Filename
stringlengths 22
64
| Paragraph
stringlengths 8
5.57k
|
|---|---|
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | a blue loop in the HR diagram (HRD) and according to Fujimoto et al. (1990) it can be explained as follows : At the ignition of the CN cycle, the core expands and both density and temperature decrease in the central regions. The surface layers thus contract and the star moves to the blue side of the HRD. As the flash decays, core contraction resumes, gravitational energy is released and a new expansion takes place ; the star moves back to the red side of the HRD, producing the loop. When the red giant branch is reached, the convective envelope deepens and the first dredge up occurs. Note that only stars with M . 1.0 M⊙ experience the first dredge-up. Its effect on the surface composition is marginal and amounts to a slight enrichment in 3He and 4He and to the destruction of the remaining 7Li. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | The evolutionary tracks of Z = 0 stars in the HR diagram is presented in Fig. 4.1. The appearance of blue loops in the tracks is present at the end of the main sequence of our 1 M⊙ model and also appears in the 3, 4 and marginally in the 5 M⊙ models at the ignition of the CNO cycle in the HBS at the beginning of the AGB phase. Note also that due to the lower opacity of the envelope, zero metallicity stars have a substantially higher effective temperature than their counterparts of higher metallicity. Their UV flux is substantially larger and massive primordial stars are now considered to be the leading candidates for the sources of reionization of the intergalactic medium (e.g. Gnedin 2000). |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | The evolution of a Z = 0 solar mass star has recently been modeled by Weiss et al. (2000; WCSS hereafter) who compared their work with the earlier simulation of Fujimoto et al. (1990; FIH hereafter). We will not reproduce that comparison here but rather extend it, by comparing our model with the most recent simulation of WCSS. During the central H burning phase, the main difference is the development of a convective core in our 1 M⊙ model, a feature absent from both the WCSS and the FIH simulations. The presence of this convective region may simply arise from the adoption of different input physics and leads to slightly lower central temperatures and densities. It also affects the age at the turn off which is slightly larger in our models (6.56 Gyr in comparison with 6.31 Gyr found by WCSS). At the ignition of the CNO cycle, we find a similar hydrogen concentration at the center (Xc = 5.8 × 10−4), and the maximum size of the CNO-driven convective core is also very comparable (∼ 0.11 M⊙ in WCSS, in comparison with ∼ 0.095 M⊙ in this study). The smaller core size characterizing our models is probably a consequence of our lower central temperatures. For comparison, at the maximum nuclear energy production during the flash, the central abundances of 12C:14N:16O are 8.30 × 10−12 : 2.48 × 10−10 : 2.78 × 10−12 in our models and 6.50 × 10−12 : 2.19 × 10−10 : 2.77 × 10−12 in WCSS. Finally, during the “CNO-flash” the luminosity generated by the CN cycle (LCNO) remains always smaller than the one generated by the p-p reactions. At its peak, LCNO represents about 16% of the total nuclear energy production. The contribution due to He burning is negligible (LHe ≃ 10−7 L⊙) and is slightly less than in WCSS (LHe = 2.6 × 10−7 L⊙) but still 2 orders of magnitude smaller than in Fujimoto et al. (1990). |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | the time of the He flash, the general properties of our model are very similar to those of WCSS, the luminosity and hydrogen-exhausted core mass being 227.75 compared to 229.4 L⊙ and 0.497 compared to 0.492 M⊙, respectively. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | The ignition of core He burning starts off-center at a mass coordinate Mr = 0.31 M⊙ which is slightly larger than the 0.25 and 0.27 M⊙ found by D’Antona (1982) and Cassisi and Castellani (1993) respectively, but significantly smaller than the 0.41 M⊙ found by FIH. The location where the first He-shell flash starts has important consequences for the subsequent evolution. Indeed, if the He-flash ignites close enough to the H-discontinuity, as has been found by FIH and later by Fujimoto et al. (2000) and Schlattl et al. (2001), the He-driven convective zone might be able to penetrate the overlaying H rich layers. In this situation, protons are engulfed in the He-driven convective zone, mixed with the abundant carbon and a H-flash occurs which subsequently leads to large modifications in the surface chemical composition and in particular to a large C and N enrichment. The location where the He flash ignites is thus a crucial issue but it depends on the star’s internal structure, which in turn reflects the imprint of the adopted physics. Recently, Schlattl et al. (2001) have analyzed in detail the conditions for proton ingestion during the He-core flash. Their analysis reveals that the occurrence of this phenomenon depends quite significantly on the initial conditions. More specifically, they showed that the use of a different EOS (which determines the electron degeneracy), of different radiative and/or conductive opacity tables, or the treatment or not of diffusion, can trigger or inhibit the penetration of the He-driven flash into the H rich layers. They also showed that the treatment of convection has very little effect on the outcome. As emphasized in FIH, the use of different treatments of plasma neutrinos turn out to have significant effects on the results. Finally, we recall that the location of the He flash also depends on the rate at which the H burning shell advances in mass. Since there are several differences in physical input between FIH and the present work (which include core overshooting), it is not easy to give a unique explanation for this different location, but we suspect that the opacities may have played a dominant role. To conclude, we didn’t find proton mixing because our He ignites too far from the H/He discontinuity and to illustrate the entangled dependence of this phenomenon on the initial conditions, let us note that Marigo et al. (2001) did not mention any evidence of extra protons mixing during the He core flash of their low-mass stars. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | The flash is initiated at the point of maximum temperature. Where the degeneracy is lifted, the temperature increases very rapidly and the nuclear energy production due to 3α reactions undergoes a runaway. A convective instability develops and the stellar structure expands. The pressure and the temperature in the degenerate core decrease as the weight of the overlaying layers decreases and when the flash dies away, core contraction resumes. The temperature peak then moves inward and advances in mass by ∼ 0.03 M⊙ between each secondary flash, a value very similar to the one found by Mengel and Sweigart (1981) in their 0.70 M⊙ stellar model. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | of the subsequent flashes is smaller due to the lower He-luminosity and the time interval between the secondary flashes also gradually increases (see also Despain 1981). Table 1 summarizes the characteristics of these flashes. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | During the major flash, only a very small amount of protons is ingested into the convective region and this prevents the H-burning luminosity from reaching the very high values encountered in the computation of Hollowell et al. (2001). Our computations shows that the H mass fraction at the outermost point of the convective region is always smaller than X < 10−8 and the hydrogen-generated luminosity never exceeds ∼ 104 L⊙ at its peak. The maximum extent of the convective flash reaches to within 0.005 M⊙ of the H rich region (where X > 0.1) which is located at Mr = 0.5075 M⊙. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | After 10 flashes, He burning starts at the center and the star rapidly moves to the blue side of the HRD along the horizontal branch. The envelope recedes and a convective core of ∼ 0.20 M⊙ forms and is sustained for ≈ 70.6 Myr until He is depleted at the center. The energy production comes largely from the 3α reactions (90%) taking place in the core but also in part from gravitational contraction (10%). The HBS, while still active, does not contribute significantly to the net energy budget. During the central He-burning phase, the convective core is subject to “breathing” pulses and convective tongues develop at the surface of the core adding additional He into the burning region and temporarily increasing the He-burning luminosity. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | We present in Tab. 2 selected parameters for our computed models during the central H and He burning phases, namely, the duration of H (He) core burning phase, the central temperature and density when the central abundance of H (He) equal 0.5, the abundance of the 3 most abundant species at the end of each core burning phase, the core mass at the tip of the RGB and the deepest extent of the convective envelope following the He core burning phase. Interestingly, we see that the central density during the H core burning and the core mass at the tip of the RGB show a minimum around M ≃ 1.5 M⊙. This value is very similar to the one found by Cassisi and Castellani (1993) and corresponds to the critical mass for the ignition of He in strongly degenerate conditions. This smaller value for the mass limit, compared for example with the 1.8 M⊙ found for Z = 10−6 (Cassisi & Castellani 1993), results from the fact that more metal deficient stars must increase their temperature (and density) to sustain a larger luminosity. They are consequently hotter and are thus less affected by electron degeneracy. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | The comparison of the nuclear lifetime with the recent paper by Marigo et al. (2001) reveals discrepancies as large as 15% in the duration of the main sequence. The largest shift is found for the critical mass M = 1.5 M⊙, and for this reason we suspect that part of these differences results from the adopted equation of state. Further, these authors use a different value for the overshooting parameter and this can also explain the discrepancies. Globally, our lower mass models (M ≤ 3 M⊙) evolve more slowly while the more massive ones tend to evolve slightly faster. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | the He burning shells increases and the H burning shell is switched-off. This situation allows the deepening of the convective envelope which reaches a mass coordinate of 0.599 M⊙. The second dredge-up is a general feature of stars in the mass range 1.0 ≤ M < 10 M⊙ and its efficiency in polluting the envelope increases with mass. Following this deep mixing, the surface layers are mainly enriched in 4He and to a lesser extent with CNO elements. Table 3 summarizes some of the properties of the envelope composition at the end of this phase. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | When He is depleted at the center, the star undergoes a large expansion and moves rapidly to the red side of the HRD. The HBS quenches and the convective envelope deepens. At its largest extent, the envelope reaches a mass coordinate of Mr ≃ 0.896 M⊙ and dredges up to the surface the products of the nucleosynthesis taking place in the H burning. As a result of the CNO cycle, the surface is mainly enriched in 14N and 4He and because the 3α reactions are also at work in the HBS, the abundance of 12C and 16O are increased while H and 3He are depleted. After the reignition of the HBS, the star undergoes a series of “weak” thermal runaways in the HeBS (LHe . 105 L⊙) which are not powerful enough to trigger the formation of a convective region, nor to extinguish the HBS (Fig. 4.1). However, at the 13th instability, a convective region forms in the HeBS, and by the mechanism described in the next section, a pollution of the envelope takes place, allowing the star to resume a “standard” AGB evolution. Interestingly, Chieffi and Tornamb´e (1984) also found that 4 small instabilities developed in their 5 M⊙ model at the beginning of the AGB phase. However, these authors did not find subsequent mixing and a steady state double shell burning evolution followed, characterized by the absence of thermal pulses. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | After the H reignition, the luminosity is mostly provided by the HBS (75%) and to a lesser extent by the HeBS (15%) and gravitational contraction (10%). When the temperature in the HBS reaches 108K, the 3α reactions start to operate more efficiently and 12C production increases. However, due to the relatively low density in the HBS (ρHBS . 350g cm−3), the 3α reactions do not contribute significantly to the nuclear energy generation, which is mainly the result of efficient CNO burning. The high temperatures found in the HBS also activate additional CN cycling. In particular, the chain of reactions 16O(p,γ)17F(β+)17O(p,γ)18F(β+)18O(p,γ)19F is responsible for some marginal 19F production. However, the main surface enrichment in this element is the result of the subsequent deep third dredge-up (3DUP) episodes. By the end of our computations, its surface mass fraction amounts to X(19F) ≃ 6.4 × 10−8, but as for 7Li (see Sect. 4), this value depends on the competitive effects resulting from its destruction by hot bottom burning and enrichment by the third dredge-up events. Note also that at the base of the HBS, the “hot” CNO cycle is operating, characterized by the main branching 13N(p,γ)14O(β+)14N. The NeNa and MgAl cycles are also activated and contribute locally to the production of 23Na and 26Al, but the abundances of these elements still remain very small (< 10−15 in mass fraction). |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | Thermal pulses during the AGB evolution are present in all of our models from 1.0 to 7 M⊙. The occurrence of thermal pulses in zero-metal stars was initially investigated theoretically by Fujimoto et al. (1984). These authors first pointed out the existence of threshold values for the core-mass and CNO abundances for the occurrence of thermal instabilities. In particular, they showed that if the core mass is smaller then a critical value M ∗ ≈ 0.73 M⊙, recurrent He shell flashes are expected to occur. They also found that for Mcore & M ∗, there exists a critical envelope CNO abundance (∼ 10−7) above which He shell flashes will be triggered. The reason for this behavior was explained in Fujimoto et al. (1984) and Chieffi and Tornamb´e (1984) and can be summarized as follows : in massive stars, for which Mcore & M ∗, the lack of CNO nuclei imposes that H is burnt at very high temperatures, allowing for carbon production in the HBS. Thus, the 3α reactions are working contemporaneously in both shells and they advance in mass at a similar rate. This prevents the mass growth of the intershell and thereby the occurrence of the instability. Conversely, in low-mass stars, the energy requirements are lower and the 3α reactions are not very active in the HBS. The two shells thus proceed outward at different rates - an unstable situation leading to the development of thermal pulses. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | Our simulations indicate that contrary to the theoretical expectations of Fujimoto et al. (1984), stars with Mcore > 0.73 M⊙ do enter the thermally pulsing AGB (TP-AGB) phase. The reason for this behavior can be found in the fact that, low- and intermediate-mass stars initially experience a series of instabilities (carbon injections) that progressively enrich the HBS and the envelope in heavy elements. As a consequence of this chemical “pollution”, the CNO cycle operates in the HBS without the requirement of carbon production by 3α reactions, as in more metal rich stars. The sequence of events leading to this carbon injection can be illustrated through a careful examination of our 3 M⊙ model. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | model, the shell has a very low C abundance (∼ 10−8) and the sudden ingestion of a large amount of carbon inevitably boost the nuclear energy production. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | The energy released during the successive (He+H) flashes produces an expansion of the over- laying layers which is followed by a deeper penetration of the convective envelope. In its descent, the envelope penetrates the region that was previously occupied by the HCS and enriches the sur- face with the elements present in the HCS. During this event, the envelope abundances of 12C, 14N and 16O increase significantly (see Tab. 3). |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | This irreversible process, which we call “carbon injection”, is characteristic of all our interme- diate mass models, even in models with Mcore . M ∗ and is activated only during the first thermal pulse of our 1, 1.5, 2, 4 and 5 M⊙ models and during the first 4 instabilities in our 3 M⊙ model (Fig. 4.1). In this latter case, the first carbon injections are not able to raise the CNO envelope abundance above the threshold value of ∼ 10−7 and the process repeats until this value is exceeded. Table 3 summarizes some properties at the end of these carbon injections. We also note that in our 7 M⊙ model, the carbon enrichment of the envelope following the second dredge-up is large enough to ensure that the CNO cycle is operating efficiently inside the HBS without the need of self carbon production by 3α reactions. Therefore, this model reaches a “standard” AGB phase and the C-injection are absent. CDLS found that their 6 M⊙ model also ended the early AGB phase with a large CNO abundance and thus, during the following evolution, it behaved normally, like our 7 M⊙ model. As far as our computations are concerned, carbon does not ignite in this model, a conclusion also reached by CDLS. This result thus indicates that the minimum mass for degenerate carbon ignition is Mup ≥ 7 M⊙. We present in Tab. 4 and 5 some properties for the 7 and 3 M⊙ models. Columns (1)-(13) list the following quantities (1) the location of the H-burning shell at the pulse ignition, (2) the location of the convective envelope relative to the HBS at the pulse ignition (Menv = MH + ∆M H env), (3) the pulse duration (∆tpul), (4) the interpulse duration (∆tinter in 103 yr), the temperature (5) and density (6) at the base of the pulse at the time of maximum nuclear energy generation, (7) the peak luminosity of the pulse, (8) the location of the top of the convective pulse relative to the HBS (M top pul), (9) the pulse mass (Mpul = M top pul − M bot pul ), (10) the location of the deepest penetration of the convective envelope env = MH + ∆MDUP) and the surface mass fractions of 12C, 14N relative to HBS after the pulse (M min and 16O at the time of pulse ignition. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | cAt the end of our computations (15 Gyr), this star had not yet ignited its core He supply. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | We also find that after the disappearance of the HCS and before the envelope penetrates the region polluted by the HCS, a secondary convective zone can develop at the border of the H rich region. This feature, illustrated in Fig. 4.1 in the 2 M⊙ model, depends on the mesh resolution and numerical scheme (see below). Although this secondary convective shell never mixes with the underlying 12C and 4He rich layers, this situation is unstable with respect to the Schwarzschild criterion because of the large positive value of ∇rad−∇ad at the bottom of the convective region (e.g. Castellani et al. 1971a, 1971b). When ∇rad > ∇ad, any perturbation of the convective boundary would give rise to extra mixing, and in particular carbon could be injected into the convective zone with the possible triggering of a new H flash and further chemical contamination of the envelope. The formation of a “secondary convective shell” above the H discontinuity has also been found by Dominguez et al. (2000), after 9 weak instabilities during the evolution of a 5 M⊙ primordial star, and it strikingly resembles the one depicted here except that in their simulation, it takes place before the C-injection (see their Fig. 7). Finally note that in our simulations this feature, if real, has no influence on the subsequent evolution since it does not mix with the underlying C-rich layers. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | The existence of C-injection episodes at the beginning of the AGB phase has recently been found by CDLS in their computations of zero-metal intermediate-mass stars. In these computations, this process follows the same course of events : a convective shell develops at the base of the H rich layer, ingests carbon and initiates an H flash. The additional energy released by this H flash subsequently leads to a deeper penetration of the convective envelope which reaches the CNO enriched HCS and pollutes the envelope. Our results thus confirm these independent computations and the unusual circumstances by which primordial stars enter the TP-AGB phase. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | the dredge- up mass can lead to significant changes after the mixing events. Basically, it is much easier to raise the metallicity from almost zero to ∼ 10−6 than from 10−6 to 10−5. The important point for the subsequent evolution is that a substantial amount of carbon has been deposited into the envelope. Remembering that an envelope mass fraction of 12C as low as X12C ≃ 10−7 is sufficient to switch H burning to the CNO cycle, we realize that an uncertainty in the amount of C dredged-up in the envelope by as much as an order of magnitude will not result in a change in the subsequent evolution, because the C-injection has already raised the 12C abundance well above this limit. The influence on the yields will also be negligible since the main chemical enrichment of the surface comes from the third dredge-up event (3DUP). As a conclusion, even though the modeling of this mixing phase suffers from some numerical uncertainties, its occurrence is well established from our simulations and those of CDLS and we are confident that the exact modeling of this process do not influence the subsequent evolution of the thermally pulsing AGB stars substantially. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | Our 4 and 5 M⊙ models exhibit a different kind of thermal pulses which we refer to as “degener- ate pulses” following the designation given by Frost et al. (1998). We illustrate their characteristics by returning to the evolution of our 5 M⊙ model. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | After a series of 12 weak instabilities, a convective pulse develops and the 5 M⊙ model under- goes the carbon injection phenomenon described previously. After ∼ 15800 yr, another standard thermal pulse takes place. Because of the relatively high degeneracy in the He shell, this second pulse ignites very deeply (at Mr ≃ 0.901 M⊙) and gives rise to a very extended convective region. The helium generated luminosity reaches LHe ≃ 1.5 × 107 L⊙ at its peak and is maintained above 103 L⊙ for ≈ 2900 yr which allows for an efficient He depletion in the intershell. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | and 8 × 105 L⊙, respectively, slightly lower than in the 5 M⊙ model and the duration of the two convective pulses, ∼ 30 and 60 yrs. We note however that in the 4 M⊙ case, the two instabilities are separated in time by only ∼ 10 yr and look like a failed pulse. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | This phenomenon is very similar to the one described in detail by Frost et al. (1998), who first reported the appearance of the so called “degenerate” thermal pulses in their computation of a low metallicity (Z = 0.004) 5 M⊙ AGB model. The occurrence of these instabilities results from the fact that the star experiences a deep dredge-up that quenches rapidly the instability, It is also favored by a higher degeneracy preventing an efficient He depletion in the intershell. which prevents efficient helium burning because of the lower temperature found at the base of the pulse. We present in Fig. 4.1 the evolution of the degeneracy parameter η and temperature, before and after the standard pulse preceding the degenerate instability, during the He accretion phase and just before the ignition of the (degenerate) pulse. As shown, the degeneracy is lifted during the pulse and the temperature increases in the intershell region (solid and dotted lines). During the He accretion phase, η increases and the temperature excess dissipates (short-dashed line). Finally, when the degenerate pulse develops, the parameter η is larger and the temperature lower than in the previous pulse. However, contrary to Frost et al. (1998), our degenerate pulse develops very early in the thermally pulsing AGB phase, probably as the result of the higher degeneracy of our zero metallicity stars. We also note that the amplitude of the generated luminosity and amount of dredge-up before the pulse are smaller in our simulations. These calculations are totally independent of those of Frost et al. (1998) and they verify that degenerate pulses indeed occur (at least in models, if not in real stars !). |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | After this series of events, our 4 and 5 M⊙ stars enter a “standard” AGB evolution phase. We note however, that the temperature and degeneracy keep increasing at the base of the helium shell. This situation is favorable for a later development of a degenerate pulse (see also Frost et al. 1998). Finally, note that CDLS do not report any “strange” behavior such as the one described here in the evolution of their 5 M⊙ Z = 0 model. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | that after the rapid depletion of 14N, some 13C is produced by 12C(n,γ) (see Tab. 6). |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | Finally, the main reactions which take place in the HCS are the CNO reactions. The fact that the elements synthesized during the pulse are mixed in the HCS at relatively high temperatures also leads, but to a much lesser extent, to marginal α-capture on CNO isotopes. However, because of the small abundance of Ne and Mg in the HCS, the NeNa and MgAl cycles are not working efficiently. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | Third dredge-up episodes are responsible for the formation of carbon stars (see e.g. Iben & Renzini 1982) and are also necessary (but not sufficient) to account for the synthesis and observa- tions of s-elements in AGB stars. However, details of 3DUP are dependent on numerous factors such as the numerical scheme (e.g. Frost & Lattanzio 1996), the spatial or temporal resolution of the simulations (Straniero et al. 1997), the input physics (choice of opacity, mass loss rate) and is favored by the presence of extra-mixing or overshooting. In past years, new formalisms for overshooting (e.g. Herwig et al. 1997) and rotationally-induced mixing (Langer et al. 1999) have been developed in order to facilitate the occurrence of 3DUP to address the issues related to the luminosity function of carbon stars (so called C-star mystery) and production of s-elements. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | One of the great interest in the first generation of stars is their effect on the chemical evolution of the universe. We have seen that these stars do, indeed, experience the third dredge-up on the AGB, and also hot bottom burning. The consequences of these events for the chemical history of the universe are our concern in this section. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | The largest uncertainties in calculating the chemical yields are the unknown mass-loss history of the stars and the amount of dredge-up material which affects both the chemical composition and the efficiency at which stellar matter is returned to the interstellar medium. If, as is believed, the mass-loss in low- and intermediate-mass stars is driven by (or at least strongly coupled to) radiation pressure on grains, then an envelope free from grains could have a very reduced mass-loss. However, if the third dredge-up takes place, as found in these simulations, the envelope is enriched in metals (mainly C, N and O) and the mass loss is switched on, although later in the evolution compared to higher metallicity stars. Thus the initial-final mass relation is likely to favor the production of more massive white dwarfs at Z = 0 than for other values of metallicity. On the other hand, the higher effective temperatures characteristic of these stars might make it more difficult for grains to form (even though nova systems show that grains form even in unlikely environments). As a consequence, the uncertainty in the form of both the mass-loss rate and the efficiency of the third dredge-up are still a major limitation at present and prevents us from calculating yields for our models. