ɛ, = u. Here, ɛ>isasall paraeter. Viscosity ethods play an iportant role in theoretical analysis on nuerical ethods for hyperbolic conservation laws. A useful technique for studying the behavior of solutions to difference equations is to odel the difference equation by a differential equation called odified equation, see for eaple Chapter of LeVeque s book [3]. The difference equation was introduced because it is easier to solve nuerically than the partial differential equation PDE. This is true if we want to generate nuerical approiations, but on the other hand it is often easier to predict qualitative behavior of a PDE than of a syste of difference equations. Since any odified equations have siilar Received by the editor Noveber 2, 995 and, in revised for, April 5, Matheatics Subject Classification. Priary 65M, 65M5, 35L65. Key words and phrases. Hyperbolic conservation laws, error estiate, viscosity ethods, piecewise sooth. Research of the first author was supported by NSERC Canada Grant OGP5545. Research of the second author was supported by the National Natural Science Foundation of China and the Science Fund of the Education Coission of China. 495 c 997 Aerican Matheatical Society License or copyright restrictions ay apply to redistribution; see

2 496 TAO TANG AND ZHEN HUAN TENG fors to equation.3 ore precise for is equation 6., see Section 6 for ore discussions, a better understanding of the solution behavior of.3 will help us to understand any difference ethods for.. For eistence and regularity properties for.3, we refer to [9, 6]. In this work, we are concerned with the accuracy of the viscosity ethods. More precisely, we will investigate the L -convergence rate of u v ɛ.itiswellknownthat the L -nor of error decays only like O ɛ if the initial data is piecewise sooth. Consider an advection diffusion equation of the for v t + av = ɛv, a = constant, with u = sign. Its eact solution is v ɛ, t = 4 at/ 4ɛt e y2 dy. π Fro this and the fact that the solution to the pure advection proble is u, t = sign at, it can be shown that u,t v ɛ,t L R =C ɛt for soe constant C independent of ɛ and t. This indicates that the L -convergence rate of the viscosity ethods is only one half for this siple linear proble. In practice, solutions of onotone difference ethods agree very well with the true solutions of related odified equations. It is epected that onotone difference ethods will have siilar convergence rate to that of viscosity ethods. Harten, Hyan and La [7] pointed out that onotone difference ethods are of at ost first-order accuracy and Kuznetsov [] provided an one half order L -convergence rate for BV bounded initial data. Tang and Teng [9] recently proved that the best L -convergence rate for onotone difference ethods to scalar conservation laws is one half if it includes the linear flu case. All these indicate that viscosity ethods and onotone difference ethods share soe coon properties on the L -convergence rate. Although it is proved that half order convergence rate is best possible for both viscosity ethods and onotone difference ethods, it is widely believed that these ethods ay have higher order of convergence rate when applying to conve conservation laws. More precisely, if the flu f is strictly conve,.5 f α>, then higher order greater than one half convergence rate is epected, see Harten [6]. Lucier [4] used one of the onotone schees, Godunov s ethod, to solve Rieann probles for. with a conve flu. He observed an order one convergence rate. Teng and Zhang [2] recently proved that for a special class of initial data, naely piecewise decreasing constants, both viscosity and strictly onotone difference ethods have order one L -convergence rate, provided that the flu f is strictly conve. Likewise, Bakhvalov [] and Harabetian [5] proved a faster convergence rate Oɛ ln ɛ for Rieann probles with a rarefaction wave. Teng and Zhang s work used the eistence of traveling waves obtained by Jennings [8]. It is noted that Jennings results do not cover Godunov s ethod since his proof relies heavily on the strict onotonicity of the schees and differentiability of the flu function. In a recent work of Fan [3], the eistence and structure of traveling License or copyright restrictions ay apply to redistribution; see

3 VISCOSITY METHODS FOR CONSERVATION LAWS 497 waves for Godunov s ethod are investigated. The order one L -convergence rate is established for Godunov s ethod on single shock solutions. In this paper, we also consider. with a strictly conve flu, but with a ore general and practical class of initial data. With this class of initial data, the entropy solution to. ay include initial central rarefaction waves, initial shocks, possible spontaneous foration of shocks in a future tie and interactions of all these patterns. We will establish Oɛ ln ɛ L -convergence rate for the piecewise sooth solutions with finitely any discontinuities. To the best of our knowledge, no global error estiates for piecewise sooth solutions have been obtained, although soe results for Rieann probles [, 3, 5], for piecewise constant solutions [2], and for BV bounded initial data [4, ] are available. For piecewise sooth solutions to hyperbolic systes with finitely any noninteracting shocks, Oɛ local error estiates have been obtained by Goodan and Xin [4]. The reainder of the paper consists of si sections. In Section, we investigate the structures of the solution to. with a class of piecewise sooth initial data. Soe useful estiates for first order and second order derivatives of the solution are obtained. In Section 2, we introduce a traveling wave solution and a stability lea, which play iportant roles in obtaining the L -convergence rate. Using the stability lea, we prove in Section 3 the following estiate: sup u,t v ɛ,t L R CT ɛ log ɛ + ɛ, t T provided that u is piecewise sooth and that f u has a finite nuber of inflection points. In our proof, we will use a atching ethod developed by Goodan and Xin [4] and Liu and Xin [2]. In Section 4, we ake soe rearks on optial error estiates. It is shown that if neither initial central rarefaction waves nor spontaneous shocks occur, then sup u,t v ɛ,t L R CTɛ. t T In Section 5, we derive sharp estiates for viscosity ethods with a special class of initial function, which includes the initial data considered in Teng and Zhang [2]. Loosely speaking, we consider a nonincreasing piecewise sooth solution with finitely any shock interactions. In this case, we obtain the following estiate: sup u,t v ɛ,t L R cɛ, t< for soe c independent of ɛ. Soe possible generalizations will be discussed in the final section.. Solution structures Throughout this paper, the nor denotes the standard L -nor, L R; C denotes a positive constant independent of ɛ; c denotes a positive constant independent of t and ɛ, but with different values at different places. Let au =f u. If = is an inflection point of au and also satisfies ȧu <, then we call a decreasing inflection point. If this inflection point satisfies ȧu, we call it an increasing inflection point. Throughout this License or copyright restrictions ay apply to redistribution; see

4 498 TAO TANG AND ZHEN HUAN TENG paper, we use the following notations: ȧu := d au ; äu := d2 d u := du d ; ü := d2 u d 2. d 2 au ; The behavior and structure of entropy solutions for scalar conve conservation laws have been studied for any years, see for eaple [2,, 5, 7]. It is well known that if the initial function is piecewise C 2 -sooth, then the entropy solutions consist of at ost a countable nuber of C 2 -sooth regions. Tador and Tassa [8] proved that if the initial speed has a finite nuber of decreasing inflection points, then it bounds the nuber of future shock discontinuities. We ake the following assuptions on the initial data: A: u is bounded and piecewise C 2 -soothwithafinitenuberof discontinuous points γ i, i I; u γ i ± and u γ i ± eist and are finite; A2: ü L R; A3: li ȧu = ; A4: äu changes signs a finite nuber of ties, i.e. au has a finite nuber of inflection points. Reark.. In order to obtain an estiate for u Lea.3, we have to assue that u is piecewise C 2. The only place to use this assuption is to obtain the estiate.3. Otherwise, a weak assuption on u, i.e. u is piecewise C, will be sufficient. Furtherore, the eistence of ü γ i ± is not required. That is, ü γ i ± can be infinity. We now introduce following notations: Denote the set of points where ȧu has a negative iniu by ζ p, p P. Without loss of generality, we assue that ȧu ζ p are distinct: ȧu ζ < ȧu ζ 2 < <ȧu ζ P. It is straightforward to etend our results to the case when there are soe ζ p,ζ q,p q, such that ȧu ζ p = ȧu ζ q. Denote t =andt p = /ȧu ζ p,p =,,P. For a fied T>t P,we let t P + = T. Reark.2. A negative iniu point ay for a shock at a future tie. However, by no eans will every negative inial point eventually generate a new shock. It is an easier case if the negative inial point does not lead to a shock. Without loss of generality, we assue that each point in the set {ζ p } corresponds to a new shock created at a later tie. The starting point of the new shock is p,t p. Since u is assued to be only piecewise C 2, it ay be discontinuous and therefore will not have a classical derivative. We refer by ȧu to the generalized derivative of au. In the sooth interval γ i,γ i+, ȧu is the classical derivative. In positive and negative jup points of u,ȧu is positive and negative infinity, respectively. In particular, in negative jup points i.e. decreasing discontinuities License or copyright restrictions ay apply to redistribution; see

5 VISCOSITY METHODS FOR CONSERVATION LAWS 499 of u,ȧu has a negative infinite iniu. If ȧu has a continuu of negative inial points it is considered as one iniu. The assuption A4 indicates that ȧu has a finite nuber of decreasing inflection points. It can be shown that under the assuptions A, A3 and A4 only a finite nuber of shocks will occur see, e.g. [7, 8]. The corresponding entropy solution consists of a finite nuber of C 2 -sooth pieces. The shocks are disjoint, ecept for coon endpoints arising fro collisions, and without loss of generality we assue that no ore than two shocks ever collide. More precisely, for each tie interval [t p,t p+ ] there eist a finite nuber of subintervals, t p,n t < t p,n+,n =,,,N p, where t p, = t p,t p,np = t p+, such that in each interval [t p,n,t p,n+ there are a finite nuber of sooth curves = X p,n t,=,,m p,n, satisfying. For t t p,n,t p,n+, X p,n t <Xp,n + t, =,,M p,n ; 2. there eists at least one such that X p,n t p,n+ =X p,n + tp,n+ ; 3. u, t issoothecepton=x p,n 4. ux p,n ±,t eist and satisfy either ux p,n X p,n is a shock, or X p,n,t=ux p,n discontinuity. Moreover, for t t p,n,t p,n+, Ẋ p,n t = fux p,n ux p,n f ux p,n t, t p,n t<t p,n+ ;,t>ux p,n +,t fuxp,n,t +,t ux p,n,t if ux p,n +,t, i.e. X p,n, +,t>ux p,n,t,,t, if ux p,n +,t=ux p,n,t, +,t, i.e. is a contact where Ẋt =dxt/dt; 5. Each shock X p,n t continues to the net subinterval [t p,n+,t p,n+2, with a possibility that it collides with another wave at t = t p,n+.consequently, each shock curve can be etended to t =. Fro the above results it follows that the solution of. and.2 is a finite cobination of the cases plotted in Figure. To proceed, we need ore inforation on shock and rarefaction discontinuities. Shocks will be fored in the following cases: i if there is an initial negative jup; ii if ȧu has a negative iniu in an interval where u is C 2 sooth; iii if u has positive jup at a point, = z say, and ȧu z+ or ȧu z is a negative iniu. A shock ay also occur as a consequence of an interaction of two waves. If = X t and=x + t are interacted with each other, then we denote the resulting shock still as = X t. It is regarded as an etension of = X t or =X + t, rather than a new shock. Consider first two cases. Assue = Xt is a shock curve. Before = Xt is interacted by other waves, say t p <t<t p,n, we can trace two characteristic lines License or copyright restrictions ay apply to redistribution; see

6 5 TAO TANG AND ZHEN HUAN TENG Figure. Solution structures fro = Xt backward to t =:. Xt=ζ +au ζ t = ζ + + au ζ + t, where ζ + t >ζ t. Differentiating both sides of the first identity in. with respect to t gives.2 or equivalently.3 Observe that.4 ζ = Ẋt au ζ +ȧu ζ t, ζ := dζ t, dt +ȧu ζ t = ζ Ẋt au ζ. Ẋt au ζ = = aθu + + θu dθ au f ξθdθu + u, License or copyright restrictions ay apply to redistribution; see