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | is exposed to high temperatures and a significant flux of α particles during the next thermal pulse, it is completely burned into 22Ne. The 22Ne is, in turn, partially burned by further α-captures into 25Mg (if subject to (α,n) reactions) or 26Mg (if subject to (α, γ) instead). Details depend on the stellar mass, of course, as the main parameter determining the shell temperatures, but we expect both to occur to some extent. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | We have also seen that for masses above about 2 M⊙ there is activation of hot bottom burning. This is in distinct contrast to the standard assumption in the literature that the minimum mass for HBB is between 4 and 5 M⊙. Although this is true for solar metallicity, a decrease in Z will cause the onset of HBB to occur at lower masses, as found in Lattanzio et al. (2001). Clearly these stars will be significant sources of 14N in the early Universe (see top panel of Figs. 4.1, 4.1, 4.1 and 4.1). |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | But HBB does more than burn 12C into 14N. It will also process the 22Ne into 23Na, as well as the 25Mg and 26Mg into 26Al and 27Al, respectively. A potentially important reaction is the production of 7Li via the Cameron-Fowler mechanism. But although 7Li is rapidly and abundantly produced by HBB it is also destroyed on subsequent passes through the bottom of the convective envelope. Hence the Li-rich phase is only temporary. Whether these stars can add to the initial Li abundance in the early universe will depend crucially on the mass-loss involved: if the mass-loss is significant while the surface Li abundance is high, then they will produce a large Li yield. If, however, the Li is depleted before the high mass-loss rates begin (a situation believed to occur in the current generation of stars) then the yields will be negligible, or even negative. The reader is referred to Travaglio et al. (2001) for a fuller discussion. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | Figures 4.1, 4.1, 4.1 and 4.1 depict the evolution of the structure concomitantly to the evolution of the surface abundances of some keys elements. As can be seen, during the initial carbon ingestion, the stars soon become N-rich as a consequence of dredging-up to the surface the products of CNO cycle which was operating in the HCS. Following a few thermal pulses, the 3DUP takes place and the abundances of C, O, Mg, Ne and Na increase. Note that the stars rapidly become C-rich. However it is not clear whether primordial stars will maintain their large C/O ratio or not. Indeed, the evolution of the C/O ratio depends on the competition between 3DUP (which increases the C abundance) and HBB (which destroys it) which in turn depends on the envelope mass, i.e. on mass loss. Frost et al. (1998) showed that, as the envelope mass decreases, HBB shuts down first while 3DUP continues. As far as our computations are concerned, our low- and intermediate- mass stars show a C/O ratio larger than one, but for how long this feature will be maintained is presently unknown. When the temperatures at the base of the convective envelope approaches 40 × 106K, HBB is activated. In the 3 M⊙ and 7 M⊙ models, the signatures of this efficient nuclear burning is well illustrated : the 12C is rapidly converted into 14N which becomes the most abundant metal in the envelope and the 22Ne after proton capture produces a large amount of 23Na. However, because of the maintenance of a high temperature at the base of the convective envelope (T & 108K, ρ > 10g cm−3) of the 7 M⊙ model, proton captures on 23Na destroy this element at the benefit of 20Ne and 24Mg. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | Finally, if mixing of protons is taking place at the base of the convective envelope (e.g. due to diffusive overshooting) the protons will react on the abundant 12C and form a “13C pocket” which will be responsible for a large release of neutrons (Iben & Renzini 1982, Busso et al. 1999 for a review). The neutrons will then be captured by the most abundant nuclei, i.e. C,N,O and Ne and lead to s-process enrichment despite the absence of iron (Goriely & Siess 2001). As a consequence, we can expect s-process elements to have formed in primordial stars. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | The evolution of low- and intermediate mass stars in the mass range 1 . M ≤ 5 M⊙ is char- acterized by the development of mixing episodes at the beginning of the AGB phase. During these events, a secondary convective shell develops at the H-He discontinuity, it expands and overlaps with the underlying carbon rich layers left by the receding pulse. The engulfed carbon initiates a H flash which further expands the structure and produces a deeper penetration of the envelope. The envelope eventually reaches the region previously occupied by this secondary shell and CNO catalysts are dredged-up to the surface by the carbon injection mechanism. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | The metal enrichment of the envelope activates mass loss, although later in the evolution compared to higher metallicity stars. As a consequence, we expect primordial stars to form more massive white dwarf cores than their more metal rich counterparts, but large uncertainties still remain on the exact form of the mass loss history. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | generation of stars pollutes the proto-globular cluster cloud, producing initial abundance anomalies which survive and are observed nowadays in population II stars. These examples illustrate only a few of the relevant issues associated with the unique evolutionary properties of the first stars. A more extended study, covering an even larger parameter space than the present work, as well as a better understanding of some of the physical processes are needed to address all of these issues. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | The authors gratefully acknowledge valuable comments by an anonymous referee. These com- ments have significantly improved the presentation of this paper. LS wishes to thanks the STScI and Monash Department of Mathematics and Statistics for their hospitality during his successive visits. This work has been funded in part by the grant 938-COS191 from the John Templeton Foundation in part by the Australian Research Council and in part by the French-Australian In- ternational Program for Scientific Collaboration “Stellar Populations, Dark Matter and Galaxy Surveys”. LS also acknowledges support from a European TMR “Marie Curie” fellowship at ULB. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | Castellani V. 2000, in “The first stars”, ESO Astrophysics Symposia, Eds. A. Weiss, T.G. Abel, V. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | Fig. 1.— Central carbon abundance as a function of central H abundance (in mass fraction). From right to left the different curves correspond to stellar masses equal to 0.8, 1.0, 1.5, 2, 3, 4, 5, 7, 10, 15 and 20 M⊙, respectively. The ignition of the CNO cycle during the main sequence occurs when log(12C) ≃ −11.5 and can be identified by the bump in the different curves, which corresponds to the appearance of a convective core. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | Fig. 2.— HRD diagram for the complete set of Z = 0 evolutionary tracks, from 0.8 to 20 M⊙. The dots and stars on the tracks indicate the end of the H and He core burning phases, respectively. The 0.8 and 1.0 M⊙ experience the first dredge-up and at the tip of their red giant branch undergo the He flash (triangle in case of the 1 M⊙ model). |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | Fig. 3.— Evolution of the convective regions (lower panel) and of the nuclear luminosity (upper panel) of the 1 M⊙ model during the He core flash. Convective regions are represented by hatched areas. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | Fig. 4.— Evolution of the structure of the 5 M⊙ model. The hatched areas correspond to a convective region, the heavy solid lines delineates the H and He burning regions where εnuc > 5 erg g s−1. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | Fig. 5.— Structural evolution of the 3 M⊙ model during the carbon injection (lower panel). The convective regions are hatched and the dashed lines delineate the H and He burning regions. The dotted line inside the burning regions corresponds to the locus of maximum nuclear energy production. The top panel shows the evolution of the H (solid) and He (dotted line) generated luminosities. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | Fig. 7.— Abundance profiles in mass fraction of 4He (solid line), 12C (dotted line) and 16O (dashed line) during the degenerate pulse of our 5 M⊙ model. In the lower and upper panels, the thick and thin lines correspond to the beginning (b) and the end (e) of the pulses and are separated in time by 10 and 40 yrs, respectively. Note the large amount of He remaining in the intershell and long tail of unburnt He left after the first pulse. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | Fig. 8.— Convective zones as a function of model number for the degenerate pulses in the 4 and 5 M⊙ models. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | Fig. 9.— Evolution of the profiles of temperature and degeneracy parameter η in the 5 M⊙ model. The solid and dotted lines correspond to the beginning and end of the pulse preceding the degenerate pulse, respectively. The short- and long-dashed depict the situation during the He accretion phase and at the beginning of the degenerate pulse, respectively. Note the higher degeneracy and lower temperature at the ignition of this latter pulse. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | generation by the HBS and HeBS. In the upper panel, the surface abundances (in mass fraction) of 12C, 16O, 14N, 22Ne, 23Na, 26Mg and 25Mg are also represented. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | Fig. 11.— Same as Fig. 4.1 but for our 3 M⊙ model. Note the large production of 14N and 23Na due to efficient HBB. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | Fig. 12.— Same as Fig. 4.1 but for our 7 M⊙ model. Note the rapid depletion of 23Na by proton captures due to the high temperature at the base of the convective envelope (T ∼ 108K). |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | Fig. 13.— Same as Fig. 4.1 but for our 1 M⊙ model. Note that dredge-up only occurs in the 4th and 8th pulse. |
Processed_Structure,_Evolution_and_Nucleosynthesis_of_Primor.txt | Table 6. Mass fractions of some elements inside selected thermal pulse of the 3 M⊙ model. The abundances are shown near the beginning and end of the convective pulse. |
Processed_Codes_for_error_detection,_good_or_not_good.txt | Abstract— Linear codes for error detection on a q-ary sym- metric channel are studied. It is shown that for given dimension k and minimum distance d, there exists a value µ(d, k) such that if C is a code of length n ≥ µ(d, k), then neither C nor its dual C ⊥ are good for error detection. For d ≫ k or k ≫ d good approximations for µ(d, k) are given. A generalization to non-linear codes is also given. |
Processed_Codes_for_error_detection,_good_or_not_good.txt | some p ∈ (0, (q − 1)/q). Note that being bad is a stronger condition than not being good. We also note that most codes are either good or bad, but a code may be neither. Lemma 1: C is bad if and only if C⊥ is bad. |
Processed_Codes_for_error_detection,_good_or_not_good.txt | Only some codes have this property. The purpose of this paper is to show that, for given dimension k and minimum distance d, there exists a value µ(d, k) such that if n ≥ µ(d, k) and C is an [n, k, d] code, then neither C nor its dual C⊥ are good for error detection. Further, for d ≫ k or k ≫ d we give good approximations of µ(d, k). At the end of the paper we give similar results for non-linear codes. |
Processed_Codes_for_error_detection,_good_or_not_good.txt | Remark 1. For q = 2, Lemma 1 is Theorem 3.4.2 part 1 in [1]. The proof for general q given above is a straightforward generalization of the proof for q = 2 given in [1]. |
Processed_Codes_for_error_detection,_good_or_not_good.txt | ln(q − 1) + d ln(1 − Qp) ≥ n ln(1 − p) + k ln(q). |
Processed_Codes_for_error_detection,_good_or_not_good.txt | Any choice of p, 0 < p < (q − 1)/q now gives a proof of the existence of µ(k, d) such that if n ≥ µ(d, k) and C is an [n, k, d] code, then neither C nor its dual C⊥ are good for error detection. |
Processed_Codes_for_error_detection,_good_or_not_good.txt | for all p ∈ (0, (q−1)/q), and so g(p) is also convex on (0, (q− 1)/q). This implies that the combined function h(p) is also convex on (0, (q −1)/q) since κ > 0, and it takes its minimum somewhere in (0, (q − 1)/q). We denote this minimum by µ(d, k). From Lemmas 1 and 2 we get the following corollary. |
Processed_Codes_for_error_detection,_good_or_not_good.txt | good for error detection. We next find approximations for µ(d, k) when d ≫ k or k ≫ d. We denote by pm the value of p where h(p) has its minimum (the minimum is by definition µ(d, k)). |
Processed_Codes_for_error_detection,_good_or_not_good.txt | All coefficients should be zero. In particular, the coefficient of y2 shows that a2 1 = 4. Since a1y2 is the dominating term in the expression for p when y is small and p > 0, we must have a1 > 0 and so a1 = 2. Next the coefficient of y3 shows that a2 = −(16Q + 4)/6. In general, we get equations in the ai which can be used to determine the ai recursively. The recursions seems to be quite complicated in general and we have not found an explicit general expression for ai. Substituting the expression for p into h(p) and taking ✷ Taylor expansion, we get the expression for µ(d, k). |
Processed_Codes_for_error_detection,_good_or_not_good.txt | A1(x) = 1, A2(x) = x + λ − 1, A3(x) = (2x2 + (4λ − 2)x + 2λ2 − 3λ)/2, B0(x) = 1, B1(x) = −(x − 1), B2(x) = −(2x − λ + 2)/2, B3(x) = −(3x2 + 3λx + λ2 − 3)/6. |
Processed_Codes_for_error_detection,_good_or_not_good.txt | i=1 αi(Λ)θi+O(θr+1). Formally treating We now write y = Λ as if it were a constant, we can take the Taylor expansion of g(y) in terms of θ and we get an expansion of the form i=1 ciθi, where the ci are polynomials of Λ. Since these polynomials ci must be identically zero, we get equations to P determine the polynomials Ai(x). Substituting the series of pm into h(p) and taking the Taylor expansion, we get the series ✷ of µ(d, k). |
Processed_Codes_for_error_detection,_good_or_not_good.txt | Remark. To formally justify that we treat Λ as if it were a constant, we should prove that Λ is algebraically independent of the other quantities involved, that is, there is no non- trivial polynomial equation in Λ with coefficients expressed as rational functions of the remaining quantities. We have not done this, but it highly likely that it is true. |