7 VISCOSITY METHODS FOR CONSERVATION LAWS 5 where u ± = uxt ±,t, ξ denotes soe interediate value. Therefore, we have Ẋt au ζ c u + u c au + au = c t ζ+ ζ, where c is soe constant independent of t, and in the last step we have used.. The above results lead to +ȧu ζ t ct ζ ζ + ζ. Graphically, it is easy to see that ζ is a decreasing function of t and ζ + is an increasing function of t. Therefore, we have ζ = ζ ζ + ζ.itisalsotrue that ζ + >ζ. Therefore, we have +ȧu ζ t ct ζ + ζ ζ + ζ. A siilar result holds for ζ +. Consequently, we conclude for both Cases i and ii that +ȧu ζ ± t ct ζ + ζ.5 ζ + ζ, where c is a constant independent of t, and the inequality holds before = Xt collides with another shock. Reark.3. The inequality.5 is an iportant result. Assue that = Xt is not interacted by other waves for t S, T. It follows fro.5 that T.6 S +ȧu ζ ± t dt CT logζ + T ζ T + logζ + S ζ S. For a finite T,wehave ζ + S ζ S ζ + T ζ T C. To bound the right-hand side of.6, we only need to find the lower bound for ζ + S ζ S. See, for eaple, step in the proof of Lea.. In Case iii, a new shock is generated on the boundary of a central rarefaction wave. Since one side of the resulting shock is within the rarefaction wave, one of the two characteristic lines fro any point on the shock will end up at the positive jup point. Consider, for eaple, u has positive jup at = z and ȧu z is a negative iniu in the sense that ȧu z < ȧu for in the left neighborhood of z. There we will have a central rarefaction wave with boundaries = X L t and=x R t. Starting at the point X L t z,t z, where t z = /ȧu z, a new shock = Xt will be generated. It satisfies.7 Now.3 holds, with Ẋt = Xt =ζ +au ζ t. aθu + + θu dθ, License or copyright restrictions ay apply to redistribution; see

8 52 TAO TANG AND ZHEN HUAN TENG where u + = a Xt z/t,u =uxt,t. It follows fro.4 and.7 that Ẋt au ζ c u + u c au + au = c t Xt z au ζ t = c t ζ z. Therefore, we have.8 +ȧu ζ t ct ζ ζ z. Siilarly, if u has positive jup at = z and ȧu z+ is a negative iniu, we have +.9 ct ζ +ȧu ζ + t ζ + z. Reark.4. Siilar to Reark.3, it follows fro.8 that T. S +ȧu ζ t dt CT logζ T z + logζ S z. Therefore, it is iportant to find a bound for ζ S z. See, for eaple, step in the proof of Lea.. A siilar idea will be used to treat.9. Fro the above discussion, we can obtain the following lea which is iportant in the error analysis. Lea.. Assue = Xt is a shock curve satisfying Ẋt = fu+ fu. u + u, u ± := uxt ±,t. Assue au is sufficiently sooth in the neighborhood of its negative iniu points. If = Xt is fored at t =,then T.2 u Xt±,t dt C. If = Xt is fored at t = t p >, then T.3 u Xt±,t dt C log δ + C, t p+δ provided that δ>is sufficiently sall. Here T>t P isafiednuber. Proof. The proof of this lea consists of three steps. Step : Initial negative jup. In this case, a shock is fored due to initial negative jup at a point, say = b. It is obvious that this shock is not going to be interacted by other waves before a finite tie t,n, n N.Itisknownthat u ζ ± +au ζ ± t, t = u ζ ±, <t<t,n. Differentiating both sides with resepect to ζ + or ζ gives.4 u Xt ±,t= u ζ ± +ȧu ζ ± t. License or copyright restrictions ay apply to redistribution; see

9 VISCOSITY METHODS FOR CONSERVATION LAWS 53 Since ȧu ± is bounded by a constant independent of, there eists a constant η such that u Xt±,t cfor all <t<η. If this shock collides with another wave at t = t,n,wechooseapositiveγsatisfying γ<inη, t,n. Obviously,.5 γ u Xt ±,t dt C. Again, since ζ is decreasing and ζ + is increasing, we have ζ + t ζ t ζ + γ ζ γ =constant>, for t γ,t,n. Using Reark.3 and the above inequality, we can obtain that.6 t,n γ u Xt ±,t dt C. At t = t,n, if the shock is interacted by another shock, then the resulting shock satisfies ζ + t ζ t ζ + γ ζ γ =constant>, for t t,n,t,n, where t,n t,n,t is the net possible shock interaction tie. Again, using Reark.3 and the above inequality gives t,n u Xt ±,t dt C..7 t,n At t = t,n, if the shock is interacted by a central rarefaction wave, u, t = a z/t, at t = t,n, say on the left boundary as in Figure G, then we have z ζ t z ζ t,n z ζ γ =constant>, for t t,n,t,n. Using Reark.4 and the fact that u Xt+,t C/t for t t,n,t,n gives the sae estiate as.7. Repeating this procedure a finite nuber of ties, we can obtain.2. Step 2: Negative iniu point in a sooth doain. Assue that ζ p z,z 2 is a negative iniu point of ȧu, and au is sufficiently sooth in z,z 2. Then there is a shock curve, = Xt, fored at p,t p, where p = ζ p + au ζ p t p with t p = /ȧu ζ p. Fro each point away fro the curve, we can trace a characteristic line backward in tie to t =. Using La geoetric condition, we know that these lines will not end up with the discontinuous points. Let = ζ + au ζt denote a characteristic line passing through an, t point. We point out that ζ is a function of and t. Forafiedtiet,itiseasyto see that a larger value of corresponds to a larger value of ζ. Thisgives.8 ζ =+ȧu ζt, ȧu ζ := d au ζ. dζ See also Daferos [2] for a siilar result. Notice that ζ p is a negative iniu point of ȧu ζ. This suggests that there eists a positive integer s such that a k u ζ p =, 2 k 2s, anda 2s+ u ζ p >, where a k u := dk d k au. License or copyright restrictions ay apply to redistribution; see

10 54 TAO TANG AND ZHEN HUAN TENG Moreover, there eists a constant γ> such that.9 <a 2s+ u ζ 2a 2s+ u ζ p, for ζ ζ p <γ. Fro.8 and Taylor s epansion, we obtain +ȧu ζ p t + 2s! a2s+ u ξζ ζ p 2s t, where ξ is between ζ and ζ p. Using the fact that ȧu ζ p = /t p, we obtain that.2 t + t p 2s! a2s+ u ξζ ζ p 2s t. If ζ ζ p <γ, we obtain fro.9 and.2 that ζ ζ p Ct t p 2s, tp <t<t p,n, where t p,n t p+ is the sallest tie level when = Xt collides with other waves t p,n = t p+ if this does not happen. Therefore, we have proved that for all ζ near ζ p, { }.2 ζ ζ p in γ,ct t p 2s, t p <t<t p,n. Tracing two characteristic lines fro = Xt backward to t =, see., we obtain fro.2 { }.22 ζ + t ζ t in γ,ct t p 2s, t p <t<t p,n. Fro.4,.22 and Reark.3, we obtain t p,n u Xt ±,t dt C log δ + C, δ+t p where δ> is sufficiently sall. By using the sae arguent as used in step of this proof, we can etend the above integration to the tie interval t p + δ, T. Step 3: Negative iniu point at a positive jup endpoint. Without loss of generality, we assue that u has a positive jup at = z, andȧu z is a negative iniu. Let X L t be the left boundary of the central rarefaction wave and t p = /ȧu z. Then a shock = Xt isforedatthepoint p,t p, where p = X L t p. Assuing that au ζ is sufficiently sooth as ζ z and using Taylor s epansion, we have { } z ζ t in γ,ct t p 2s, where s is a positive constant. This, together with Reark.4, yield that t,n u Xt,t dt C log δ + C, δ+t p where t,n indicates the sallest tie level when the shock collides with another wave. It is clear that u, t =f z/t for Xt + and hence u Xt+,t C/t. We can then conclude that.3 is true for this case. License or copyright restrictions ay apply to redistribution; see

11 VISCOSITY METHODS FOR CONSERVATION LAWS 55 Reark.5. It is observed fro the proof that if the discontinuities are purely due to the initial negative jup, then the O log δ factor is not included in the estiate for the integral of u, see.2. The optial error estiate sees possible in the case when the solution u contains neither a central rarefaction wave nor an original shock generated at a later tie. This will be investigated in Section 4. Reark.6. Assue = Xt is a shock curve generated at t =ort=t p and = ζ ± + au ζ ± t are the corresponding characteristic lines. We have proved in the above lea that.23 T t p+δ +ȧu ζ ± dt C log δ + C. t The partial derivative u will also be discontinuous at a point z if there is an initial positive jup at z, i.e. u z + > u z. This corresponds to a central rarefaction wave. The case that the rarefaction wave collides with a shock is included in Step of the proof for Lea.. Here we are interested in the central rarefaction wave before it is interacted by a shock. Lea.2. Assue a central rarefaction wave is fored at = z. Let=X L t and = X R t be left and right boundaries of the rarefaction wave, respectively. If ȧu z +is not a negative iniu, then u X R t+,t C, u X R t,t Ct. The above results hold before the rarefaction wave is interacted by a shock. ȧu z +is a negative iniu, then u X R t+,t Ct t p, u X R t,t Ct, for t<t p,wheret p = /ȧu z+. The curve = X R t will becoe a shock after t = t p. Siilar results, based on ȧu z, hold for X L t. Proof. Observe that u, t =f z t, X L t X R t, where X L t =z+f u z t, X R t =z+f u z+t. The above result gives that u X R t,t Ct. Moreover, if ȧu z + orȧu z+ < but not a local iniu, then we can show that u X R t+,t Cbefore = X R t collides with other waves. Now consider the case that ȧu z +< is a local iniu. Then a new shock will be fored at t p = /ȧu z +. Siilar to.4 we have u z+ u X R t+,t= +ȧu z+t, <t<t p. This and the fact that t p = /ȧu z + yield The proof is coplete. u X R t+,t Ct p t, <t<t p. If License or copyright restrictions ay apply to redistribution; see

12 56 TAO TANG AND ZHEN HUAN TENG Reark.7. Let = X R t be the right boundary of a central rarefaction wave. It ay collide with a shock at a certain tie, say t = t. We still denote the resulting shock as X R t. Hence, we have etended the curve = X R t tothewholetie interval [,T]. Using Leas. and.2 gives tp+ δ.24 u X R t ±,t dt C log δ + C. t p+δ Moreover, fro each point on = X R t we can trace a right characteristic line = ζ + + au ζ + t back to t =. Here ζ + = z +beforex R t is interacted by a shock. Fro the proofs for Leas. and.2, we can show that tp+ δ.25 t p+δ +ȧu ζ + dt C log δ + C. t We ay etend = X L t in a siilar way and trace a left characteristic line = ζ + au ζ t back to t =. Here, ζ = z before = X L t collides with a shock. We also have tp+ δ.26 u X L t ±,t dt C log δ + C,.27 t p+δ tp+ δ t p+δ +ȧu ζ dt C log δ + C. t In order to provide a sharp L -error bound, soe properties on second-order derivatives are also needed, e.g. see Sections 3.2 and 3.3. Lea.3. Assue that u satisfies the assuptions A A4 and let au be sufficiently sooth near its negative iniu points. Then the solution to. and.2 satisfies.28 tp+ δ t p+δ provided δ is sufficiently sall. u,t dt C log δ + C, p P, Proof. For any fied tie t t p,t p+, there are finite intervals Z l t,z l+ t the nuber of the intervals can be bounded by a constant independent of t such that the transforation = ζ + au ζt aps Z l t,z l+ t in space to θ l t,θ l+ t in ζ space. The solution u, t is continuously differentiable in Z l,z l+ but not at the end points Z l t andz l+ t. It is observed that in Z l t =in l θ l =, and a l l Z l t =aθ l =. l In the case that Z l t andz l+ t are boundaries of a central rarefaction wave, there eists an integer s such that θ s t =θ s+ t =z s and the solution is of the for u, t =f zs, Z l t Z l+ t, t where Z l+ t Z l t [f u z s + f u z s ]t. It follows fro f α> that [f ] ζ C for ζ C. Hence, we have Zl+ t Zl+ t d u, t d C t 2 = C Z l+ t Z l t t 2 Ct, Z l t Z l t License or copyright restrictions ay apply to redistribution; see

13 VISCOSITY METHODS FOR CONSERVATION LAWS 57 which leads to tp+ Zl+ t t p+δ Z l t u, t ddt C log δ + C. If θ l <θ l+,wehave+ȧu ζt >forζ θ l,θ l+. Differentiating the equation uζ + au ζt, t =u ζ with respect to ζ gives u ζ + au ζt, t + ȧu t = u äu t + ȧu t 2 + ü +ȧu t. It follows fro the above result that Zl+ t Z l t u, t d = θl+ t θ l t θl+ t θ l t u ζ + au ζt, t + ȧu t dζ u äu t θl+ t + ȧu t 2 dζ + θ l t ü +ȧu t dζ =: I + I 2. Observe that äu t = ζ +ȧu t + ȧu t 2. This suggests that the aiu of / + ȧu t possibly occurs at i increasing inflection points of au ; ii decreasing inflection points of au ; and iii the end points θ l t orθ l+ t. For the possibility i, we have / + ȧu t. For the possibility ii, noting that + ȧu t>, we have +ȧu t a. t j>t t/t j Since t t p,t p+, we obtain fro the above result that +ȧu t C t p+ t. Consequently, we obtain for the above three possibilities that.29 +ȧu t + C t p+ t + +ȧu θ l t + +ȧu θ l+ t. Notice that there is no new shock being generated for t t p,t p+. If = Z l t corresponds to a shock curve, then θ l t =ζ + t following our notation in Lea.. Then using Reark.6 we have tp+ dt C log δ + C. t p+δ +ȧu θ l t If = Z l t is a right boundary of a rarefaction wave and collides with a shock at t = t t p,t p+, then using Reark.7 gives tp+ δ dt C log δ + C. t p+δ +ȧu θ l t Siilar results hold for the last ter in.29. These, together with the assuption A2, yield tp+ δ t p+δ I 2 dt C log δ + C. License or copyright restrictions ay apply to redistribution; see

14 58 TAO TANG AND ZHEN HUAN TENG Observe that b äu t c + ȧu t 2 dζ = +ȧu ct +ȧu bt, provided that au C 2 b, c. It follows fro A3, A4 and the above identity that.3 θl+ t äu t θ l t + ȧu t 2 dζ +ȧu θ l tt + +ȧu θ l+ tt + j 2 +ȧu y j t, where y j θ j,θ j+ are inflection points of au. Siilar to the discussion for I 2, we have tp+ δ θl+ t äu t dζdt C log δ + C. t p+δ θ l t + ȧu t 2 Using A and A2, we can show that u is bounded alost everywhere. This and the above inequality yield that tp+ δ t p+δ The proof of Lea.3 is thereby coplete. I dt C log δ + C. 2. Stability and traveling wave Our error estiates are based on a stability lea for nonhoogeneous viscous equations and a traveling wave lea. We first introduce the stability lea. Lea 2.. Let v i, t,i =,2,be continuous and piecewise sooth solutions of the following equations: 2. v i + fv i ɛ v i = g i, t, t > a, i =,2. t The above equation holds for all values of ecept on soe curves X t, M, wherev i ay not eist. If w := v v 2 as,then 2.2 w,t w,a + +ɛ M = Proof. It follows fro 2. that 2.3 t a t a g,τ g 2,τ dτ w X τ+,τ w X τ,τ dτ. w t +f ζw ɛw = g, t g 2, t, where ζ is soe interediate value between v and v 2. If w orw for all, then straightforward integration on the above equation gives 2.2. Let p t <p t<p 2 t< be the points such that at those points w changes signs. License or copyright restrictions ay apply to redistribution; see

15 VISCOSITY METHODS FOR CONSERVATION LAWS 59 Let α j be the sign of w in p j,p j+. Multiplying 2.3 by α j and integrating the resulting equation over p j,p j+ gives pj+ 2.4 α j w t d = ɛ α j w p j+,t α j w p j +,t + p j p j<x <p j+ ɛα j pj+ +α j g, t g 2, t d. p j w X t+,t w X,t Since wp j,t=wp j+,t=andα j w for p j,p j+, we have d pj+ pj+ w d = α j w t d. dt p j p j Moreover, observing that α j w p j+,t andα j w p j +,t, we obtain fro 2.4 that d pj+ w d w X +,t w X,t dt p j ɛ p j<x <p j+ pj+ + p j ɛ p <X <p p g, t g 2, t d. Since the above inequality is true for all j, we have d p w d w X +,t w X,t dt p + p g, t g 2, t d, where p =sup j p j.ifp <, using a siilar ethod as above gives d w d ɛ w X t+,t w X,t dt p X >p + p g, t g 2, t d. In obtaning the last inequality, we have used the fact that w as.using the above two inequalities, we obtain w, t d ɛ t w X +,τ w X,τ dτ p a X >p t + g, τ g 2, τ ddτ + w, a d. a p p A siilar result which replaces p, by,p, respectively, in the above inequality, can be established. These results lead to 2.2. Assue X t is a sooth curve satisfying the Rankine-Hugoniot jup condition: Ẋ t = fu+ fu 2.5 u + u, License or copyright restrictions ay apply to redistribution; see

16 5 TAO TANG AND ZHEN HUAN TENG where Ẋt :=dx/dt is the shock speed, u+ = ux t+,,u = ux t, with u + <u.ifu ± are independent of t, it is shown in [2] that there is a traveling wave solution of the for V ɛ St, where the wave speed S is a constant, satisfying.3 and V ɛ =u,v ɛ =u +. In the case when u± are functions of t, we need the following generalization. Lea 2.2. Let u + t <u tbe two given functions and X t be defined by 2.5. LetVζ;u +,u be defined iplicitly by ζ = V 2 u+ +u [ Φu; u +,u ] du, Φu; u +,u =fu fu Ẋu u. With respect to ζ, the function V ζ; u +,u is a decreasing function satisfying 2.8 V ; u +,u =u, V ; u +,u = 2 u+ +u, V ;u +,u =u +. It also satisfies the following inequalities V ζ; u +,u Hζ;u +,,u u 2.9 u + ep{ αu u + ζ /2}, V ɛ X ;u +,u H X ;u +,u 2. 4 α ɛ, where α is defined by.5. In the above inequalities, V ɛ ζ; u +,u := Vζ/ɛ;u +,u, H is the so-called Heaviside function defined by { H; u +,u = u + if, u if <. Denote V ɛ X t; u +,u byvɛ. Direct calculation shows that 2. V ɛ t + f V ɛ ɛ V ɛ = V ɛ u + u + + V ɛ u u, where V ɛ indicate the partial derivatives for V ɛ with respect to the paraeters u ± ɛ u ±, respectively. Since V H X ; u +,u asɛ +see 2., it is epected that the right-hand side of 2. approaches H X t; u +, u as ɛis sufficiently sall. Lea 2.3. Let V ɛ ζ; u +,u be defined by that in Lea 2.2. We have V ɛ u u ; +,u u V ɛ 2.2 u ;u +,u u H ; u +, u Cɛ u X +,t + u X,t, where u ± = du ± /dt. We defer the proofs of the above two leas to the Appendi. License or copyright restrictions ay apply to redistribution; see

17 VISCOSITY METHODS FOR CONSERVATION LAWS 5 3. Main result and outline of proof In the preceding sections we were concerned with the solution structure, traveling wave solution and stability lea. Using these results, we are now ready to state and prove the ain result of this paper. Theore 3.. Let the flu f be strictly conve. Assue that the initial data u satisfies the requireents A-A4 stated in Section and that au is sufficiently sooth near its negative iniu points. If v ɛ is the solution to.3 and.4 and u is the solution to. and.2, then the following error estiate holds: 3. sup t T v ɛ,t u,t CT where CT is a constant independent of ɛ. ɛ log ɛ + ɛ, Before giving a proof for the above estiate, we point out a direct application of Theore 3.. Corollary 3.. Let the flu f be strictly conve and u C 2 R. If au has a finite nuber of inflection points and is sufficiently sooth at the decreasing inflection points, then the following error estiate holds: 3.2 sup v ɛ,t u,t CT ɛ log ɛ + ɛ. t T Reark 3.. Loosely speaking, the above result iplies that for scalar conve conservation laws, if the initial data u is sooth and f u has a finite nuber of inflection points, then the L -convergence rate of viscosity ethods is Oɛ log ɛ. The class of initial data described in Corollary 3. is quite general, and is widely used in practical coputations. We will use Lea 2. to prove Theore 3.. An essential requireent in Lea 2. is that the solution to 2. should be continuous. It is known that the solution of.3 is continuous for t> if the initial data is piecewise sooth. However, the solution of. ay be discontinuous for t> even if the initial data is sooth. Therefore, we cannot apply Lea 2. directly to equations. and.3. As discussed before, there are two kinds of discontinuity, naely rarefaction wave and shock wave for solutions of. and.2. Across the rarefaction region, the solution is continuous, but on the shock curves, the solution is discontinuous. We need to construct a reasonable approiation to u so that it can get rid of the shock discontinuities. We will consider the cases when u possesses shock discontinuity. For any interval t p,t p+, p P here t =,t P+ = T, we will construct v ɛ,p, t, an approiate function to u, t, such that B: v ɛ,p, t CR t p,t p+, v ɛ,p, t v ɛ, as ± ; B2: v ɛ,p, t is piecewise sooth and satisfies 3.3 v ɛ,p t +f v ɛ,p ɛ v ɛ,p =ḡ, t in its sooth regions; where the right-hand side function satisfies 3.4 tp+ ɛ t p+ɛ ḡ,t dt Cɛ log ɛ + Cɛ; License or copyright restrictions ay apply to redistribution; see

18 52 TAO TANG AND ZHEN HUAN TENG B3: u and v ɛ,p are discontinuous on the sae curves, naely, they are discontinuous on X t, M p,for, t R t p,t p+ see Section for the definition of X ; here for ease of notation we ignore the superscript p and n. Moreover, 3.5 v ɛ,p X t+,t v ɛ,p X t,t = u X t+,t u X t,t ; B4: For t t p,t p+, 3.6 v ɛ,p,t u,t Cɛ. It is known that v ɛ, the solution to.3 and.4, belongs to C R t p,t p+. If v ɛ,p satisfies B, B2 and B3, using Lea 2. gives v ɛ,t v ɛ,p,t v ɛ,t p +ɛ v ɛ,p,t p +ɛ + t +ɛ u X τ+,τ u X τ,τ dτ, t p+ɛ for t t p + ɛ, t p+ ɛ. Using Leas.,.2 and 3.4 gives t t p+ɛ ḡ,τ dτ v ɛ,t v ɛ,p,t v ɛ,t p +ɛ v ɛ,p,t p +ɛ +Cɛ log ɛ + Cɛ, for t t p + ɛ, t p+ ɛ. It follows fro the above inequality and B4 that v ɛ,t u,t v ɛ,t p +ɛ u,t p +ɛ +Cɛ log ɛ + Cɛ, for t t p + ɛ, t p+ ɛ. Since u and v ɛ satisfy the following stability results, we obtain u,s u,s 2 C s s 2, v ɛ,s v ɛ,s 2 C s s 2, v ɛ,t u,t v ɛ,t p u,t p +Cɛ log ɛ + Cɛ, t [t p,t p+ ]. Noting that the above inequality is true for all p P, we obtain that for any t [,T], v ɛ,t u,t v ɛ, u, + Cɛ log ɛ + Cɛ = Cɛ log ɛ + Cɛ. To prove Theore 3., what reains is to construct v ɛ,p, p P, satisfying B B4. We will concentrate on t p,t p+, one of the intervals in P p= t p,t p+. Reark 3.2. Once again we point out that the ain difficulty in using Lea 2. is that the solution u ay be discontinuous. This happens when u has shock discontinuities. If the solution u contains contact discontinuities only i.e. rarefaction waves, the solution is still continuous and Lea 2. can be used directly. Therefore, the construction of v ɛ is to reove shock discontinuities and no efforts need be ade to reove the contact discontinuities. For ease of notation, we denote v ɛ,p by v ɛ in the reainder of this section. License or copyright restrictions ay apply to redistribution; see