Processed_Codes_for_error_detection,_good_or_not_good.txt | V. NON-LINEAR CODES To get results for a non-linear code C, we have to do some modifications. In stead of the weight distribution, we let {Ai} denote the distance distribution (for linear codes these are the same). Then equation (2) is still valid for non-linear codes. |
Processed_Codes_for_error_detection,_good_or_not_good.txt | this Pue(C⊥, p) = P ⊥ of Pue(C⊥, π) and P ⊥ proof of Lemmas 1 and 2, we get the following result. |
Processed_Codes_for_error_detection,_good_or_not_good.txt | This illustrate the results we have found that for large d, µ(d, k) and µ(d, k)N are both approximately 2d (taking only the first term of the approximation) and for large k, µ(d, k) is approximately k whereas µ(d, k)N is approximately 2k (reflecting the fact that κ is approximately twice as large in the non-linear case). |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | Gauß cubature (multidimensional numerical integration) rules are the nat- ural generalisation of the 1D Gauß rules. They are optimal in the sense that they exactly integrate polynomials of as high a degree as possible for a par- ticular number of points (function evaluations). For smooth integrands, they are accurate, computationally efficient formulae. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | The construction of the points and weights of a Gauß rule requires the so- lution of a system of moment equations. In 1D, this system can be converted to a linear system, and a unique solution is obtained, for which the points lie within the region of integration, and the weights are all positive. These properties help ensure numerical stability, and we describe the rules as ‘good’. In the multidimensional case, the moment equations are nonlinear algebraic equations, and a solution is not guaranteed to even exist, let alone be good. The size and degree of the system grow with the degree of the desired cuba- ture rule. Analytic solution generally becomes impossible as the degree of the polynomial equations to be solved goes beyond 4, and numerical approxima- tions are required. The uncertainty of the existence of solutions, coupled with the size and degree of the system makes the problem daunting for numerical methods. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | The construction of Gauß rules for (fully symmetric) n-dimensional regions is easily specialised to the case of U3, the unit sphere in 3D. Despite the problems described above, for degrees up to 17, good Gauß rules for U3 have been constructed/discovered. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | ZRn R are their respective The xi ∈ weights. We want rules that are as accurate as possible for a given number of points (function evaluations). Smooth integrands may be accurately approximated by polynomials, hence Gauß rules, which exactly integrate all polynomials of as high a degree as possible, are a natural choice. For numerical stability, we also want rules with positive weights, and points which lie within Rn (these things are automatic in 1D). Up until about 1975, the best rules known were the (Cartesian) product rules, which although good, are not optimal. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | Rn is a natural generalisation of the well-known 1D case. It originates in a foundation paper by Mantel and Rabinowitz (1977) [18]. The material has been applied to the case of U3 [9], although no computed rules have been published. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | Before progressing, we sketch an application of the use of cubature for U3: the solution of the interior Dirichlet problem using a boundary integral equation. More significant applications are from statistics, where cubatures over many dimensions are required, and efficiency is critical. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | The interior Dirichlet problem is a convenient model problem to work with. Its solution represents a potential function, that is readily related to physical observ- ables (e.g. electrostatic force). It is mathematically attractive, as the existence of a unique solution is known. Where the boundary ∂G, and the boundary data f are simple, it may even be possible to find an analytic solution. In general, however, this is not possible, and it is more sensible to construct a numerical approxima- tion. Knowledge of the existence of a unique solution, for even quite nonsmooth boundaries, greatly encourages this. Techniques to (approximately) solve the inte- rior Dirichlet problem may be able to be used as models for the solution of more sophisticated boundary value PDEs. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | A typical choice is (a, b, c) = (1, 2, 1.1). As c decreases towards 1, the shape be- comes less convex, eventually becoming like a peanut, and numerical methods lose accuracy. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | Accurate approximation of multidimensional integrals is in general a computation- ally expensive task. The rapid growth in expense with n has been called the ‘curse of dimensionality’ (see Rn has some sym- metry (consider U3), and this can greatly simplify both the construction and the application of rules. Here, we consider the case where Rn has some symmetry (not an issue in 1D!), and the construction of Gauß rules, which minimise computational expense for a desired accuracy. The philosophy is that if Rn and ω have a certain symmetry, then it will be natural to place equally-weighted cubature points within Rn according to the same symmetry. This symmetry should considerably reduce the computations required to construct the rule. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | The use of symmetry for cubature rules dates back to Russian works in the 1960s, primarily in papers by Sobolev [24, 25, 23]. Other symmetries for 3D can be defined in terms of the symmetry groups of other Platonic solids. The immedi- ately attractive case is that of the dodecahedron/icosahedron, but this is harder to visualise than that of the octahedron (where the vertices all lie on the coordinate axes), and leads to more difficult algebra to disentangle. In any case, alternative symmetries have little application beyond 3D, whilst the notion of full symmetry generalises perfectly. The notion of full symmetry dates back to Lyness (1965) [15], and the use of octahedral symmetry for cubature on U3 was first put on a clear foundation by Lebedev (1976) [14]. Shortly after this, the foundation paper of Man- tel and Rabinowitz (1977) [18] generalised the notion to arbitrary fully symmetric domains, implicitly using the octahedral symmetry, although not acknowledging the intellectual heritage. This presentation of the theory closely follows both [18] and a complementary paper by Keast and Lyness (1979) [11]. To begin, we formalise the notion of full symmetry. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | Definition 2.1 (Full Symmetry between Two Points) x and y are a pair of fully symmetric (FS) points in Rn, denoted x y, if y can be reached from x by permutations and/or sign changes of the entries of x. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | The number of elements in each class thus varies between 1 (for the equivalence class [0]) and 2nn! (in general). The exact number will be important for our work. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | Definition 2.2 (Full Symmetry for a Set of Points) A set of points is called fully symmetric (FS) if x ∈ Rn and y Rn contains only complete equivalence classes. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | Recall that the degree of a polynomial in n variables is the maximum sum of the exponents in any of its terms, not the maximum exponent of any one variable appearing in its terms. Thus 3x3 2 + x5 1x3 is a polynomial in 3 (or more!) variables, that is of degree 7 (not 5), whilst x5 1 and x2 2 are monomials of degree 5. The rule I N is exact for a function f , if I N [f ] = I[f ]. If it is exact for all polynomials in n variables of degree up to and including k, then it is called an integration rule of degree (of exactness) k. Cubature rules of high degrees of exactness should accurately integrate smooth functions (which can be accurately approximated by polynomials), but this does not necessarily carry over to non-smooth functions. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | Definition 2.4 (Fully Symmetric Integration Rule) A cubature rule I N is called a fully symmetric integration rule if the evaluation points form an FS set, and all points in an FS equivalence class within the rule have the same weight. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | Cubature rules that have a minimal number of points for a specified degree are called ‘minimal rules’. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | The first of these conditions is familiar from 1D Gauß quadrature, whilst the second is new. These properties seem natural, however there is nothing in the as- sumptions for cubature that demands them. In 1D, na¨ıve approaches to quadrature, such as the (equally-spaced) Newton-Cotes family, do not preserve the positivity of weights for larger N , whilst both these properties are satisfied by 1D Gauß rules. In higher dimensions, Gauß rules are not necessarily good. The properties of good rules, in particular the latter, assist numerical stability. Whilst a rule may theoret- ically exactly integrate the relevant polynomials, in practice, summation of large 3, we consider the terms of alternating sign tends to reduce numerical precision. In case where Rn is U3, so the internal points condition will simply require points to lie on the surface of the unit sphere. We use the concept of full symmetry to define several classes of rules for the cubature of integrals with FS weight functions where the moments 2 exist (commonly ω 1). A fully symmetric cubature rule of degree 2m + 1 exactly integrates all polynomials of degree up to and including 2m. A fully symmetric minimal (FSM) rule does so using a minimal number of cubature points. A fully symmetric good (FSG) rule does so where all weights are positive and all points lie within Rn. A fully symmetric minimal good (FSMG) rule is both minimal and good. A fully symmetric good minimal (FSGM) rule is a good rule that is minimal, that is, although FS rules on fewer points may exist, none are good. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | We seek firstly FSMG rules, and if there are none of these, FSGM rules, which 4), which (generally) always exist. At worst, FSGM rules are product rules (see 1]n = (4m + 1)n points to be of degree 2m + 1. For low- require [2(2m + 1) dimensional applications (small n), this may not be terribly inefficient, e.g. product rules for U3 actually require 2(m + 1)2 points. Apart from such considerations, a rule that is minimal (or almost so) might be almost good in the sense that negative Rn, or that failing these weights are very small, or that points are only just outside conditions, errors in integrating polynomials are minor. The quest for good rules may be an arduous search through these almost good rules. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | The procedure for constructing Gauß rules becomes more complicated as the di- mension n increases. Here, we fully develop the 3D case, specialising this to the case of U3 in 3. We will set up a system of (moment) equations to express the fact R3 will exactly integrate all polynomials of up to a specified that an FS rule for degree, without presupposing the number or distribution (beyond being an FS set) of points, or the sign of weights. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | N = K0 + 6K1 + 12K2 + 24K3 + 8K4 + 24K5 + 48K6. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | that will Ki} lead to a system ( { ∗ ) is consistent. This is an optimisation problem, with to minimise N , such that ( ∗ constraints that will ensure the loosest possible consistency of ( ). It turns out [11, ∗ 18] that the optimisation problem has linear constraints. Clearly the cost function are integers. The optimisation problem is thus (4) is linear, and the solutions a linear integer programming problem, with linear cost function. This is routine to solve, and many examples of solutions are provided by Rabinowitz et al. in [20] (for 2D) and [17, 18] (for 3D). |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | of dimensions is presented in [11], it seems to be essential to use a machine to do this. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | Figure 2: The constraints for system ( ∗ p 398], in which it is called system ¯C). We apply the convention that if q < p. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | We specialise the theory presented in Rn is U3, and the weight- ing function is unity. As the ‘volume’ of U3 is finite, integrals over it of any bounded function will be defined. We describe the spherical harmonic polynomials, which our cubature rule should integrate exactly, and comment that this is equivalent to our previous requirement that our rules integrate polynomials exactly. The construction of Gauß cubature rules follows, with simplifying assumptions that are obtained by considering the region of integration. The material presented is related to that of Keast [10, 9]. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | we can relate any spherical harmonic to an associated Legendre polynomial P m with m > 0. Table 3 lists some of the simplest spherical harmonics. As l and m increase, the degree of the trigonometric functions increases. Note that for m = 0, they are purely real. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | Table 3: The first few spherical harmonics and the associated Legendre polynomials (after [19, p 195]). |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | i = 1, . . . , K3 i = 1, . . . , K5 i = 1, . . . , K6. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | In general for FS regions R3, there are a total of 13 consistency conditions (Figure 2), but for U3 these simplify drastically. There are only 4, and these listed in Figure 3. For m = 1, . . . , 20, the appropriate right hand sides of the constraint equations in Figure 3 are presented in Table 4. |
Processed_Gauß_Cubature_for_the_Surface_of_the_Unit_Sphere.txt | Table 4: The four elements of the columns of the right hand side in Figure 3, for various m, as generated by appropriate sums of the pν(r). |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.