19 VISCOSITY METHODS FOR CONSERVATION LAWS Zero shock. In this case, no shock occurs in the doain R t p,t p+. We siply choose v ɛ = u. Since there is no shock discontinuity for u, wehave v ɛ = u CR t p,t p+. It is straightforward to verify that this v ɛ satisfies the requireents B, B3 and B4. It is also obvious that ḡ, t in 3.3 equals ɛu. Using Lea.3 we can verify that this ḡ satisfies 3.4. Therefore, the requireent B2 is also satisfied One shock. In this case, we assue that u is sooth in R t p,t p+, ecept on = X t. Here the function X t satisfies 2.5. As entioned before, Lea 2. cannot be applied directly since u CR t p,t p+. In order to overcoe the difficulty, we introduce the following approiate solution: 3.7 [ v ɛ, t =u, t+ V ɛ X ;u +,u H X ;u + ],u. We will verify that this v ɛ satisfies the requireents B-B4. First, we observe that v ɛ X +,t v ɛ X,t=V ɛ +; u +,u Vɛ ; u +,u =, which indicates that v ɛ CR t p,t p+. It is also easy to see that v ɛ u as ±, which iplies that v ɛ v ɛ as ±. Hence, B is true. We can also show that B3 is true. Using Lea 2.2 will give B4. We now need to show that v ɛ satisfies B2. Proof of B2. We denote V ɛ X ; u +,u byvɛ,andh X ;u +,u by H. It follows fro 2. that V ɛ t + f V ɛ ɛ V ɛ = V ɛ u + u + + V ɛ This, together with., gives v ɛ t +f v ɛ ɛ v ɛ =ḡ, t, t t p,t p+, where ḡ, t =I +I 2 ɛu, I = fu f V ɛ + f v ɛ, I 2 = u + + V ɛ u H X t; u +, u. V ɛ u + u Notice that [H X t; u +,u ] =forallecept on the curve = X t. Away fro = X t, we have I = f v ɛ u + V ɛ f uu f V ɛ V ɛ = f v ɛ f u u + f v ɛ f V ɛ V ɛ = f ξ V ɛ H u + f ξ 2 u H V ɛ =: I + I 2, where ξ and ξ 2 are soe interediate values. For >X t, although there are no shock curves across the interval X,, there ay eist a finite nuber of central rarefaction waves in the interval. Let = Y l t, l L, denote left u u. License or copyright restrictions ay apply to redistribution; see

20 54 TAO TANG AND ZHEN HUAN TENG or right boundaries of the rarefaction waves for t t p,t p+. If >X t but Y l t, u, t =u X t+,t+ u Y l t+,t u Y l t,t + u, td. l X It follows fro the above result that, for >X t, 3.8 u, t u X t+,t + l Siilarly, we can show that, for <X t, 3.9 u, t u X t,t + l u Y l t±,t + u,t. u Y l t±,t + u,t. The above results and 2. lead to I,t Cɛ u X t ±,t + l u Y l t±,t + u,t. An application of Leas.,.2 and.3 yields tp+ ɛ t p+ɛ I,t dt Cɛ log ɛ + Cɛ. Using the facts that [H X t; u +,u ] =,a.e. and V ɛ, we have 3. I 2,t C u H H V ɛ d. For any X t, we can show that u, t H a u z ±,t X t. z R For X t, using 3.8 and 3.9 yields u, t H u X t±,t + u Y l t±,t + u,t X t. l Hence, fro 3. we have I 2,t C u X t±,t + l X t u Y l t±,t + u,t H V ɛ d. Using integration by parts and Lea 2.2, we can show that the integral on the right-hand side can be bounded by Cɛ. Further, using Leas.,.2 and.3 gives tp+ ɛ t p+ɛ I 2,t dt Cɛ log ɛ + Cɛ. License or copyright restrictions ay apply to redistribution; see

21 VISCOSITY METHODS FOR CONSERVATION LAWS 55 We have therefore obtained an estiate for I. An application of Leas. and 2.3 yields tp+ ɛ t p+ɛ I 2,t dt Cɛ log ɛ + Cɛ. Using the estiates for I and I 2, together with Lea.3, we establish the inequality 3.4. Reark 3.3. Fro the definition of ḡ, t, we find that it is necessary to estiate u. Therefore, Lea.3 is useful in obtaining our error estiates Two shocks. In this case, we have two sooth shock curves, X t and X + t, satisfying one of the following possibilities: P: X + t >X tfort [t p,t p+ ], see Case H in Figure ; P2: X + t >X tfort [t p,t p+ andx t p+ =X + t p+, see Case D in Figure ; P3: there eists a tie level t p, t p,t p+ such that there are two shock waves for t t p,t p,, but the two shocks eet at t = t p, and for one shock for t [t p,,t p+, see Cases C and D in Figure. For the first two possibilities, we introduce the following approiate solution: [ v ɛ, t =u, t+ V ɛ X t; u +,u H X t; u + ] 3.,u [ + V ɛ X + t; u + + +,u H X+ t; u ],u. We can verify that v ɛ satisfies the requireents B, B3 and B4. Again, we will show that v ɛ satisfies B2. Proof of B2. We denote V ɛ X t; u +,u andh X t; u +,u by V ɛ and H, respectively. It follows fro 2. that V ɛ t + f V ɛ ɛ V ɛ + V ɛ + t + f V ɛ + ɛ V ɛ + = V ɛ u + + V ɛ u + V ɛ + u V ɛ + u +. u + This, together with., gives where u u + + v ɛ t + f v ɛ ɛ v ɛ =ḡ, t, t t p,t p+, u ḡ, t =I +I 2 ɛu, I = fu f V ɛ f V ɛ + + f v ɛ, I 2 = V ɛ u + u + + V ɛ u u H X t; u +, u + V ɛ + u + u V ɛ + u + u + + H X + t; u + +, u +. License or copyright restrictions ay apply to redistribution; see

22 56 TAO TANG AND ZHEN HUAN TENG We only need to estiate I ; other ters on the right-hand side of 3.2 can be handled by Leas 2.3,.,.2 and.3. Observe that I = f v ɛ u + V ɛ + V ɛ + f uu f V ɛ V ɛ f V ɛ + V ɛ + = f ξ v ɛ uu +f ξ 2 v ɛ V ɛ V ɛ +f ξ 3 v ɛ V ɛ + V ɛ + =: I + I 2 + I 3, where ξ,ξ 2,ξ 3 are soe interediate values. Siilar to the proof of 3.8 and 3.9, we have 3.3 u ±,t u X ±,t + u X + ±,t + u Y l t ±,t + u,t, l for any, t R t p,t p+, where = Y l t, l L, denotes left or right boundaries of possible rarefaction waves. This, together with Lea 2.2 ore precisely 2., leads to tp+ ɛ t p+ɛ Observe that I C u H =: J + J 2. Fro 2.9 we have J 2,t =C Cβ + ep I,t dt Cɛ log ɛ + Cɛ. V ɛ V ɛ + H + X + t β + 2ɛ [ = Cβ + + X +t δt/2 Cβ + [ X + δ/2 Cβ + [V V ɛ δt 2ɛ ; u+,u V ɛ + C V + H + V ɛ d V ɛ ] X +t δt/2 d +ep 3δt 2ɛ ;u+,u d δt β + 4ɛ +β ep V ɛ V ɛ β + δt 4ɛ d ], where β = u u +,β + = u + u+ +,δt=x +t X t. It follows fro 2.9 that δt 3δt V 2ɛ ; u+,u V 2ɛ ;u+,u δt V 2ɛ ;u+,u u + δt β ep β. 4ɛ ] License or copyright restrictions ay apply to redistribution; see

23 VISCOSITY METHODS FOR CONSERVATION LAWS 57 Consequently, we obtain 3.5 } δt δt J 2,t Cβ + β {ep β +ep β +. 4ɛ 4ɛ Consider the following function: t p+ t zt = X + t X t, t [t p,t p+ ]. It is obvious that zt C[t p,t p+. For the first possibility, P, X + t X t > for all t [t p,t p+ ], which indicates that z C[t p,t p+ ]. For the second possibility, P2, we have u + t p+ =u + t p+. It follows fro the definition 2.5 that 3.6 Ẋ t p+ Ẋ+t p+ = f ξθdθ u t p+ u + + t p+, where ξ is an interediate value. Since X + t p+ =X t p+, we have <ζ + + t p+ ζ t p+ =t p+ au t p+ au + + t p+ C u t p+ u + + t p+. This and 3.6 lead to Fro this and L Hospital s rule, we have Ẋ t p+ Ẋ+t p+ C>. li zt = t t p+ Ẋ t p+ Ẋ+t p+ C. This indicates that z C[t p,t p+ ]. Therefore, for both possibilities P and P2, there eists a constant γ>, such that zt γfor all t [t p,t p+ ]. It follows fro the definition of zt that δt=x + t X t Ct p+ t, t p t t p+. Fro this and 3.5, we have J 2,t Cβ + β { ep Cβ + t p+ t/ɛ+ep Cβ t p+ t/ɛ This yields tp+ t p J 2,t dt Cɛ. The estiate for J is a little sipler than that for J 2. It follows fro the facts { az<x+t u u, t H z ±,t X t, if <X + t, C, if X + t, }. License or copyright restrictions ay apply to redistribution; see

24 58 TAO TANG AND ZHEN HUAN TENG that J,t =C u H C a u z ±,t z<x +t +C X + V ɛ X+ d V ɛ C a u z ±,t z<x +t +C V δt/ɛ; u +,u u +. d X t V ɛ X t H d V ɛ Using integration by parts and Lea 2.2, we can show that the last integral can be bounded by Cɛ. Siilar to the proof for J 2,wehave <V δt/ɛ; u +,u u + β ep Cβ t p+ t/ɛ. These results, together with 3.3, yield tp+ ɛ t p+ɛ J,t dt Cɛ log ɛ + Cɛ. Therefore, for both possibilities P and P2, we have tp+ ɛ t p+ɛ I 2,t dt C log ɛ + Cɛ. In a siilar anner we can show that the above estiate holds also for I 3. Hence, we have proved that in cases P and P2 the function v ɛ given by 3. satisfied the requireent B2. If the possibility P3 is the case, naely there are two shock curves, X t and X + t, for t t p,t p,, and one shock, X t, for t t p,,t p+, we construct v ɛ in the following way: 3.7 u, t+ V ɛ H + v ɛ = u, t+ d V ɛ + H +, if t t p,t p,, V ɛ H, if t [t p,,t p+. This case can be regarded as a cobination of possibility P2 analyzed above and the one shock case analyzed in the last subsection. It is then not difficult to verify that this v ɛ satisfies the requireents B-B More shocks. Assue that we have M shock curves X t, M for t t p,t p,, where t p, t p+. These shocks ay eerge and there will be a lesser nuber of shocks for t t p,,t p+. Siilar to the last two subsections, we let 3.8 v ɛ, t =u, t+ M = V ɛ X t; u +,u H X t; u +,u, for, t R t p,t p,. We found that it is straightforward to etend the ethod of proof in the last subsection to show that the function v ɛ given by 3.8 satisfies License or copyright restrictions ay apply to redistribution; see

25 VISCOSITY METHODS FOR CONSERVATION LAWS 59 the requireents B-B4. It is found that the following estiates are particularly useful: M 3.9 v ɛ,t u,t H Cɛ; = V ɛ M 3.2 u ±,t u X t ±,t + u Y l t±,t + u,t ; = l v ɛ,t V ɛ k V ɛ 3.2 k u,t H k V ɛ k + V ɛ H V ɛ k k a u z ±,t Cɛ + Cβ k ep Cβ k t p+ t/ɛ X k <z<x k+ + { } β β k ep Cβ t p+ t/ɛ+ep Cβ k t p+ t/ɛ, k for k M, whereβ =u u+,β k = u k u+ k. HerewehavesetX t=,x M+ t =. For t [t p,,t p+, the nuber of shock curves is less than M and we ay use atheatical induction based on the nuber of shock curves to show that an appropriate v ɛ can be found cf. [2]. Therefore, Theore 3. is proved. Reark 3.4. Under the present fraework, the factor O log ɛ cannotbereoved fro our results. We found the factor is due to the estiates for u and u,see Leas. to.3. In soe special cases, these estiates can be iproved so that optial convergence rates can be obtained. 4. Rearks on sharp estiate In this section, we consider the case when the initial function u satisfies the following conditions: C: u is bounded and piecewise C 2 -sooth with a finite nuber of discontinuous points γ i, i I; u γ i ± eist and are finite; u γ i > u γ i +; C2: ü L R; C3: li ȧu = ; C4: au has no decreasing inflection points, but ay have a finite nuber of increasing inflection points. The assuption C4 iplies that there is no original shock generated after t =. The assuption C iplies that there is no central rarefaction waves fored at t =. Therefore, only discontinuities are initial shocks and the interactions of the initial shocks. Assue = Xt is a shock curve. Following Lea., we can show that T T u Xt ±,t dt C, +ȧu ζ ± dt C. t License or copyright restrictions ay apply to redistribution; see

26 52 TAO TANG AND ZHEN HUAN TENG We now use notations introduced in Section. The interval [,T] can be divided into several subintervals: = t < t < < t N = T. In each interval [t n,t n+, there are a finite nuber of sooth curves = X n t satisfying X n t <Xn + t, =,,M n. Fro each point on = X n t, we can trace two characteristic lines back to t = : X n t=ζ +au ± ζ t. ± Using.29 and the assuption C4 gives +ȧu ζt + +ȧu ζ t + + +ȧu ζ+ t, for ζ + ζ ζ+ with =,,M n, where we have set ζ + =,ζ M =. n+ It follows fro 4.2 that the integral for the right-hand side functions over [,T] can be bounded by a constant C. Following the proof for Lea.3 we can show that 4.3 T u,t dt C. Using and repeating the procedures in the last section yield the following theore. Theore 4.. Let the flu f be strictly conve. Assue that the initial data u satisfies the requireents C-C4 stated above. If v ɛ is the solution to.3 and.4 and u is the solution to. and.2, then the following error estiate holds: 4.4 sup v ɛ,t u,t CTɛ, t T where CT is a constant independent of ɛ. It is easy to show that Oɛ convergence rate is the best possible one for viscosity ethods. The above result indicates that if the solution to. and.2 includes neither central rarefaction waves nor spontaneous shocks, then viscosity ethods yield the optial convergence rate. 5. Results on nonincreasing initial data Throughout this section, c denotes a positive constant independent of t and ɛ, but with different values at different places. We consider a special case when the initial function satisfies the following conditions: D: u is bounded, nonincreasing and piecewise C 2 sooth with a finite nuber of discontinuous points γ < <γ I ; u γ i ± eist and are bounded; D2: ü does not change sign when Λ, with Λ a positive constant; D3: li ȧu = ; D4: äu has no inflection points. Since äu does not change sign in each of its sooth doains, we obtain fro D and D3 that äu L R. The assuption D also indicates that u ± is bounded for Λ. Observe that äu =a u u 2 + a u ü, License or copyright restrictions ay apply to redistribution; see

27 VISCOSITY METHODS FOR CONSERVATION LAWS 52 where a u =f u α>. The above results yield that ü L Λ, Λ. This, together with D2 and D3, iply that ü L R. The assuption D also indicates that u γ i > u γ i +. Since u is nonincreasing, it is known that there eists a tie level T such that there is only one shock after t = T. The solution is sooth away fro this shock for t>t. Fro each point on the shock curve, we can trace two characteristic lines Xt =ζ ± +au ζ ± back to t =. Itisobviousthat 5. ζ t<γ, ζ + t >γ I, T t<. It follows fro D that u ζ u γ, u ζ + u γ I +, T t<, which leads to 5.2 u ζ u ζ + u γ u γ I +=constant>, for T t<.fro.2,.4 and 5.2, we have 5.3 ζ c +ȧu ζ t c>, wherewehaveusedthefactȧu due to D. Since ζ <, we obtain that 5.4 li t ζ t =. Moreover, fro D, D3 and D4, we have 5.5 äu ζ, ζ,γ, and äu ζ, ζ γ I,. We need to prove the following results for soe finite values T,T 2 >T : 5.6 u Xt,t dt c, u Xt+,t dt c; T T 2 ζ t u äu t 5.7 u äu t dζdt c, dζdt c; T + ȧu t 2 T 2 ζ + t + ȧu t 2 ζ t ü 5.8 ü dζdt c, dζdt c. T +ȧu t T 2 ζ + t +ȧu t It follows fro 5.4 that there eists a tie level t = T >T such that ζ t Λ fort T. Fro.4 and 5.3, we obtain 5.9 u Xt,t dt c ζ t u ζ dt T T c u u ζ T c, where in the second inequality we have used the facts that ζ and u ζ. Hence the first inequality in 5.6 is proved. Further, using 5.5 gives ζ t äu t + ȧu t 2 dζ = +ȧu ζ t +ȧu ζ t. It follows fro D2 and D3 that u ζ u ζ for ζ,ζ. License or copyright restrictions ay apply to redistribution; see

28 522 TAO TANG AND ZHEN HUAN TENG The above two inequalities lead to ζ t u äu t + ȧu t 2 dζ u ζ +ȧu ζ t = u Xt,t. This and 5.9 yield the first inequality in 5.7. Using D2 and D3 gives ζ t ü dζ = u ζ, for t T. This and 5.5 yield ζ t ü +ȧu t dζ u ζ +ȧu ζ t = u Xt,t. Consequently, the first inequality in 5.8 is proved. The second inequalities in can be proved in a siilar way. Let T =a{t,t 2 }. Using and following the proofs for Leas. and.3, we can show that 5. T u Xt ±,t dt c, Repeating the proof in Section 3., we can obtain 5. v ɛ,t u,t v ɛ,t u,t +cɛ, T u,t dt c. t [T,, where c is independent of t and ɛ. Since the conditions C C4 include D D4, Theore 4. iplies that 5.2 v ɛ,t u,t CTɛ, Cobining 5. and 5.2 gives the following theore. t [,T]. Theore 5.. Let the flu f be strictly conve. Assue that the initial data u satisfies the requireents D-D4 stated above. If v ɛ is the solution to.3 and.4 and u is the solution to. and.2, then the following error estiate holds: 5.3 sup t< v ɛ,t u,t cɛ, where c is a constant independent of ɛ. 6. Generalizations In this section, we will briefly ention two possible generalizations. The first one is a straightforward etension. The ideas in this paper can be applied to a ore general viscosity equation of the for 6. v ɛ t + fv ɛ = ɛbv ɛ v ɛ, where Bu > for all u R and B C R. We only need to change Φ defined by 2.7 to 6.2 Φu; u +,u =[fu fu Ẋu u ]/Bu. License or copyright restrictions ay apply to redistribution; see

c 2000 Society for Industrial and Applied Mathematics

c 2000 Society for Industrial and Applied Mathematics SIAM J. NUME. ANAL. Vol. 38, No. 5, pp. 1483 1495 c 2000 Society for Industrial and Applied Matheatics ON THE EGULAITY OF APPOXIMATE SOLUTIONS TO CONSEVATION LAWS WITH PIECEWISE SMOOTH SOLUTIONS TAO TANG

More information

A Simple Regression Problem

A Simple Regression Problem A Siple Regression Proble R. M. Castro March 23, 2 In this brief note a siple regression proble will be introduced, illustrating clearly the bias-variance tradeoff. Let Y i f(x i ) + W i, i,..., n, where

More information

Polygonal Designs: Existence and Construction

Polygonal Designs: Existence and Construction Polygonal Designs: Existence and Construction John Hegean Departent of Matheatics, Stanford University, Stanford, CA 9405 Jeff Langford Departent of Matheatics, Drake University, Des Moines, IA 5011 G

More information

Solutions of some selected problems of Homework 4

Solutions of some selected problems of Homework 4 Solutions of soe selected probles of Hoework 4 Sangchul Lee May 7, 2018 Proble 1 Let there be light A professor has two light bulbs in his garage. When both are burned out, they are replaced, and the next

More information

4 = (0.02) 3 13, = 0.25 because = 25. Simi-

4 = (0.02) 3 13, = 0.25 because = 25. Simi- Theore. Let b and be integers greater than. If = (. a a 2 a i ) b,then for any t N, in base (b + t), the fraction has the digital representation = (. a a 2 a i ) b+t, where a i = a i + tk i with k i =

More information

1 Bounding the Margin

1 Bounding the Margin COS 511: Theoretical Machine Learning Lecturer: Rob Schapire Lecture #12 Scribe: Jian Min Si March 14, 2013 1 Bounding the Margin We are continuing the proof of a bound on the generalization error of AdaBoost

More information

A Note on Online Scheduling for Jobs with Arbitrary Release Times

A Note on Online Scheduling for Jobs with Arbitrary Release Times A Note on Online Scheduling for Jobs with Arbitrary Release Ties Jihuan Ding, and Guochuan Zhang College of Operations Research and Manageent Science, Qufu Noral University, Rizhao 7686, China dingjihuan@hotail.co

More information

3.8 Three Types of Convergence

3.8 Three Types of Convergence 3.8 Three Types of Convergence 3.8 Three Types of Convergence 93 Suppose that we are given a sequence functions {f k } k N on a set X and another function f on X. What does it ean for f k to converge to

More information

Soft Computing Techniques Help Assign Weights to Different Factors in Vulnerability Analysis

Soft Computing Techniques Help Assign Weights to Different Factors in Vulnerability Analysis Soft Coputing Techniques Help Assign Weights to Different Factors in Vulnerability Analysis Beverly Rivera 1,2, Irbis Gallegos 1, and Vladik Kreinovich 2 1 Regional Cyber and Energy Security Center RCES

More information

1. Introduction. We study the convergence of vanishing viscosity solutions governed by the single conservation law

1. Introduction. We study the convergence of vanishing viscosity solutions governed by the single conservation law SIAM J. NUMER. ANAL. Vol. 36, No. 6, pp. 1739 1758 c 1999 Society for Industrial and Applied Mathematics POINTWISE ERROR ESTIMATES FOR SCALAR CONSERVATION LAWS WITH PIECEWISE SMOOTH SOLUTIONS EITAN TADMOR

More information

A Note on Scheduling Tall/Small Multiprocessor Tasks with Unit Processing Time to Minimize Maximum Tardiness

A Note on Scheduling Tall/Small Multiprocessor Tasks with Unit Processing Time to Minimize Maximum Tardiness A Note on Scheduling Tall/Sall Multiprocessor Tasks with Unit Processing Tie to Miniize Maxiu Tardiness Philippe Baptiste and Baruch Schieber IBM T.J. Watson Research Center P.O. Box 218, Yorktown Heights,

More information

New upper bound for the B-spline basis condition number II. K. Scherer. Institut fur Angewandte Mathematik, Universitat Bonn, Bonn, Germany.

New upper bound for the B-spline basis condition number II. K. Scherer. Institut fur Angewandte Mathematik, Universitat Bonn, Bonn, Germany. New upper bound for the B-spline basis condition nuber II. A proof of de Boor's 2 -conjecture K. Scherer Institut fur Angewandte Matheati, Universitat Bonn, 535 Bonn, Gerany and A. Yu. Shadrin Coputing

More information

FAST DYNAMO ON THE REAL LINE

FAST DYNAMO ON THE REAL LINE FAST DYAMO O THE REAL LIE O. KOZLOVSKI & P. VYTOVA Abstract. In this paper we show that a piecewise expanding ap on the interval, extended to the real line by a non-expanding ap satisfying soe ild hypthesis

More information

Chapter 6 1-D Continuous Groups

Chapter 6 1-D Continuous Groups Chapter 6 1-D Continuous Groups Continuous groups consist of group eleents labelled by one or ore continuous variables, say a 1, a 2,, a r, where each variable has a well- defined range. This chapter explores:

More information

On the hyperbolicity of two- and three-layer shallow water equations. 1 Introduction

On the hyperbolicity of two- and three-layer shallow water equations. 1 Introduction On the hyperbolicity of two- and three-layer shallow water equations M. Castro, J.T. Frings, S. Noelle, C. Parés, G. Puppo RWTH Aachen, Universidad de Málaga, Politecnico di Torino E-ail: frings@igp.rwth-aachen.de

More information

Multi-Dimensional Hegselmann-Krause Dynamics

Multi-Dimensional Hegselmann-Krause Dynamics Multi-Diensional Hegselann-Krause Dynaics A. Nedić Industrial and Enterprise Systes Engineering Dept. University of Illinois Urbana, IL 680 angelia@illinois.edu B. Touri Coordinated Science Laboratory

More information

A BLOCK MONOTONE DOMAIN DECOMPOSITION ALGORITHM FOR A NONLINEAR SINGULARLY PERTURBED PARABOLIC PROBLEM

A BLOCK MONOTONE DOMAIN DECOMPOSITION ALGORITHM FOR A NONLINEAR SINGULARLY PERTURBED PARABOLIC PROBLEM INTERNATIONAL JOURNAL OF NUMERICAL ANALYSIS AND MODELING Volue 3, Nuber 2, Pages 211 231 c 2006 Institute for Scientific Coputing and Inforation A BLOCK MONOTONE DOMAIN DECOMPOSITION ALGORITHM FOR A NONLINEAR

More information

2 Q 10. Likewise, in case of multiple particles, the corresponding density in 2 must be averaged over all

2 Q 10. Likewise, in case of multiple particles, the corresponding density in 2 must be averaged over all Lecture 6 Introduction to kinetic theory of plasa waves Introduction to kinetic theory So far we have been odeling plasa dynaics using fluid equations. The assuption has been that the pressure can be either

More information

Model Fitting. CURM Background Material, Fall 2014 Dr. Doreen De Leon

Model Fitting. CURM Background Material, Fall 2014 Dr. Doreen De Leon Model Fitting CURM Background Material, Fall 014 Dr. Doreen De Leon 1 Introduction Given a set of data points, we often want to fit a selected odel or type to the data (e.g., we suspect an exponential

More information

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and This article appeared in a ournal published by Elsevier. The attached copy is furnished to the author for internal non-coercial research and education use, including for instruction at the authors institution

More information

List Scheduling and LPT Oliver Braun (09/05/2017)

List Scheduling and LPT Oliver Braun (09/05/2017) List Scheduling and LPT Oliver Braun (09/05/207) We investigate the classical scheduling proble P ax where a set of n independent jobs has to be processed on 2 parallel and identical processors (achines)

More information

13.2 Fully Polynomial Randomized Approximation Scheme for Permanent of Random 0-1 Matrices

13.2 Fully Polynomial Randomized Approximation Scheme for Permanent of Random 0-1 Matrices CS71 Randoness & Coputation Spring 018 Instructor: Alistair Sinclair Lecture 13: February 7 Disclaier: These notes have not been subjected to the usual scrutiny accorded to foral publications. They ay

More information

THE AVERAGE NORM OF POLYNOMIALS OF FIXED HEIGHT

THE AVERAGE NORM OF POLYNOMIALS OF FIXED HEIGHT THE AVERAGE NORM OF POLYNOMIALS OF FIXED HEIGHT PETER BORWEIN AND KWOK-KWONG STEPHEN CHOI Abstract. Let n be any integer and ( n ) X F n : a i z i : a i, ± i be the set of all polynoials of height and

More information

Estimation of the Mean of the Exponential Distribution Using Maximum Ranked Set Sampling with Unequal Samples

Estimation of the Mean of the Exponential Distribution Using Maximum Ranked Set Sampling with Unequal Samples Open Journal of Statistics, 4, 4, 64-649 Published Online Septeber 4 in SciRes http//wwwscirporg/ournal/os http//ddoiorg/436/os4486 Estiation of the Mean of the Eponential Distribution Using Maiu Ranked

More information

TABLE FOR UPPER PERCENTAGE POINTS OF THE LARGEST ROOT OF A DETERMINANTAL EQUATION WITH FIVE ROOTS. By William W. Chen

TABLE FOR UPPER PERCENTAGE POINTS OF THE LARGEST ROOT OF A DETERMINANTAL EQUATION WITH FIVE ROOTS. By William W. Chen TABLE FOR UPPER PERCENTAGE POINTS OF THE LARGEST ROOT OF A DETERMINANTAL EQUATION WITH FIVE ROOTS By Willia W. Chen The distribution of the non-null characteristic roots of a atri derived fro saple observations

More information

World's largest Science, Technology & Medicine Open Access book publisher

World's largest Science, Technology & Medicine Open Access book publisher PUBLISHED BY World's largest Science, Technology & Medicine Open Access book publisher 2750+ OPEN ACCESS BOOKS 95,000+ INTERNATIONAL AUTHORS AND EDITORS 88+ MILLION DOWNLOADS BOOKS DELIVERED TO 5 COUNTRIES

More information

Hyperbolic Systems of Conservation Laws. in One Space Dimension. II - Solutions to the Cauchy problem. Alberto Bressan

Hyperbolic Systems of Conservation Laws. in One Space Dimension. II - Solutions to the Cauchy problem. Alberto Bressan Hyperbolic Systems of Conservation Laws in One Space Dimension II - Solutions to the Cauchy problem Alberto Bressan Department of Mathematics, Penn State University http://www.math.psu.edu/bressan/ 1 Global

More information

ON THE OSCILLATION OF DIFFERENTIAL EQUATIONS WITH PERIODIC COEFFICIENTS

ON THE OSCILLATION OF DIFFERENTIAL EQUATIONS WITH PERIODIC COEFFICIENTS proceedings of the aerican atheatical society Volue 111, Nuber 2, February 1991 ON THE OSCILLATION OF DIFFERENTIAL EQUATIONS WITH PERIODIC COEFFICIENTS CH. G. PHILOS (Counicated by Kenneth R. Meyer) Abstract.

More information

Numerically repeated support splitting and merging phenomena in a porous media equation with strong absorption. Kenji Tomoeda

Numerically repeated support splitting and merging phenomena in a porous media equation with strong absorption. Kenji Tomoeda Journal of Math-for-Industry, Vol. 3 (C-), pp. Nuerically repeated support splitting and erging phenoena in a porous edia equation with strong absorption To the eory of y friend Professor Nakaki. Kenji

More information

ON SEQUENCES OF NUMBERS IN GENERALIZED ARITHMETIC AND GEOMETRIC PROGRESSIONS

ON SEQUENCES OF NUMBERS IN GENERALIZED ARITHMETIC AND GEOMETRIC PROGRESSIONS Palestine Journal of Matheatics Vol 4) 05), 70 76 Palestine Polytechnic University-PPU 05 ON SEQUENCES OF NUMBERS IN GENERALIZED ARITHMETIC AND GEOMETRIC PROGRESSIONS Julius Fergy T Rabago Counicated by

More information

Lecture 21. Interior Point Methods Setup and Algorithm

Lecture 21. Interior Point Methods Setup and Algorithm Lecture 21 Interior Point Methods In 1984, Kararkar introduced a new weakly polynoial tie algorith for solving LPs [Kar84a], [Kar84b]. His algorith was theoretically faster than the ellipsoid ethod and

More information

In this chapter, we consider several graph-theoretic and probabilistic models

In this chapter, we consider several graph-theoretic and probabilistic models THREE ONE GRAPH-THEORETIC AND STATISTICAL MODELS 3.1 INTRODUCTION In this chapter, we consider several graph-theoretic and probabilistic odels for a social network, which we do under different assuptions

More information

ANALYSIS OF A FULLY DISCRETE FINITE ELEMENT METHOD FOR THE PHASE FIELD MODEL AND APPROXIMATION OF ITS SHARP INTERFACE LIMITS

ANALYSIS OF A FULLY DISCRETE FINITE ELEMENT METHOD FOR THE PHASE FIELD MODEL AND APPROXIMATION OF ITS SHARP INTERFACE LIMITS ANALYSIS OF A FULLY DISCRETE FINITE ELEMENT METHOD FOR THE PHASE FIELD MODEL AND APPROXIMATION OF ITS SHARP INTERFACE LIMITS XIAOBING FENG AND ANDREAS PROHL Abstract. We propose and analyze a fully discrete

More information

Generalized eigenfunctions and a Borel Theorem on the Sierpinski Gasket.

Generalized eigenfunctions and a Borel Theorem on the Sierpinski Gasket. Generalized eigenfunctions and a Borel Theore on the Sierpinski Gasket. Kasso A. Okoudjou, Luke G. Rogers, and Robert S. Strichartz May 26, 2006 1 Introduction There is a well developed theory (see [5,

More information

e-companion ONLY AVAILABLE IN ELECTRONIC FORM

e-companion ONLY AVAILABLE IN ELECTRONIC FORM OPERATIONS RESEARCH doi 10.1287/opre.1070.0427ec pp. ec1 ec5 e-copanion ONLY AVAILABLE IN ELECTRONIC FORM infors 07 INFORMS Electronic Copanion A Learning Approach for Interactive Marketing to a Custoer

More information

Physically Based Modeling CS Notes Spring 1997 Particle Collision and Contact

Physically Based Modeling CS Notes Spring 1997 Particle Collision and Contact Physically Based Modeling CS 15-863 Notes Spring 1997 Particle Collision and Contact 1 Collisions with Springs Suppose we wanted to ipleent a particle siulator with a floor : a solid horizontal plane which

More information

Computational and Statistical Learning Theory

Computational and Statistical Learning Theory Coputational and Statistical Learning Theory Proble sets 5 and 6 Due: Noveber th Please send your solutions to learning-subissions@ttic.edu Notations/Definitions Recall the definition of saple based Radeacher

More information

CHAPTER 8 CONSTRAINED OPTIMIZATION 2: SEQUENTIAL QUADRATIC PROGRAMMING, INTERIOR POINT AND GENERALIZED REDUCED GRADIENT METHODS

CHAPTER 8 CONSTRAINED OPTIMIZATION 2: SEQUENTIAL QUADRATIC PROGRAMMING, INTERIOR POINT AND GENERALIZED REDUCED GRADIENT METHODS CHAPER 8 CONSRAINED OPIMIZAION : SEQUENIAL QUADRAIC PROGRAMMING, INERIOR POIN AND GENERALIZED REDUCED GRADIEN MEHODS 8. Introduction In the previous chapter we eained the necessary and sufficient conditions

More information

Distributed Subgradient Methods for Multi-agent Optimization

Distributed Subgradient Methods for Multi-agent Optimization 1 Distributed Subgradient Methods for Multi-agent Optiization Angelia Nedić and Asuan Ozdaglar October 29, 2007 Abstract We study a distributed coputation odel for optiizing a su of convex objective functions

More information

Pointwise convergence rate for nonlinear conservation. Eitan Tadmor and Tao Tang

Pointwise convergence rate for nonlinear conservation. Eitan Tadmor and Tao Tang Pointwise convergence rate for nonlinear conservation laws Eitan Tadmor and Tao Tang Abstract. We introduce a new method to obtain pointwise error estimates for vanishing viscosity and nite dierence approximations

More information

Finite fields. and we ve used it in various examples and homework problems. In these notes I will introduce more finite fields

Finite fields. and we ve used it in various examples and homework problems. In these notes I will introduce more finite fields Finite fields I talked in class about the field with two eleents F 2 = {, } and we ve used it in various eaples and hoework probles. In these notes I will introduce ore finite fields F p = {,,...,p } for

More information

A Better Algorithm For an Ancient Scheduling Problem. David R. Karger Steven J. Phillips Eric Torng. Department of Computer Science

A Better Algorithm For an Ancient Scheduling Problem. David R. Karger Steven J. Phillips Eric Torng. Department of Computer Science A Better Algorith For an Ancient Scheduling Proble David R. Karger Steven J. Phillips Eric Torng Departent of Coputer Science Stanford University Stanford, CA 9435-4 Abstract One of the oldest and siplest

More information

ROTH S THEOREM ON ARITHMETIC PROGRESSIONS

ROTH S THEOREM ON ARITHMETIC PROGRESSIONS ROTH S THEOREM ON ARITHMETIC PROGRESSIONS ADAM LOTT ABSTRACT. The goal of this paper is to present a self-contained eposition of Roth s celebrated theore on arithetic progressions. We also present two

More information

IN modern society that various systems have become more

IN modern society that various systems have become more Developent of Reliability Function in -Coponent Standby Redundant Syste with Priority Based on Maxiu Entropy Principle Ryosuke Hirata, Ikuo Arizono, Ryosuke Toohiro, Satoshi Oigawa, and Yasuhiko Takeoto

More information

Curious Bounds for Floor Function Sums

Curious Bounds for Floor Function Sums 1 47 6 11 Journal of Integer Sequences, Vol. 1 (018), Article 18.1.8 Curious Bounds for Floor Function Sus Thotsaporn Thanatipanonda and Elaine Wong 1 Science Division Mahidol University International

More information

arxiv: v1 [math.nt] 14 Sep 2014

arxiv: v1 [math.nt] 14 Sep 2014 ROTATION REMAINDERS P. JAMESON GRABER, WASHINGTON AND LEE UNIVERSITY 08 arxiv:1409.411v1 [ath.nt] 14 Sep 014 Abstract. We study properties of an array of nubers, called the triangle, in which each row

More information

Applications of Maximum Entropy Principle in Formulating Solution of Constrained Non-Linear Programming Problem

Applications of Maximum Entropy Principle in Formulating Solution of Constrained Non-Linear Programming Problem Journal of Statistical and Econoetric Methods, vol.4, no., 05, 45-60 ISSN: 4-084 (print), 4-076 (online) Scienpress Ltd, 05 Applications of Maiu Entropy Principle in Forulatin Solution of Constrained Non-Linear

More information

PRÜFER SUBSTITUTIONS ON A COUPLED SYSTEM INVOLVING THE p-laplacian

PRÜFER SUBSTITUTIONS ON A COUPLED SYSTEM INVOLVING THE p-laplacian Electronic Journal of Differential Equations, Vol. 23 (23), No. 23, pp. 9. ISSN: 72-669. URL: http://ejde.ath.txstate.edu or http://ejde.ath.unt.edu ftp ejde.ath.txstate.edu PRÜFER SUBSTITUTIONS ON A COUPLED

More information

1. INTRODUCTION AND RESULTS

1. INTRODUCTION AND RESULTS SOME IDENTITIES INVOLVING THE FIBONACCI NUMBERS AND LUCAS NUMBERS Wenpeng Zhang Research Center for Basic Science, Xi an Jiaotong University Xi an Shaanxi, People s Republic of China (Subitted August 00

More information

Support Vector Machine Classification of Uncertain and Imbalanced data using Robust Optimization

Support Vector Machine Classification of Uncertain and Imbalanced data using Robust Optimization Recent Researches in Coputer Science Support Vector Machine Classification of Uncertain and Ibalanced data using Robust Optiization RAGHAV PAT, THEODORE B. TRAFALIS, KASH BARKER School of Industrial Engineering

More information

A := A i : {A i } S. is an algebra. The same object is obtained when the union in required to be disjoint.

A := A i : {A i } S. is an algebra. The same object is obtained when the union in required to be disjoint. 59 6. ABSTRACT MEASURE THEORY Having developed the Lebesgue integral with respect to the general easures, we now have a general concept with few specific exaples to actually test it on. Indeed, so far

More information

Sharp Time Data Tradeoffs for Linear Inverse Problems

Sharp Time Data Tradeoffs for Linear Inverse Problems Sharp Tie Data Tradeoffs for Linear Inverse Probles Saet Oyak Benjain Recht Mahdi Soltanolkotabi January 016 Abstract In this paper we characterize sharp tie-data tradeoffs for optiization probles used

More information

On Certain C-Test Words for Free Groups

On Certain C-Test Words for Free Groups Journal of Algebra 247, 509 540 2002 doi:10.1006 jabr.2001.9001, available online at http: www.idealibrary.co on On Certain C-Test Words for Free Groups Donghi Lee Departent of Matheatics, Uni ersity of

More information

THE CONSTRUCTION OF GOOD EXTENSIBLE RANK-1 LATTICES. 1. Introduction We are interested in approximating a high dimensional integral [0,1]

THE CONSTRUCTION OF GOOD EXTENSIBLE RANK-1 LATTICES. 1. Introduction We are interested in approximating a high dimensional integral [0,1] MATHEMATICS OF COMPUTATION Volue 00, Nuber 0, Pages 000 000 S 0025-578(XX)0000-0 THE CONSTRUCTION OF GOOD EXTENSIBLE RANK- LATTICES JOSEF DICK, FRIEDRICH PILLICHSHAMMER, AND BENJAMIN J. WATERHOUSE Abstract.

More information

An Approximate Model for the Theoretical Prediction of the Velocity Increase in the Intermediate Ballistics Period

An Approximate Model for the Theoretical Prediction of the Velocity Increase in the Intermediate Ballistics Period An Approxiate Model for the Theoretical Prediction of the Velocity... 77 Central European Journal of Energetic Materials, 205, 2(), 77-88 ISSN 2353-843 An Approxiate Model for the Theoretical Prediction

More information

arxiv: v1 [cs.cc] 19 Jan 2016

arxiv: v1 [cs.cc] 19 Jan 2016 Extended Abstract SOLUTIONS TO THE PROBLEM OF K-SAT/K-COL PHASE TRANSITION LOCATION Changqing Liu ClinTFL Ltd. c.liu@clintfl.co arxiv:1601.04794v1 [cs.cc] 19 Jan 2016 As general theories, currently there

More information

On the Existence of Pure Nash Equilibria in Weighted Congestion Games

On the Existence of Pure Nash Equilibria in Weighted Congestion Games MATHEMATICS OF OPERATIONS RESEARCH Vol. 37, No. 3, August 2012, pp. 419 436 ISSN 0364-765X (print) ISSN 1526-5471 (online) http://dx.doi.org/10.1287/oor.1120.0543 2012 INFORMS On the Existence of Pure

More information

Stability Ordinates of Adams Predictor-Corrector Methods

Stability Ordinates of Adams Predictor-Corrector Methods BIT anuscript No. will be inserted by the editor Stability Ordinates of Adas Predictor-Corrector Methods Michelle L. Ghrist Jonah A. Reeger Bengt Fornberg Received: date / Accepted: date Abstract How far

More information

Block designs and statistics

Block designs and statistics Bloc designs and statistics Notes for Math 447 May 3, 2011 The ain paraeters of a bloc design are nuber of varieties v, bloc size, nuber of blocs b. A design is built on a set of v eleents. Each eleent

More information

On Process Complexity

On Process Complexity On Process Coplexity Ada R. Day School of Matheatics, Statistics and Coputer Science, Victoria University of Wellington, PO Box 600, Wellington 6140, New Zealand, Eail: ada.day@cs.vuw.ac.nz Abstract Process

More information

Global Existence and Nonexistence in a System of Petrovsky

Global Existence and Nonexistence in a System of Petrovsky Journal of Matheatical Analysis and Applications 265, 296 308 (2002) doi:10.1006/jaa.2001.7697, available online at http://www.idealibrary.co on Global Existence and Nonexistence in a Syste of Petrovsky

More information

The Weierstrass Approximation Theorem

The Weierstrass Approximation Theorem 36 The Weierstrass Approxiation Theore Recall that the fundaental idea underlying the construction of the real nubers is approxiation by the sipler rational nubers. Firstly, nubers are often deterined

More information

Probability Distributions

Probability Distributions Probability Distributions In Chapter, we ephasized the central role played by probability theory in the solution of pattern recognition probles. We turn now to an exploration of soe particular exaples

More information

Hermite s Rule Surpasses Simpson s: in Mathematics Curricula Simpson s Rule. Should be Replaced by Hermite s

Hermite s Rule Surpasses Simpson s: in Mathematics Curricula Simpson s Rule. Should be Replaced by Hermite s International Matheatical Foru, 4, 9, no. 34, 663-686 Herite s Rule Surpasses Sipson s: in Matheatics Curricula Sipson s Rule Should be Replaced by Herite s Vito Lapret University of Lublana Faculty of

More information

An Optimal Family of Exponentially Accurate One-Bit Sigma-Delta Quantization Schemes

An Optimal Family of Exponentially Accurate One-Bit Sigma-Delta Quantization Schemes An Optial Faily of Exponentially Accurate One-Bit Siga-Delta Quantization Schees Percy Deift C. Sinan Güntürk Felix Kraher January 2, 200 Abstract Siga-Delta odulation is a popular ethod for analog-to-digital

More information

Lost-Sales Problems with Stochastic Lead Times: Convexity Results for Base-Stock Policies

Lost-Sales Problems with Stochastic Lead Times: Convexity Results for Base-Stock Policies OPERATIONS RESEARCH Vol. 52, No. 5, Septeber October 2004, pp. 795 803 issn 0030-364X eissn 1526-5463 04 5205 0795 infors doi 10.1287/opre.1040.0130 2004 INFORMS TECHNICAL NOTE Lost-Sales Probles with

More information

CHARACTER SUMS AND RAMSEY PROPERTIES OF GENERALIZED PALEY GRAPHS. Nicholas Wage Appleton East High School, Appleton, WI 54915, USA.

CHARACTER SUMS AND RAMSEY PROPERTIES OF GENERALIZED PALEY GRAPHS. Nicholas Wage Appleton East High School, Appleton, WI 54915, USA. INTEGERS: ELECTRONIC JOURNAL OF COMBINATORIAL NUMBER THEORY 6 006, #A18 CHARACTER SUMS AND RAMSEY PROPERTIES OF GENERALIZED PALEY GRAPHS Nicholas Wage Appleton East High School, Appleton, WI 54915, USA

More information

M ath. Res. Lett. 15 (2008), no. 2, c International Press 2008 SUM-PRODUCT ESTIMATES VIA DIRECTED EXPANDERS. Van H. Vu. 1.

M ath. Res. Lett. 15 (2008), no. 2, c International Press 2008 SUM-PRODUCT ESTIMATES VIA DIRECTED EXPANDERS. Van H. Vu. 1. M ath. Res. Lett. 15 (2008), no. 2, 375 388 c International Press 2008 SUM-PRODUCT ESTIMATES VIA DIRECTED EXPANDERS Van H. Vu Abstract. Let F q be a finite field of order q and P be a polynoial in F q[x

More information

ON THE TWO-LEVEL PRECONDITIONING IN LEAST SQUARES METHOD

ON THE TWO-LEVEL PRECONDITIONING IN LEAST SQUARES METHOD PROCEEDINGS OF THE YEREVAN STATE UNIVERSITY Physical and Matheatical Sciences 04,, p. 7 5 ON THE TWO-LEVEL PRECONDITIONING IN LEAST SQUARES METHOD M a t h e a t i c s Yu. A. HAKOPIAN, R. Z. HOVHANNISYAN

More information

DEPARTMENT OF ECONOMETRICS AND BUSINESS STATISTICS

DEPARTMENT OF ECONOMETRICS AND BUSINESS STATISTICS ISSN 1440-771X AUSTRALIA DEPARTMENT OF ECONOMETRICS AND BUSINESS STATISTICS An Iproved Method for Bandwidth Selection When Estiating ROC Curves Peter G Hall and Rob J Hyndan Working Paper 11/00 An iproved

More information

A proposal for a First-Citation-Speed-Index Link Peer-reviewed author version

A proposal for a First-Citation-Speed-Index Link Peer-reviewed author version A proposal for a First-Citation-Speed-Index Link Peer-reviewed author version Made available by Hasselt University Library in Docuent Server@UHasselt Reference (Published version): EGGHE, Leo; Bornann,

More information

Structured signal recovery from quadratic measurements: Breaking sample complexity barriers via nonconvex optimization

Structured signal recovery from quadratic measurements: Breaking sample complexity barriers via nonconvex optimization Structured signal recovery fro quadratic easureents: Breaking saple coplexity barriers via nonconvex optiization Mahdi Soltanolkotabi Ming Hsieh Departent of Electrical Engineering University of Southern

More information

Derivative at a point

Derivative at a point Roberto s Notes on Differential Calculus Capter : Definition of derivative Section Derivative at a point Wat you need to know already: Te concept of liit and basic etods for coputing liits. Wat you can

More information

Constrained Consensus and Optimization in Multi-Agent Networks arxiv: v2 [math.oc] 17 Dec 2008

Constrained Consensus and Optimization in Multi-Agent Networks arxiv: v2 [math.oc] 17 Dec 2008 LIDS Report 2779 1 Constrained Consensus and Optiization in Multi-Agent Networks arxiv:0802.3922v2 [ath.oc] 17 Dec 2008 Angelia Nedić, Asuan Ozdaglar, and Pablo A. Parrilo February 15, 2013 Abstract We

More information

On Poset Merging. 1 Introduction. Peter Chen Guoli Ding Steve Seiden. Keywords: Merging, Partial Order, Lower Bounds. AMS Classification: 68W40

On Poset Merging. 1 Introduction. Peter Chen Guoli Ding Steve Seiden. Keywords: Merging, Partial Order, Lower Bounds. AMS Classification: 68W40 On Poset Merging Peter Chen Guoli Ding Steve Seiden Abstract We consider the follow poset erging proble: Let X and Y be two subsets of a partially ordered set S. Given coplete inforation about the ordering

More information

Homotopy Analysis Method for Solving Fuzzy Integro-Differential Equations

Homotopy Analysis Method for Solving Fuzzy Integro-Differential Equations Modern Applied Science; Vol. 7 No. 3; 23 ISSN 93-844 E-ISSN 93-82 Published by Canadian Center of Science Education Hootopy Analysis Method for Solving Fuzzy Integro-Differential Equations Ean A. Hussain

More information

The Hilbert Schmidt version of the commutator theorem for zero trace matrices

The Hilbert Schmidt version of the commutator theorem for zero trace matrices The Hilbert Schidt version of the coutator theore for zero trace atrices Oer Angel Gideon Schechtan March 205 Abstract Let A be a coplex atrix with zero trace. Then there are atrices B and C such that

More information

Computable Shell Decomposition Bounds

Computable Shell Decomposition Bounds Coputable Shell Decoposition Bounds John Langford TTI-Chicago jcl@cs.cu.edu David McAllester TTI-Chicago dac@autoreason.co Editor: Leslie Pack Kaelbling and David Cohn Abstract Haussler, Kearns, Seung

More information

12 Towards hydrodynamic equations J Nonlinear Dynamics II: Continuum Systems Lecture 12 Spring 2015

12 Towards hydrodynamic equations J Nonlinear Dynamics II: Continuum Systems Lecture 12 Spring 2015 18.354J Nonlinear Dynaics II: Continuu Systes Lecture 12 Spring 2015 12 Towards hydrodynaic equations The previous classes focussed on the continuu description of static (tie-independent) elastic systes.

More information

arxiv: v2 [math.nt] 5 Sep 2012

arxiv: v2 [math.nt] 5 Sep 2012 ON STRONGER CONJECTURES THAT IMPLY THE ERDŐS-MOSER CONJECTURE BERND C. KELLNER arxiv:1003.1646v2 [ath.nt] 5 Sep 2012 Abstract. The Erdős-Moser conjecture states that the Diophantine equation S k () = k,

More information

TRAVELING WAVE SOLUTIONS OF THE POROUS MEDIUM EQUATION. Laxmi P. Paudel. Dissertation Prepared for the Degree of DOCTOR OF PHILOSOPHY

TRAVELING WAVE SOLUTIONS OF THE POROUS MEDIUM EQUATION. Laxmi P. Paudel. Dissertation Prepared for the Degree of DOCTOR OF PHILOSOPHY TRAVELING WAVE SOLUTIONS OF THE POROUS MEDIUM EQUATION Laxi P Paudel Dissertation Prepared for the Degree of DOCTOR OF PHILOSOPHY UNIVERSITY OF NORTH TEXAS May 013 APPROVED: Joseph Iaia, Major Professor

More information

Fairness via priority scheduling

Fairness via priority scheduling Fairness via priority scheduling Veeraruna Kavitha, N Heachandra and Debayan Das IEOR, IIT Bobay, Mubai, 400076, India vavitha,nh,debayan}@iitbacin Abstract In the context of ulti-agent resource allocation

More information

Algorithms for parallel processor scheduling with distinct due windows and unit-time jobs

Algorithms for parallel processor scheduling with distinct due windows and unit-time jobs BULLETIN OF THE POLISH ACADEMY OF SCIENCES TECHNICAL SCIENCES Vol. 57, No. 3, 2009 Algoriths for parallel processor scheduling with distinct due windows and unit-tie obs A. JANIAK 1, W.A. JANIAK 2, and

More information

Acyclic Colorings of Directed Graphs

Acyclic Colorings of Directed Graphs Acyclic Colorings of Directed Graphs Noah Golowich Septeber 9, 014 arxiv:1409.7535v1 [ath.co] 6 Sep 014 Abstract The acyclic chroatic nuber of a directed graph D, denoted χ A (D), is the iniu positive

More information

APPROXIMATION BY GENERALIZED FABER SERIES IN BERGMAN SPACES ON INFINITE DOMAINS WITH A QUASICONFORMAL BOUNDARY

APPROXIMATION BY GENERALIZED FABER SERIES IN BERGMAN SPACES ON INFINITE DOMAINS WITH A QUASICONFORMAL BOUNDARY NEW ZEALAND JOURNAL OF MATHEMATICS Volue 36 007, 11 APPROXIMATION BY GENERALIZED FABER SERIES IN BERGMAN SPACES ON INFINITE DOMAINS WITH A QUASICONFORMAL BOUNDARY Daniyal M. Israfilov and Yunus E. Yildirir

More information

DRAFT. Memo. Contents. To whom it may concern SVN: Jan Mooiman +31 (0) nl

DRAFT. Memo. Contents. To whom it may concern SVN: Jan Mooiman +31 (0) nl Meo To To who it ay concern Date Reference Nuber of pages 219-1-16 SVN: 5744 22 Fro Direct line E-ail Jan Mooian +31 )88 335 8568 jan.ooian@deltares nl +31 6 4691 4571 Subject PID controller ass-spring-daper

More information

Best Arm Identification: A Unified Approach to Fixed Budget and Fixed Confidence

Best Arm Identification: A Unified Approach to Fixed Budget and Fixed Confidence Best Ar Identification: A Unified Approach to Fixed Budget and Fixed Confidence Victor Gabillon Mohaad Ghavazadeh Alessandro Lazaric INRIA Lille - Nord Europe, Tea SequeL {victor.gabillon,ohaad.ghavazadeh,alessandro.lazaric}@inria.fr

More information

Note on generating all subsets of a finite set with disjoint unions

Note on generating all subsets of a finite set with disjoint unions Note on generating all subsets of a finite set with disjoint unions David Ellis e-ail: dce27@ca.ac.uk Subitted: Dec 2, 2008; Accepted: May 12, 2009; Published: May 20, 2009 Matheatics Subject Classification:

More information

Modeling the Structural Shifts in Real Exchange Rate with Cubic Spline Regression (CSR). Turkey

Modeling the Structural Shifts in Real Exchange Rate with Cubic Spline Regression (CSR). Turkey International Journal of Business and Social Science Vol. 2 No. 17 www.ijbssnet.co Modeling the Structural Shifts in Real Exchange Rate with Cubic Spline Regression (CSR). Turkey 1987-2008 Dr. Bahar BERBEROĞLU

More information

Comparison of Stability of Selected Numerical Methods for Solving Stiff Semi- Linear Differential Equations

Comparison of Stability of Selected Numerical Methods for Solving Stiff Semi- Linear Differential Equations International Journal of Applied Science and Technology Vol. 7, No. 3, Septeber 217 Coparison of Stability of Selected Nuerical Methods for Solving Stiff Sei- Linear Differential Equations Kwaku Darkwah

More information

AN APPLICATION OF CUBIC B-SPLINE FINITE ELEMENT METHOD FOR THE BURGERS EQUATION

AN APPLICATION OF CUBIC B-SPLINE FINITE ELEMENT METHOD FOR THE BURGERS EQUATION Aksan, E..: An Applıcatıon of Cubıc B-Splıne Fınıte Eleent Method for... THERMAL SCIECE: Year 8, Vol., Suppl., pp. S95-S S95 A APPLICATIO OF CBIC B-SPLIE FIITE ELEMET METHOD FOR THE BRGERS EQATIO by Eine

More information

RANDOM GRADIENT EXTRAPOLATION FOR DISTRIBUTED AND STOCHASTIC OPTIMIZATION

RANDOM GRADIENT EXTRAPOLATION FOR DISTRIBUTED AND STOCHASTIC OPTIMIZATION RANDOM GRADIENT EXTRAPOLATION FOR DISTRIBUTED AND STOCHASTIC OPTIMIZATION GUANGHUI LAN AND YI ZHOU Abstract. In this paper, we consider a class of finite-su convex optiization probles defined over a distributed

More information

Course Notes for EE227C (Spring 2018): Convex Optimization and Approximation

Course Notes for EE227C (Spring 2018): Convex Optimization and Approximation Course Notes for EE7C (Spring 018: Convex Optiization and Approxiation Instructor: Moritz Hardt Eail: hardt+ee7c@berkeley.edu Graduate Instructor: Max Sichowitz Eail: sichow+ee7c@berkeley.edu October 15,

More information

Algebraic Montgomery-Yang problem: the log del Pezzo surface case

Algebraic Montgomery-Yang problem: the log del Pezzo surface case c 2014 The Matheatical Society of Japan J. Math. Soc. Japan Vol. 66, No. 4 (2014) pp. 1073 1089 doi: 10.2969/jsj/06641073 Algebraic Montgoery-Yang proble: the log del Pezzo surface case By DongSeon Hwang

More information

Concentration of ground states in stationary Mean-Field Games systems: part I

Concentration of ground states in stationary Mean-Field Games systems: part I Concentration of ground states in stationary Mean-Field Gaes systes: part I Annalisa Cesaroni and Marco Cirant June 9, 207 Abstract In this paper we provide the existence of classical solutions to soe

More information

8.1 Force Laws Hooke s Law

8.1 Force Laws Hooke s Law 8.1 Force Laws There are forces that don't change appreciably fro one instant to another, which we refer to as constant in tie, and forces that don't change appreciably fro one point to another, which

More information

The Fundamental Basis Theorem of Geometry from an algebraic point of view

The Fundamental Basis Theorem of Geometry from an algebraic point of view Journal of Physics: Conference Series PAPER OPEN ACCESS The Fundaental Basis Theore of Geoetry fro an algebraic point of view To cite this article: U Bekbaev 2017 J Phys: Conf Ser 819 012013 View the article

More information

Principles of Optimal Control Spring 2008

Principles of Optimal Control Spring 2008 MIT OpenCourseWare http://ocw.it.edu 16.323 Principles of Optial Control Spring 2008 For inforation about citing these aterials or our Ters of Use, visit: http://ocw.it.edu/ters. 16.323 Lecture 10 Singular

More information

Biostatistics Department Technical Report

Biostatistics Department Technical Report Biostatistics Departent Technical Report BST006-00 Estiation of Prevalence by Pool Screening With Equal Sized Pools and a egative Binoial Sapling Model Charles R. Katholi, Ph.D. Eeritus Professor Departent

More information

arxiv:math/ v1 [math.nt] 6 Apr 2005

arxiv:math/ v1 [math.nt] 6 Apr 2005 SOME PROPERTIES OF THE PSEUDO-SMARANDACHE FUNCTION arxiv:ath/05048v [ath.nt] 6 Apr 005 RICHARD PINCH Abstract. Charles Ashbacher [] has posed a nuber of questions relating to the pseudo-sarandache function

More information